From 40143ac2f9cdd57ecdbd9ae623f2d8c7f91023f4 Mon Sep 17 00:00:00 2001 From: mengleiz Date: Mon, 14 Aug 2023 17:41:29 -0700 Subject: [PATCH] update website --- README.html | 14 +- _images/network_gym_overview.png | Bin 0 -> 128155 bytes _images/network_gym_overview2.png | Bin 0 -> 160272 bytes _images/network_gym_uml.png | Bin 35012 -> 34992 bytes _modules/index.html | 16 +- _modules/network_gym_client/adapter.html | 14 +- _modules/network_gym_client/env.html | 55 +- .../envs/network_slicing/adapter.html | 16 +- .../envs/nqos_split/adapter.html | 23 +- .../envs/qos_steer/adapter.html | 16 +- .../northbound_interface.html | 213 +----- _modules/network_gym_env/custom_env.html | 495 ++++++++++++++ _modules/network_gym_env/dummy_sim.html | 562 ++++++++++++++++ .../northbound_interface.md | 17 +- _sources/api/network_gym_env.md | 30 +- _sources/api/network_gym_env/custom_env.md | 17 + _sources/api/network_gym_env/dummy_sim.md | 19 + _sources/content/overview.md | 8 +- _sources/index.md | 2 + .../customizing_observation_and_reward.md | 98 +++ .../implementing_custom_environment.md | 12 + _sources/tutorials/training_agents.md | 7 +- api/network_gym_client.html | 14 +- api/network_gym_client/adapter.html | 14 +- .../adapter/network_slicing.html | 20 +- .../adapter/nqos_split.html | 14 +- api/network_gym_client/adapter/qos_steer.html | 16 +- api/network_gym_client/env.html | 14 +- .../northbound_interface.html | 93 ++- api/network_gym_env.html | 51 +- api/network_gym_env/custom_env.html | 491 ++++++++++++++ api/network_gym_env/dummy_sim.html | 536 +++++++++++++++ api/network_gym_server.html | 18 +- content/motivation.html | 14 +- content/overview.html | 22 +- content/quickstart.html | 14 +- environments/mx_network_slicing.html | 14 +- .../cellular_network_slicing.html | 14 +- environments/mx_traffic_management.html | 20 +- .../mx_qos_traffic_steering.html | 14 +- .../mx_traffic_splitting.html | 14 +- genindex.html | 49 +- index.html | 14 +- objects.inv | Bin 1168 -> 1344 bytes old-index.html | 14 +- search.html | 14 +- searchindex.js | 2 +- .../customizing_observation_and_reward.html | 608 ++++++++++++++++++ tutorials/handling_time_limits.html | 14 +- .../implementing_custom_environment.html | 418 ++++++++++++ tutorials/training_agents.html | 29 +- 51 files changed, 3733 insertions(+), 470 deletions(-) create mode 100644 _images/network_gym_overview.png create mode 100644 _images/network_gym_overview2.png create mode 100644 _modules/network_gym_env/custom_env.html create mode 100644 _modules/network_gym_env/dummy_sim.html create mode 100644 _sources/api/network_gym_env/custom_env.md create mode 100644 _sources/api/network_gym_env/dummy_sim.md create mode 100644 _sources/tutorials/customizing_observation_and_reward.md create mode 100644 _sources/tutorials/implementing_custom_environment.md create mode 100644 api/network_gym_env/custom_env.html create mode 100644 api/network_gym_env/dummy_sim.html create mode 100644 tutorials/customizing_observation_and_reward.html create mode 100644 tutorials/implementing_custom_environment.html diff --git a/README.html b/README.html index 79f0169..ce91aa6 100644 --- a/README.html +++ b/README.html @@ -145,20 +145,24 @@
  • network_slicing Adapter
    • -
  • NorthboundInterface
    • +
  • NorthBoundClient
  • network_gym_server
    • -
  • network_gym_env
    • +
  • network_gym_env +

    • Environments

    Environments

  • network_gym_server
    • -
  • network_gym_env
    • +
  • network_gym_env +

    • Environments

    Environments

      -
    • Traffic Management
        +
      • Traffic Management
        • -
      • Network Slicing
          +
        • Network Slicing
          • @@ -166,6 +170,8 @@

          Development

          @@ -289,7 +295,7 @@

          Source code for network_gym_client.env

           import json
 import sys
 import os
-from network_gym_client.northbound_interface import NorthboundInterface
+from network_gym_client.northbound_interface import NorthBoundClient
 np.set_printoptions(precision=3)
 
 FILE_PATH = pathlib.Path(__file__).parent
@@ -382,7 +388,7 @@ 
          Source code for network_gym_client.env
                   self.action_space = self.adapter.get_action_space()
         self.observation_space = self.adapter.get_observation_space()
 
-        self.northbound_interface_client = NorthboundInterface(id, config_json) #initial northbound_interface_client
+        self.northbound_interface_client = NorthBoundClient(id, config_json) #initial northbound_interface_client
         #self.max_counter = int(config_json['env_config']['simulation_time_s'] * 1000/config_json['env_config']['GMA']['measurement_interval_ms'])# Already checked the interval for Wi-Fi and LTE in the main file
 
         #self.link_type = config_json['rl_config']['link_type'] 
@@ -415,6 +421,8 @@ 
          Source code for network_gym_client.env
                   self.current_step = 1
         self.current_ep += 1
 
+        print("reset() at episode:" + str(self.current_ep) + ", step:" + str(self.current_step))
+
         # if a new simulation starts (first episode) in the reset function, we need to connect to server
         # else a new episode of the same simulation.
             # do not need to connect, send action directly
@@ -423,24 +431,14 @@ 
          Source code for network_gym_client.env
                   else:
             self.northbound_interface_client.send(self.last_policy) #send action to network gym server
 
-        measurement_report = self.northbound_interface_client.recv()#first measurement
-        if (measurement_report == None):
-            #empty measurement, env termination.
-            return [], {"network_stats": []}
-        ok_flag = measurement_report.ok_flag
-        df_list =  measurement_report.df_list
-
-        print("reset() at episode:" + str(self.current_ep) + ", step:" + str(self.current_step))
-
-        if self.enable_rl_agent and not ok_flag:
-            print("[WARNING], some users may not have a valid measurement, for qos_steering case, the qos_test is not finished before a measurement return...")
+        network_stats = self.northbound_interface_client.recv()#first measurement
 
-        observation = self.adapter.get_observation(df_list)
+        observation = self.adapter.get_observation(network_stats)
 
         # print(observation.shape)
 
 
-        return observation.astype(np.float32), {"network_stats": df_list}
+        return observation.astype(np.float32), {"network_stats": network_stats}
 
     def step(self, action):
         """Run one timestep of the environment's dynamics using the agent actions.
@@ -470,6 +468,8 @@ 
          Source code for network_gym_client.env
                           a certain timelimit was exceeded, or the physics simulation has entered an invalid state.
         """
         self.current_step += 1
+        
+        print("----------| step() at episode:" + str(self.current_ep) + ", step:" + str(self.current_step) + " |----------")
 
         #1.) Get action from RL agent and send to network gym server
         if not self.enable_rl_agent or action.size == 0:
@@ -482,21 +482,14 @@ 
          Source code for network_gym_client.env
                       self.northbound_interface_client.send(policy) #send network policy to network gym server
 
         #2.) Get measurements from gamsim and obs and reward
-        measurement_report = self.northbound_interface_client.recv()
-        ok_flag = measurement_report.ok_flag
-        df_list =  measurement_report.df_list
-       
-        if self.enable_rl_agent and not ok_flag:
-            print("[WARNING], some users may not have a valid measurement, for qos_steering case, the qos_test is not finished before a measurement return...")
-
-        print("----------| step() at episode:" + str(self.current_ep) + ", step:" + str(self.current_step) + " |----------")
-
-        observation = self.adapter.get_observation(df_list)
+        network_stats = self.northbound_interface_client.recv()
+        
+        observation = self.adapter.get_observation(network_stats)
 
         # print(observation)
 
         #Get reward
-        reward = self.adapter.get_reward(df_list)
+        reward = self.adapter.get_reward(network_stats)
 
 
         self.adapter.wandb_log()
@@ -516,7 +509,7 @@ 
          Source code for network_gym_client.env
                           time.sleep(1) # sleep 1 second to let the server disconnect client and env worker. In case the client restart connection right after a env termination.
         #4.) return observation, reward, done, info
         # print("terminated:" + str(terminated) + " truncated:" + str(truncated))
-        return observation.astype(np.float32), reward, terminated, truncated, {"network_stats": df_list}
          
+        return observation.astype(np.float32), reward, terminated, truncated, {"network_stats": network_stats}
          
 
          
 
                 
diff --git a/_modules/network_gym_client/envs/network_slicing/adapter.html b/_modules/network_gym_client/envs/network_slicing/adapter.html
index 7c63911..11c47e7 100644
--- a/_modules/network_gym_client/envs/network_slicing/adapter.html
+++ b/_modules/network_gym_client/envs/network_slicing/adapter.html
@@ -144,20 +144,24 @@
 
        • network_slicing Adapter
          •
        • -
      • NorthboundInterface
        • +
      • NorthBoundClient
    • network_gym_server
      • -
    • network_gym_env
      • +
    • network_gym_env +

    Environments

    Environments

    Environments

    Environments

      -
    • Traffic Management
        +
      • Traffic Management
        • -
      • Network Slicing
          +
        • Network Slicing
          • @@ -166,6 +170,8 @@

          Development

          @@ -285,20 +291,7 @@

          Source code for network_gym_client.northbound_interface

          import json import pandas as pd -
          [docs]class MeasurementReport: - """Data structure to store the network stats measurement report. - """ - def __init__(self, ok_flag, df_list): - """Initialize measurement report. - - Args: - ok_flag (Bool): indicate whehter the measurement is valid. - df_list (list[pandas.dataframe]): a list of dataframe that stores the network ststs - """ - self.ok_flag = ok_flag - self.df_list = df_list
          - -
          [docs]class NorthboundInterface(): +
          [docs]class NorthBoundClient(): """NetworkGym northbound interface client. Northbound interface connects the network gym client to the network gym server. Client sends the network policy to the Sever/Env. @@ -306,7 +299,7 @@

          Source code for network_gym_client.northbound_interface

          """ def __init__(self, id, config_json): - """Initialize NorthboundInterface. + """Initialize NorthBoundClient. Args: id (int): client ID @@ -315,11 +308,10 @@

          Source code for network_gym_client.northbound_interface

          self.identity = u'%s-%d' % (config_json["session_name"], id) self.config_json=config_json self.socket = None - self.end_ts = None # sync the timestamp between obs and action #connect to network gym server using ZMQ socket def connect(self): - """Connect to the network gym server. + """Connect to the network gym server. Send the Start Env request, where the configuration is loaded from json file. """ context = zmq.Context() self.socket = context.socket(zmq.DEALER) @@ -360,7 +352,7 @@

          Source code for network_gym_client.northbound_interface

          """Receive a message from the network gym server. Returns: - MeasurementReport: the network stats :class:`network_gym_client.MeasurementReport` from the environment + pd.DataFrame: the network stats measurement from the environment """ reply = self.socket.recv() @@ -397,181 +389,8 @@

          Source code for network_gym_client.northbound_interface

          quit() def process_measurement (self, reply_json): - df = pd.json_normalize(reply_json['metric_list']) - if 'end_ts' in df.columns: - self.end_ts = int(df["end_ts"][0]) - ok_flag = True - return MeasurementReport(ok_flag, df) - - ''' - #process measurement from network gym server - def process_measurement (self, reply_json): - """Process the raw network measurements (json) and translate them to pandas.dataframe format. - - Args: - reply_json (json): the message from the server - - Returns: - MeasurementReport: the network stats :class:`network_gym_client.MeasurementReport` from the environment - """ - df_list = [] - ok_flag = True - df = pd.json_normalize(reply_json['metric_list']) - # print(df) - - if self.config_json['env_config']['downlink']: - df = df[df['direction'] == 'DL'].reset_index(drop=True) - else: - df = df[df['direction'] == 'UL'].reset_index(drop=True) - - df_phy = df[df['group'] == 'PHY'].reset_index(drop=True) - df_phy_lte = df_phy[df_phy['cid'] == 'LTE'].reset_index(drop=True) - - df_phy_lte_max_rate = [] - df_phy_wifi_max_rate = [] - - if not df_phy_lte.empty: - # process PHY LTE measurement - - df_phy_lte_start_ts = df_phy_lte[df_phy_lte['name'] == 'start_ts'].reset_index(drop=True) - df_phy_lte_end_ts = df_phy_lte[df_phy_lte['name'] == 'end_ts'].reset_index(drop=True) - - #check PHY LTE timestamps of all users are the same - if 1==len(set(df_phy_lte_start_ts['value'])) and 1==len(set(df_phy_lte_end_ts['value'])): - start_ts = df_phy_lte_start_ts['value'][0] - end_ts = df_phy_lte_end_ts['value'][0] - self.end_ts = end_ts - df_phy_lte_max_rate = df_phy_lte[df_phy_lte['name'] == 'max_rate'].reset_index(drop=True) - df_phy_lte_max_rate.insert(0,'end_ts', end_ts) - df_phy_lte_max_rate.insert(0,'start_ts', start_ts) - df_phy_lte_max_rate['unit'] = 'mbps' - #print(df_phy_lte_max_rate) - - df_phy_lte_slice_id = df_phy_lte[df_phy_lte['name'] == 'slice_id'].reset_index(drop=True) - df_phy_lte_slice_id.insert(0,'end_ts', end_ts) - df_phy_lte_slice_id.insert(0,'start_ts', start_ts) - #print(df_phy_lte_slice_id) - - df_phy_lte_rb_usage = df_phy_lte[df_phy_lte['name'] == 'rb_usage'].reset_index(drop=True) - df_phy_lte_rb_usage.insert(0,'end_ts', end_ts) - df_phy_lte_rb_usage.insert(0,'start_ts', start_ts) - df_phy_lte_rb_usage['unit'] = '%' - # print(df_phy_lte_rb_usage) - - else: - print(self.identity+" "+"ERROR, PHY LTE timestamp is not the same") - - df_phy_wifi = df_phy[df_phy['cid'] == 'Wi-Fi'].reset_index(drop=True) - if not df_phy_wifi.empty: - # process PHY Wi-Fi measurement - df_phy_wifi_start_ts = df_phy_wifi[df_phy_wifi['name'] == 'start_ts'].reset_index(drop=True) - df_phy_wifi_end_ts = df_phy_wifi[df_phy_wifi['name'] == 'end_ts'].reset_index(drop=True) - - #check PHY Wi-Fi timestamps of all users are the same - if 1==len(set(df_phy_wifi_start_ts['value'])) and 1==len(set(df_phy_wifi_end_ts['value'])): - start_ts = df_phy_wifi_start_ts['value'][0] - end_ts = df_phy_wifi_end_ts['value'][0] - self.end_ts = end_ts - df_phy_wifi_max_rate = df_phy_wifi[df_phy_wifi['name'] == 'max_rate'].reset_index(drop=True) - df_phy_wifi_max_rate.insert(0,'end_ts', end_ts) - df_phy_wifi_max_rate.insert(0,'start_ts', start_ts) - df_phy_wifi_max_rate['unit'] = 'mbps' - #print(df_phy_wifi_max_rate) - - else: - print(self.identity+" "+"ERROR, PHY LTE timestamp is not the same") - - df_gma = df[df['group'] == 'GMA'].reset_index(drop=True) - if not df_gma.empty: - # process GMA measurement - df_gma_start_ts = df_gma[df_gma['name'] == 'start_ts'].reset_index(drop=True) - df_gma_end_ts = df_gma[df_gma['name'] == 'end_ts'].reset_index(drop=True) - - #check GMA timestamps of all users are the same - if 1==len(set(df_gma_start_ts['value'])) and 1==len(set(df_gma_end_ts['value'])): - start_ts = df_gma_start_ts['value'][0] - end_ts = df_gma_end_ts['value'][0] - - self.end_ts = end_ts - df_load = df_gma[df_gma['name'] == 'tx_rate'].reset_index(drop=True) - df_load.insert(0,'end_ts', end_ts) - df_load.insert(0,'start_ts', start_ts) - df_load['unit'] = 'mbps' - #print(df_load) - - df_rate = df_gma[df_gma['name'] == 'rate'].reset_index(drop=True) - #df_rate = df_rate[df_rate['cid'] == 'All'].reset_index(drop=True) - - df_rate.insert(0,'end_ts', end_ts) - df_rate.insert(0,'start_ts', start_ts) - df_rate['unit'] = 'mbps' - #print(df_rate) - - df_qos_rate = df_gma[df_gma['name'] == 'qos_rate'].reset_index(drop=True) - - df_qos_rate.insert(0,'end_ts', end_ts) - df_qos_rate.insert(0,'start_ts', start_ts) - df_qos_rate['unit'] = 'mbps' - #print(df_qos_rate) - - df_owd = df_gma[df_gma['name'] == 'owd'].reset_index(drop=True) - #df_owd = df_owd[df_owd['cid'] == 'All'].reset_index(drop=True) - df_owd.insert(0,'end_ts', end_ts) - df_owd.insert(0,'start_ts', start_ts) - df_owd['unit'] = 'ms' - #print(df_owd) - - df_max_owd = df_gma[df_gma['name'] == 'max_owd'].reset_index(drop=True) - df_max_owd.insert(0,'end_ts', end_ts) - df_max_owd.insert(0,'start_ts', start_ts) - df_max_owd['unit'] = 'ms' - #print(df_max_owd) - - df_split_ratio = df_gma[df_gma['name'] == 'split_ratio'].reset_index(drop=True) - df_split_ratio.insert(0,'end_ts', end_ts) - df_split_ratio.insert(0,'start_ts', start_ts) - #print(df_split_ratio) - - df_ap_id = df_gma[df_gma['name'] == 'ap_id'].reset_index(drop=True) - df_ap_id.insert(0,'end_ts', end_ts) - df_ap_id.insert(0,'start_ts', start_ts) - #print(df_ap_id) - - df_delay_violation = df_gma[df_gma['name'] == 'delay_violation'].reset_index(drop=True) - df_delay_violation.insert(0,'end_ts', end_ts) - df_delay_violation.insert(0,'start_ts', start_ts) - df_delay_violation['unit'] = '%' - - #print(df_delay_violation) - - df_ok = df[df['name'] == 'measurement_ok'].reset_index(drop=True) - df_ok.insert(0,'end_ts', end_ts) - df_ok.insert(0,'start_ts', start_ts) - - if(df_ok['value'].min() < 1): - #print("[WARNING], some users may not have a valid measurement, for qos_steering case, the qos_test is not finished before a measurement return...") - #print(df_ok) - ok_flag = False - else: - print(self.identity+" "+"ERROR, GMA timestamp is not the same") - - #return True, df_phy_lte_max_rate, df_phy_wifi_max_rate, df_load, df_rate, df_qos_rate, df_owd, df_split_ratio - - df_list.append(df_phy_lte_max_rate) - df_list.append(df_phy_wifi_max_rate) - df_list.append(df_load) - df_list.append(df_rate) - df_list.append(df_qos_rate) - df_list.append(df_owd) - df_list.append(df_split_ratio) - df_list.append(df_ap_id) - df_list.append(df_phy_lte_slice_id) - df_list.append(df_phy_lte_rb_usage) - df_list.append(df_delay_violation) - df_list.append(df_max_owd) - - return MeasurementReport(ok_flag, df_list) - '''
          + network_stats = pd.json_normalize(reply_json['metric_list']) + return network_stats
          diff --git a/_modules/network_gym_env/custom_env.html b/_modules/network_gym_env/custom_env.html new file mode 100644 index 0000000..9c5f4ec --- /dev/null +++ b/_modules/network_gym_env/custom_env.html @@ -0,0 +1,495 @@ + + + + + + + + + + + network_gym_env.custom_env — NetworkGym Documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
          +
          +
          +
          + + + Ctrl+K +
          +
          + + + +
          +
          + +
          + + + +
          + + + + +
          + +
          +
          + + +
          +
          +
          + + +
          +
          + + + + +
          + +
          + + + +
          + +
          +
          + +
          +
          + +
          + +
          + +
          + + +
          + +
          + +
          + + + + + + + + + + +
          + +
          + +
          +
          + + + +
          +

          + +
          +
          + +
          +
          +
          + + + + +
          + +

          Source code for network_gym_env.custom_env

          +#Copyright(C) 2023 Intel Corporation
+#SPDX-License-Identifier: Apache-2.0
+#File : dummy_worker.py
+
+import zmq
+import threading
+import json
+import traceback
+"""
+1. The "env_config" establishes a connection with the server and periodically dispatches the message "env-hello."
+2. Upon reception of an "env-start" signal, the "env_config" terminates its socket connection and initiates the "env_sim" simulator.
+3. Upon launching, the "env_sim" establishes a connection with the server, utilizing the identical identification as the "env_config."
+    Subsequently, it commences the exchange of measurement/action messages with the client.
+4. Following the conclusion of the "env_sim" simulation, the socket is closed, and the "env_config" reconnects.
+
+From the server's point, the "env_config" and "env_sim" are perceived as identical entities.
+The division between the "env_config" and "env_sim" components provides the advantage of facilitating straightforward expansion
+ of the "env_sim" to other simulators (e.g., ns-3) or test environments, all the while utilizing the same underlying "env_config" code.
+
+Flow Chart:
+-------------------------------------------------
+env_config.connect()
+[every T seconds]:
+    env_config -- "env-hello" --> server
+env_config <-- "env-start" -- server
+env_config.close()
+
+env_sim.connect()
+[every interval]:
+    env_sim -- "env-measurement" --> server
+    env_sim <-- "env-action" -- server
+env_sim.close()
+
+env_config.connect()
+env_config -- "env-end" --> server
+[every T seconds]:
+    env_config -- "env-hello" --> server
+...
+-------------------------------------------------
+"""
+
+
          [docs]class CustomEnv(threading.Thread):
+    """Custom environment that send random measurement metrics."""
+    def __init__(self, identity, NetworkGymSim, env_list, port_num=8091):
+        """Initialize custom environment.
+
+        Args:
+            identity (int): environment identity
+            NetworkGymSim (simulator): network simulator
+            env_list (list[str]): a list of supported environments
+            port_num (int): the port to connect
+        """
+        threading.Thread.__init__ (self)
+        self.identity = identity
+        self.env_list = env_list
+        self.port_num = port_num
+        self.NetworkGymSim = NetworkGymSim
+
+    def run(self):
+        """Run custom environement
+        """
+        context = zmq.Context()
+        env_config = context.socket(zmq.DEALER)
+        identity = str(self.identity)
+        env_config.identity = str(self.identity).encode('utf-8')
+        env_config.connect('tcp://localhost:'+str(self.port_num))
+        print(identity + ': env_config socket connected.')
+
+        poller = zmq.Poller()
+        poller.register(env_config, flags=zmq.POLLIN)
+        
+        while True:
+
+            hello_msg = json.loads('{"type":"env-hello"}')
+            hello_msg["env_list"] = self.env_list # add supported env_list to hello msg
+            env_config.send_multipart([b'', json.dumps(hello_msg, indent=2).encode('utf-8')])# Hello msg does not include client
+            print(identity + ': send env-hello msg.')
+            print(hello_msg)
+
+            # resend hello msg after 5 seconds if no msg is received.
+            # [Warning] always use a poll with timeout before receiving a msg (recv_multipart)! otherwise, it may stuck forever!
+            if poller.poll(timeout=5*1000):
+
+                # received a new msg
+                msg = env_config.recv_multipart()
+                config_json = json.loads(msg[1])  
+                if config_json["type"] == "env-start":
+                    # In idle mode (periodic sending env-hello msg), the first msg should be "env-start"
+
+                    # check if the env is supported (in the env_list)
+                    if config_json["env"] not in self.env_list:
+                        # not supported env
+                        print("Unkown Environment!")
+                        error_msg = json.loads('{"type":"env-error", "error_msg": "Unkown Environment!"}')
+                        msg[1]=json.dumps(error_msg, indent=2).encode('utf-8')
+                        env_config.send_multipart(msg)
+                        continue
+
+                    print(identity + ': Recv.')
+                    print(config_json)
+                    env_config.close()
+
+                    # start dummy simulator
+                    print(identity + ': env_config socket closed.')
+
+                    # The dummy simmulator will start a new socket to send measurement and receive action.
+
+                    # use try except such that if there is an error in the simulator, the thread will not stop and we can report the error to the client.
+                    sim_error_msg = ''
+                    try:
+                        self.NetworkGymSim(identity, self.port_num, msg[0].decode(), config_json) # replace it with your own simulator.
+                    except Exception:
+                        traceback.print_exc()
+                        sim_error_msg = traceback.format_exc()
+
+                    # simulator terminated, env_config reconnect the server.
+                    env_config = context.socket(zmq.DEALER)
+                    env_config.identity = identity.encode('utf-8')
+                    env_config.connect('tcp://localhost:'+str(self.port_num))
+
+                    poller.register(env_config, flags=zmq.POLLIN)
+
+                    if sim_error_msg != '':
+                        # simualtor crashed with error msg
+                        error_msg = json.loads('{"type":"env-error"}')
+                        error_msg["error_msg"] = sim_error_msg
+                        msg[1]=json.dumps(error_msg, indent=2).encode('utf-8')
+                        env_config.send_multipart(msg)# Env End msg
+                    else:
+                        # no error, send env-end to terminate client to env worker mapping
+                        end_msg = json.loads('{"type":"env-end"}')
+                        env_config.send_multipart([msg[0], json.dumps(end_msg, indent=2).encode('utf-8')])# Env End msg
+
+                    print(identity + ': env_config socket connected.')
+
+
+                else:
+                    print("Unkown MSG type, Expecting env-start!")
+                    error_msg = json.loads('{"type":"env-error", "error_msg": "Unkown MSG type, Expecting env-start!"}')
+                    msg[1]=json.dumps(error_msg, indent=2).encode('utf-8')
+                    env_config.send_multipart(msg)
+        env_config.close()
          
+
          + +
          + + + + +
          + +
          + +
          +
          +
          + +
          + +
          + +
          + + + + +
          +
          + +
          + +
          + +

          +By Menglei Zhang +

          + +
          + +
          + +

          + + © Copyright 2023, Intel Labs Multi-Access Intelligence. +
          + +

          + +
          + +
          + +
          + +
          + +
          + +
          +
          + + +
          +
          +
          + + + + + +
          +
          + + \ No newline at end of file diff --git a/_modules/network_gym_env/dummy_sim.html b/_modules/network_gym_env/dummy_sim.html new file mode 100644 index 0000000..d6943f5 --- /dev/null +++ b/_modules/network_gym_env/dummy_sim.html @@ -0,0 +1,562 @@ + + + + + + + + + + + network_gym_env.dummy_sim — NetworkGym Documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
          +
          +
          +
          + + + Ctrl+K +
          +
          + + + +
          +
          + +
          + + + +
          + + + + +
          + +
          +
          + + +
          +
          +
          + + +
          +
          + + + + +
          + +
          + + + +
          + +
          +
          + +
          +
          + +
          + +
          + +
          + + +
          + +
          + +
          + + + + + + + + + + +
          + +
          + +
          +
          + + + +
          +

          + +
          +
          + +
          +
          +
          + + + + +
          + +

          Source code for network_gym_env.dummy_sim

          +#Copyright(C) 2023 Intel Corporation
+#SPDX-License-Identifier: Apache-2.0
+#File : start_dummy_worker.py
+
+
+"""
+1. Upon launching, the "env_sim" establishes a connection with the server, utilizing the identical identification as the "env_config."
+    Subsequently, it commences the exchange of measurement/action messages with the client.
+2. Following the conclusion of the "env_sim" simulation, the socket is closed, and the "env_config" reconnects.
+
+
+Flow Chart:
+-------------------------------------------------
+
+env_sim.connect()
+[every interval]:
+    env_sim -- "env-measurement" --> server
+    env_sim <-- "env-action" -- server
+env_sim.close()
+
+-------------------------------------------------
+"""
+
+import pandas as pd
+import json 
+import random
+import zmq
+
+
          [docs]class DummySim:
+    """A dummy simulator that generate random measurement samples. When the simulation terminates, resume to the env_config.
+
+        The simulator will create a new env_sim socket using the env_idenntity and connects to the env_port.
+        The simulator is configured using the config_json file.
+        The first measurement will be send to the client_identity.
+    """
+    def __init__(self, env_identity, env_port, client_identity, config_json):
+        """Initilize dummy simulator.
+
+        Args:
+            env_identity (str): the identity of the environement socket
+            env_port (int): the port number to connect
+            client_identity (str): the identity of the client socket who started the env
+            config_json (json): environment configuration
+        """
+        # use the config_json to config the simulator
+        self.interval = config_json['measurement_interval_ms'] + config_json['measurement_guard_interval_ms'] # measurement interval
+        self.start_ts = config_json['app_and_measurement_start_time_ms'] # start timestamp of a measurement
+        self.end_ts = self.start_ts + self.interval # end timestamp of a measurement
+        self.sim_end_ts = self.start_ts + config_json['simulation_time_s']*1000
+        self.num_users = config_json['num_users']
+        self.start_simulation(env_identity, env_port, client_identity)
+
+    def start_simulation(self, env_identity, env_port, client_identity):
+        """Start simulation. Connect to the server using SouthBound API. Report network stats measurment and receive action.
+
+        Args:
+            env_identity (str): the identity of the environement socket
+            env_port (int): the port number to connect
+            client_identity (str): the identity of the client socket who started the env
+        """
+        # open a socket with the same env_identity and connect to the same env_port.
+        context = zmq.Context()
+        env_sim = context.socket(zmq.DEALER)
+        env_sim.identity = env_identity.encode('utf-8')
+        env_sim.connect('tcp://localhost:'+str(env_port))
+        print(env_identity + ': env_sim socket connected.')
+
+        poller = zmq.Poller()
+        poller.register(env_sim, flags=zmq.POLLIN)
+        
+        # running simualtor
+        while True:
+            dummy_report = json.loads('{"type":"env-measurement"}')
+            dummy_report["metric_list"] = self.run_one_interval()
+
+            # the first part of the msg is the client_identity, the second part is the measurement report.
+            msg = [client_identity.encode('utf-8'), json.dumps(dummy_report, indent=2).encode('utf-8')]
+            env_sim.send_multipart(msg) # send measurement
+            #print(identity + ': Send.')
+            #print(json_formatted_str)
+
+            if self.end_ts >= self.sim_end_ts:
+                # simulation ends. stop the while loop
+                break
+            
+            # simulation continues, wait for a new action.
+            # [Warning] always use a poll with timeout before receiving a msg (recv_multipart)! otherwise, it may stuck forever!
+            if poller.poll(timeout=30*1000):
+
+                msg = env_sim.recv_multipart() # receives an action
+                action_json = json.loads(msg[1])
+                if action_json["type"] == "env-action":
+                    #dummy simulator do not take any action.
+                    action_df = pd.json_normalize(action_json["action_list"]) 
+                    print(action_df)
+                else:
+                    print("Unkown MSG type, Expecting env-action!")
+                    error_msg = json.loads('{"type":"env-error", "error_msg": "Unkown MSG type, Expecting env-action!"}')
+                    msg[1]=json.dumps(error_msg, indent=2).encode('utf-8')
+                    env_sim.send_multipart(msg)
+                    break
+            else:
+                print("Timeout: No env-action received!")
+                error_msg = json.loads('{"type":"env-error", "error_msg": "Timeout: No env-action received!"}')
+                msg[1]=json.dumps(error_msg, indent=2).encode('utf-8')
+                env_sim.send_multipart(msg)
+                break
+            
+        env_sim.close()
+        print(env_identity + ': env_sim socket closed.')
+    
+    def run_one_interval(self):
+        """Running the simulator for one interval.
+
+        Returns:
+            json: the measurement list
+        """
+        # you can modify the tags for your own measurement
+        tags = {}
+
+        # all tags are optional
+        tags['unit']='mbps'
+        tags['start_ts']=self.start_ts
+        self.start_ts += self.interval
+        tags['end_ts']=self.end_ts
+        self.end_ts = self.start_ts + self.interval
+        tags['direction']='DL'
+        tags['group']='GMA'
+        tags['cid']='All'
+        
+
+        output1 = self.generate_dummy_measurement('rate', tags, self.num_users);
+        tags['cid']='Wi-Fi'
+        output2 = self.generate_dummy_measurement('rate', tags, self.num_users);
+        output3 = self.generate_dummy_measurement('qos_rate', tags, self.num_users);
+
+        tags['cid']='LTE'
+        output4 = self.generate_dummy_measurement('rate', tags, self.num_users);
+
+        tags['group']='PHY'
+        tags['cid']='LTE'
+        output5 = self.generate_dummy_measurement('max_rate', tags, self.num_users);
+
+        tags['cid']='Wi-Fi'
+        output6 = self.generate_dummy_measurement('max_rate', tags, self.num_users);
+
+        tags['unit']=''
+        tags['group']='GMA'
+        output7 = self.generate_dummy_measurement('split_ratio', tags, self.num_users);
+
+
+        tags['unit']='ms'
+        tags['group']='GMA'
+        tags['cid']='All'
+        output8 = self.generate_dummy_measurement('owd', tags, self.num_users);
+
+        merged_list = output1 + output2 + output3 + output4 + output5 + output6 + output7 + output8
+
+        #print(terminated)
+        #print(merged_list)
+        #json_formatted_str = json.dumps(merged_list, indent=2)
+        #print(json_formatted_str)
+
+
+        #data_recv = pd.json_normalize(merged_list)
+        #print(data_recv)
+
+        #data_recv_flat = data_recv.explode(column=['user', 'value'])
+        #print(data_recv_flat)
+        return merged_list
+
+        
+    def generate_dummy_measurement(self, name, tags, num_users):
+        """Generate random measurement.
+
+        Args:
+            name (str): the name of measurement
+            tags (dict): the tags added to the measurement, e.g., timestamps
+            num_users (int): the number of users
+
+        Returns:
+            json: the measurement results
+        """
+        #print (name)
+        #print (tags)
+        #print (user_num)
+
+        data = []
+        for id in range(num_users):
+            data.append([id, random.randint(3, 9)])
+        df = pd.DataFrame(data, columns=['user', 'value'])
+        for key, value in reversed(tags.items()):
+            df.insert(0, key, value)
+        df.insert(0,'name', name)
+        #print(df)
+
+        #tranform the json to a nested structure for lower overhead
+        group_by_list = list(tags.keys())
+        group_by_list.insert(0, 'name')
+        json_data = (df.groupby(group_by_list)
+                .apply(lambda x: x[['user', 'value', ]].to_dict(orient='list'))
+                .reset_index()
+                .rename(columns={0: ''})
+                .to_json(orient='records'))
+
+        json_object = json.loads(json_data)
+        
+        return json_object
          
+
+
          + +
          + + + + +
          + +
          + +
          +
          +
          + +
          + +
          + +
          + + + + +
          +
          + +
          + +
          + +

          +By Menglei Zhang +

          + +
          + +
          + +

          + + © Copyright 2023, Intel Labs Multi-Access Intelligence. +
          + +

          + +
          + +
          + +
          + +
          + +
          + +
          +
          + + +
          +
          +
          + + + + + +
          +
          + + \ No newline at end of file diff --git a/_sources/api/network_gym_client/northbound_interface.md b/_sources/api/network_gym_client/northbound_interface.md index 205e928..ea9eba8 100644 --- a/_sources/api/network_gym_client/northbound_interface.md +++ b/_sources/api/network_gym_client/northbound_interface.md @@ -1,27 +1,26 @@ --- -title: NorthboundInterface +title: NorthBoundClient --- -# NorthboundInterface +# NorthBoundClient -## network_gym_client.NorthboundInterface +## network_gym_client.NorthBoundClient ```{eval-rst} -.. autoclass:: network_gym_client.NorthboundInterface -.. autoclass:: network_gym_client.MeasurementReport +.. autoclass:: network_gym_client.NorthBoundClient ``` ### Methods ```{eval-rst} -.. autofunction:: network_gym_client.NorthboundInterface.connect -.. autofunction:: network_gym_client.NorthboundInterface.send -.. autofunction:: network_gym_client.NorthboundInterface.recv +.. autofunction:: network_gym_client.NorthBoundClient.connect +.. autofunction:: network_gym_client.NorthBoundClient.send +.. autofunction:: network_gym_client.NorthBoundClient.recv ``` ### Additional Methods ```{eval-rst} -.. autofunction:: network_gym_client.NorthboundInterface.process_measurement +.. autofunction:: network_gym_client.NorthBoundClient.process_measurement ``` diff --git a/_sources/api/network_gym_env.md b/_sources/api/network_gym_env.md index 0f8b496..abac711 100644 --- a/_sources/api/network_gym_env.md +++ b/_sources/api/network_gym_env.md @@ -4,26 +4,32 @@ title: network_gym_env # network_gym_env +```{toctree} +:hidden: +network_gym_env/custom_env +network_gym_env/dummy_sim + +``` ```{mermaid} flowchart LR subgraph network_gym_env -southbound_api[[southbound Interface]] -simulator -emulator -testbed +southbound_interface[[southbound_interface]] +southbound_interface <--> configuration +southbound_interface <--> simulator +southbound_interface <-..-> emulator +southbound_interface <-..-> testbed end ``` +The NetworkGymEnv comprises two core components: the **Environment Configuration** and the **NetworkGym Simulator**. These components establish a connection with the Server via the designated **SouthBound Interface**. -NetworkGymEnv includes the two components, a **southbound interface**, and a **NetworkGym simulator**. -- The **southbound interface** connects the env to the server, receives the environment parameters, communicate network stats and policy between env and server. -- At present, we only support **NetworkGym simulator**. The emulator and testbed option will be released in the future. +- The **SouthBound Interface** serves as the bridge between the environment and the server. It facilitates communication of network configurations, network statistics, and policies between the environment and the server. +- The **Environment Configuration** module keeps the connection alive by periodically dispatching an "Env Hello" message to the server. Upon receiving a simulation start request, it disengages from the server and initiates the simulator. +- Upon activation, the ns-3 based **NetworkGym Simulator** establishes a connection with the server and instigates the exchange of measurements. Plans for incorporating an emulator and testbed alternative are earmarked for future release. +## NetworkGym UML Diagram +![network_gym_uml](network_gym_uml.png) -```{note} -TODO: provide southbound interface example code -``` -## NetworkGym UML Class Diagram -![network_gym_uml](network_gym_uml.png) \ No newline at end of file +The UML diagram depicts the flow of messages for a complete cycle within a simulated network environment session. In this cycle, a Client initiates the process by sending an "Env Start" message to request the launch of a network environment. Upon receiving this message, the Env_Config component triggers the simulator to generate "Env Measurement" data and receive incoming "Action" message. Once the simulation concludes, the Env_Config component dispatches an "Env End" message to signal the termination of the Client-to-Env mapping. Importantly, only one of the components, Env_Config or Env_Sim, can establish a connection with the Server at any given moment, and these connections cannot occur simultaneously. From the Server's standpoint, it treats both the Env_Config and Env_Sim components equivalently, identifying them both as the Env entity. \ No newline at end of file diff --git a/_sources/api/network_gym_env/custom_env.md b/_sources/api/network_gym_env/custom_env.md new file mode 100644 index 0000000..dcb4ac1 --- /dev/null +++ b/_sources/api/network_gym_env/custom_env.md @@ -0,0 +1,17 @@ +--- +title: Custom Environment +--- + +# Custom Environment + +## network_gym_env.CustomEnv + +```{eval-rst} +.. autoclass:: network_gym_env.CustomEnv +``` + +### Methods + +```{eval-rst} +.. autofunction:: network_gym_env.CustomEnv.run +``` diff --git a/_sources/api/network_gym_env/dummy_sim.md b/_sources/api/network_gym_env/dummy_sim.md new file mode 100644 index 0000000..983c2cb --- /dev/null +++ b/_sources/api/network_gym_env/dummy_sim.md @@ -0,0 +1,19 @@ +--- +title: Dummy Simulator +--- + +# Dummy Simulator + +## network_gym_env.DummySim + +```{eval-rst} +.. autoclass:: network_gym_env.DummySim +``` + +### Methods + +```{eval-rst} +.. autofunction:: network_gym_env.DummySim.generate_dummy_measurement +.. autofunction:: network_gym_env.DummySim.start_simulation +.. autofunction:: network_gym_env.DummySim.run_one_interval +``` diff --git a/_sources/content/overview.md b/_sources/content/overview.md index c34dcbd..f6c6e89 100644 --- a/_sources/content/overview.md +++ b/_sources/content/overview.md @@ -18,10 +18,10 @@ northbound <--> southbound[[southbound_interface]] end subgraph network_gym_env -southbound_interface -simulator -emulator -testbed +southbound_interface <--> configuration +southbound_interface <--> simulator +southbound_interface <-..-> emulator +southbound_interface <-..-> testbed end agent <--> gymnasium.env diff --git a/_sources/index.md b/_sources/index.md index 1173215..7400a90 100644 --- a/_sources/index.md +++ b/_sources/index.md @@ -79,6 +79,8 @@ environments/mx_network_slicing tutorials/handling_time_limits tutorials/training_agents +tutorials/customizing_observation_and_reward +tutorials/implementing_custom_environment Gymnasium Tutorial ``` diff --git a/_sources/tutorials/customizing_observation_and_reward.md b/_sources/tutorials/customizing_observation_and_reward.md new file mode 100644 index 0000000..270c38f --- /dev/null +++ b/_sources/tutorials/customizing_observation_and_reward.md @@ -0,0 +1,98 @@ +--- +title: Customizing Observation and Reward +--- + +# Customizing Observation and Reward +You have the flexibility to tailor the observation and reward functions to suit your experimentation goals within the environments we offer. To illustrate, we present a comprehensive example of network statistics measurement. The observation and reward functions can be constructed utilizing this information. + +```python + cid direction end_ts group name start_ts user value +0 LTE DL 6000 GMA ap_id 5900 [0, 1, 2, 3] [255.0, 255.0, 255.0, 255.0] +1 Wi-Fi DL 6000 GMA ap_id 5900 [0, 1, 2, 3] [0.0, 0.0, 1.0, 1.0] +2 All DL 6000 GMA delay_violation 5900 [0, 1, 2, 3] [59.64912280701754, 100.0, 100.0, 100.0] +3 All DL 6000 GMA max_owd 5900 [0, 1, 2, 3] [90.0, 341.0, 153.0, 379.0] +4 LTE DL 6000 GMA max_owd 5900 [0, 1, 2, 3] [90.0, 341.0, 153.0, 379.0] +5 Wi-Fi DL 6000 GMA max_owd 5900 [0, 1, 2, 3] [11.0, 11.0, 1.0, 2.0] +6 LTE DL 6000 PHY max_rate 5900 [0, 1, 2, 3] [35.0, 35.0, 29.0, 35.0] +7 Wi-Fi DL 6000 PHY max_rate 5900 [0, 1, 2, 3] [78.0, 78.0, 78.0, 78.0] +8 All DL 6000 GMA measurement_ok 5900 [0, 1, 2, 3] [1.0, 1.0, 1.0, 1.0] +9 All DL 6000 GMA missed_action 5900 [0, 1, 2, 3] [0.0, 0.0, 0.0, 0.0] +10 All DL 6000 GMA owd 5900 [0, 1, 2, 3] [37.0, 296.0, 128.0, 353.0] +11 LTE DL 6000 GMA owd 5900 [0, 1, 2, 3] [38.0, 292.0, 127.0, 359.0] +12 Wi-Fi DL 6000 GMA owd 5900 [0, 1, 2, 3] [5.0, 2.0, 0.0, 0.0] +13 LTE DL 6000 GMA qos_marking 5900 [0, 1, 2, 3] [1.0, 1.0, 1.0, 1.0] +14 Wi-Fi DL 6000 GMA qos_marking 5900 [0, 1, 2, 3] [1.0, 1.0, 0.0, 0.0] +15 All DL 6000 GMA qos_rate 5900 [0, 1, 2, 3] [0.0, 0.0, 0.0, 0.0] +16 LTE DL 6000 GMA qos_rate 5900 [0, 1, 2, 3] [0.0, 0.0, 0.0, 0.0] +17 Wi-Fi DL 6000 GMA qos_rate 5900 [0, 1, 2, 3] [0.0, 0.0, 0.0, 0.0] +18 All DL 6000 GMA rate 5900 [0, 1, 2, 3] [13.24, 18.464, 6.968, 19.28] +19 LTE DL 6000 GMA rate 5900 [0, 1, 2, 3] [10.8, 7.776, 6.272, 7.776] +20 Wi-Fi DL 6000 GMA rate 5900 [0, 1, 2, 3] [4.76, 18.464, 1.856, 7.2] +21 LTE DL 6000 GMA split_ratio 5900 [0, 1, 2, 3] [26.0, 15.0, 24.0, 20.0] +22 Wi-Fi DL 6000 GMA split_ratio 5900 [0, 1, 2, 3] [6.0, 17.0, 8.0, 12.0] +23 LTE DL 6000 GMA traffic_ratio 5900 [0, 1, 2, 3] [69.0, 30.0, 77.0, 52.0] +24 Wi-Fi DL 6000 GMA traffic_ratio 5900 [0, 1, 2, 3] [31.0, 70.0, 23.0, 48.0] +25 All DL 6000 GMA tx_rate 5900 [0, 1, 2, 3] [25.09, 29.04, 7.434, 18.004] +``` + +The network statistics measurement contains 8 columns, with each field's explanation provided below: + +| Column | Description | +| ----- | ---- | +| cid | Connection ID: Wi-Fi, LTE, or All, where All signifies the combination of all links. | +| direction | DL: downlink; UL: uplink. | +| end_ts | The ending timestamp of this measurement. | +| group | GMA: GMA convergence layer measurement; PHY: measurement from the PHY layer. | +| name | Name of the measurement.| +| start_ts | The starting timestamp of this measurement. | +| user | A list of user IDs for this measurement. | +| value | A list of values for this measurement (user and value should have the same size). | + +Now, we elaborate on each row: + +| Row | Name | Description | +| ----- | ---- | ---- | +| 0,1 | ap_id | LTE cell ID and Wi-Fi access point ID as measured by each user, encompassing LTE and Wi-Fi links. *(NOTE: LTE isn't measured, 255 is the default value.)*| +| 2 | delay_violation | One-way delay violation percentage (%) as measured by each user; delay bound (delay_bound_ms) can be configured in the JSON file. *(NOTE: Only for QoS use case)* | +| 3,4,5 | max_owd | Maximum one-way delay measured by each user, involving LTE link, Wi-Fi link, and ALL. *(ALL: after reordering out-of-order packets from both links)*. | +| 6,7 | max_rate | LTE/Wi-Fi link capacity as measured by each user. | +| 8 | measurement_ok | 1: Measurement is valid; 0: An issue exists with the measurement.| +| 9 | missed_action | Count of missed actions by the environment. *(NOTE: Applicable in testbed environment where RL agent might take too long to compute an action. Simulator waits for action by default in simulations, therefore should always be 0.)* | +| 10,11,12 | owd | One-way delay measured by each user, involving LTE link, Wi-Fi link, and ALL. *(ALL: after reordering out-of-order packets from both links)*. | +| 13,14 | qos_marking | 1: QoS user; 0: Non-QoS user.| +| 15,16,17 | qos_rate | QoS delivery rate (output traffic throughput meeting QoS requirement) as measured by each user, comprising traffic over LTE link, Wi-Fi link, and ALL (combined). | +| 18,19,20 | rate | Delivery rate (output traffic throughput) measured by each user, including traffic over LTE link, Wi-Fi link, and ALL (combined).| +| 21,22 | split_ratio | Traffic split ratio (in range of [0, 32]) as measured by each user, covering LTE link, Wi-Fi link. LTE split ratio + Wi-Fi split ratio equals 32. | +| 23,34 | traffic_ratio | Traffic ratio (%) sending to each link measured by each user, involving LTE link, Wi-Fi link. traffic_ratio might not equal split_ratio, since the system takes sine time to apply the new action. | +| 25 | tx_rate | Load (input traffic throughput) measured by each user. | + + +## Working with the DataFrame Format + +To retrieve a specific network statistic, such as `rate``, you can utilize the following code: +```python +df_rate = df[df['name'] == 'rate'].reset_index(drop=True) # get the rate +``` + +To isolate the link for the rate measurement, such as `All`, you can execute the subsequent code: +```python +df_rate = df_rate[df_rate['cid'] == 'All'].reset_index(drop=True) # get the rate for the all link combined +``` + +Lastly, employ the `explode` function to transform the nested data frame into its original structure: + +```python +df_rate = df_rate.explode(column=['user', 'value']) #keep the flow rate. +``` +Here is the final output: +```python + cid direction end_ts group name start_ts user value +0 All DL 6000 GMA rate 5900 0 13.24 +0 All DL 6000 GMA rate 5900 1 18.464 +0 All DL 6000 GMA rate 5900 2 6.968 +0 All DL 6000 GMA rate 5900 3 19.28 +``` + +## Customizing Observation + +## Customizing Reward \ No newline at end of file diff --git a/_sources/tutorials/implementing_custom_environment.md b/_sources/tutorials/implementing_custom_environment.md new file mode 100644 index 0000000..5f683f0 --- /dev/null +++ b/_sources/tutorials/implementing_custom_environment.md @@ -0,0 +1,12 @@ +--- +title: Implementing Custom Environment +--- + +# Implementing Custom Environment + +```{figure} network_gym_overview2.png +--- +width: 60% +--- +``` +NetworkGym provides a range of network Environments, it also supports Custom Environment via the SouthBound (SB) Interface. \ No newline at end of file diff --git a/_sources/tutorials/training_agents.md b/_sources/tutorials/training_agents.md index 7829498..42dbb3f 100644 --- a/_sources/tutorials/training_agents.md +++ b/_sources/tutorials/training_agents.md @@ -4,11 +4,16 @@ title: Training Agents # Training Agents +```{figure} network_gym_overview.png +--- +width: 40% +--- +``` NetworkGym provides a range of network agents, each designed to fulfill specific roles when interacting with NetworkGym Environments. ## System Default Agent -The System Default Agent enables the collection of data for offline training purposes. To activate system default algorithms, a agent simply sends an "empty" action to the environment. You can use the [demo](https://github.com/IntelLabs/networkgym/blob/main/network_gym_client_demo.py) code and modify the action as shown below: +The System Default Agent enables the collection of data for offline training purposes. To activate system default algorithms, a agent simply sends an "empty" action to the environment. You can use the [demo](https://github.com/IntelLabs/networkgym/blob/main/start_client_demo.py) code and modify the action as shown below: ```python # action = env.action_space.sample() # agent policy that uses the observation and info diff --git a/api/network_gym_client.html b/api/network_gym_client.html index cd3ac13..eb183c3 100644 --- a/api/network_gym_client.html +++ b/api/network_gym_client.html @@ -147,20 +147,24 @@
        • network_slicing Adapter
        • -
      • NorthboundInterface
        • +
      • NorthBoundClient
    • network_gym_server
      • -
    • network_gym_env
      • +
    • network_gym_env +

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

    Environments

      -
    • Traffic Management
        +
      • Traffic Management
        • -
      • Network Slicing @@ -328,6 +335,10 @@

        D

        +
        @@ -343,6 +354,8 @@

        E

        G

        -

        M

        - - -
        -

        N

          @@ -407,7 +412,7 @@

          N
        @@ -423,7 +428,7 @@

        O

        P

        @@ -431,11 +436,15 @@

        P

        R

        @@ -443,10 +452,12 @@

        R

        S

        diff --git a/index.html b/index.html index 9510302..6761c5f 100644 --- a/index.html +++ b/index.html @@ -146,20 +146,24 @@
      • network_slicing Adapter
        • -
      • NorthboundInterface
        • +
      • NorthBoundClient
      • network_gym_server
        • -
      • network_gym_env
        • +
      • network_gym_env +

        • Environments

          -
        • Traffic Management
            +
          • Traffic Management
            • -
          • Network Slicing
              +
            • Network Slicing
              • @@ -168,6 +172,8 @@

              Development

              diff --git a/objects.inv b/objects.inv index 2b325254d5944709b4e4b7c641eadaa677e4d2e3..000a25f5bd1194690618fec94d99f6528de806dc 100644 GIT binary patch delta 1236 zcmV;_1S|WH3BU@Fe1GF66y579NF`g@m2K}l-CfC4rRUL}s#&Tq!ZtNLJObJ6S>-qK zd-;-p&|668;pxuKhRZnzxc30!SY1T&EjnCTgebvWtu`?{DV%`Q^$t-eCHNh{6It_4 zu(?ocrV&2-bYGAF+s`z?^BKx5&SmT7s%sXM1RHMcF>RfJYoSbKl@Nt zFjEuB6GYfRLZuS2OBG%STkX<~5V=UyaVJt9M}%wWy%wDJ-;hP@Bu>uor>seWB_5ex zW8kF1r`w=?FiZQ+EbV={o}Scl?ZzaI6%w4|vE{a$?vx0GrHmMcGEFecr}-Mx2}bAF z;*yxhKZ&dGrGJhAhvTfQ-yz+6CTkl$7y1*t$13~q<*I0xlpw+n(@&Rm8)0Y(hbUJE ztfICXk3KDKE2HLu{|^;!z9J`~8?SFDAlV%vNq)TL8YK7vllYYd%y%6(_f#VoQ2Jd+ zotKoTIj&I+_Ag0?XsK5$C!g;SH|2h#+0Sx=<+QU3aDN%C4iy(2nkqtHg?LH5w-_fB zZojlFR$-SSub^tIcLh2$D+DP@DO5PRkUfb^K&taT;whW95sOqTGn&QJvu);<%x^OD zITo{+bdJHyB3)4BExz$}>(u3Z1@*s&!x_)1F%RyhHuIdhPo)yf^PNWru2B{k^}g{N zV5=nc6@SuL%!ib*>y|K|zSzX+t1I(E=^W%ZF_vcsjeo08CDO|f36AYnoguw3yi_nz zCe|3Z-@8V_*LKB`9pSjG?9a0Z^d?OhJU?z8o*scz@m(yC!Lu#dK=3Xl zGzJd>rDpf-B7Vg?NfZv;6<9}dM_A-;dob)cet+P`cMg#IO)%|sGM>fxt^sYhu`0^D z&=dj9ix#MXjHE5yc>-(G_(28X;naCIf~6=Bpi3c`;PE4%dyfXtJ_iCvNevN{qFar@ zuoHM7+nxceN0^l%iQue9XjS_@{H95O%oM=Zx~1suqq&F;XNryJE`@CmkiOK&kiCQm zQ-7*SlCWNM6z(uJsJNP+T(g&`RZ{C>bky@v<38^PA8bF^+Wv}P=WM7FxZe{LX1VLt zkVi)ha(Kv2Lepw;zvwA&3KdT&RV3UX^R*j%7F+Pe27+xKSjj1+n9@}@q>%Ks&R(0^NeCivnSUi8_)AM=0Z(C;Zh-?GEnHMr`v z!#|h*15c{d(LABK5wtR9ZYoY79!}@}WcFvpkce)yr7QB|scaACm%=mwg9SCBoMx8E zbdl~#^VONYQxhKJmu77jtrp_F`59DUxvW}5o&|qg*&TSmOdx$tCqK+Iiu~BT2Y=64 zVwj*KfL!0*@EKEa%C?e;*rDNc4U3MVxZ{-KT#cY{*ilqA+5ulNMS@|VL@7fn{P*vF zB#7%*$6taPmq)~>t2$~gYG}!))fi~o?3k_X|D-o7P^mt{2&6JUMzd9Fd`{4-Q@pfm zQ=2lbek|Kv_*nZ;Vpjh-i>rw_>P=Q^5EWrKqJ$A8-Sng+`rLn#O-0r!6rU4J$UYeU~1Ha^>@ax-7-l&jh5_M56@Jh$PiRkp4eC6ph=ztPrJgsv^@ zqFfz7d(}6<;5XE=GD?&8|DodUI}!@Bc)6nhxjT4@{VDTvsBC*G_$EEkC`Qz z@ABkRM$Bx|DFbE_=?s@w_=fwoQCH(FsN)XqPPk7Ey5$d7*LitgNM93&>%2T|9=A_O zY5&0&B!A|zD+vtxEuzdvkAg(;V_DCBg`J`roMugBq~;8ZTwiy(o1`E3>S~J87IDUX zP!4Bl{1io1xS=ZY*Vh;U*Y8GD0vSk~8)^jWwO3IA(ZgvRJb)#;^&@k|cSdym0O(e; zpay5uxO?R1x+{w{ocImYz27#Yj%L>#Vy3xrgMT-d0AuE3Lb~PFWq1cHMm5}23Le{5 zt%EnO(wOV4+H1MZ{hr3rM;k}4HeT?XlXtbm`#mLY(gD31(rhol93PTF8L!42ltnU} zOvN^0nz|bj|FaK$7O&`wO+>G0Vj(A&5*t8I)RODSKpKVQBC;%p2-7$*gwQnov?-l% zJ%4IXcbost0faf?6NPhAc^0$9f2F_j((e%^-;~BuHx(CATmCu!?ew%t8LcNZwNs^I zrnce;;_h~CZ|454=n_%Swscv4Z1Z(zx)sI==rq&-J5997c#+Lc>+{(GIORUPU3)uQ zwGhwopH7jM$|_}~UGU)&JfKGi1d+ysKY!xK$ogaRj9x$imgI;qH(x+-0w%)bCnetwork_slicing Adapter
            • -
          • NorthboundInterface
            • +
          • NorthBoundClient
        • network_gym_server
          • -
        • network_gym_env
          • +
        • network_gym_env +

        Environments

        Environments

          -
        • Traffic Management
            +
          • Traffic Management
            • -
          • Network Slicing
              +
            • Network Slicing
              • @@ -168,6 +172,8 @@

              Development

              diff --git a/searchindex.js b/searchindex.js index 9119e2f..2062c0d 100644 --- a/searchindex.js +++ b/searchindex.js @@ -1 +1 @@ -Search.setIndex({"docnames": ["README", "api/network_gym_client", "api/network_gym_client/adapter", "api/network_gym_client/adapter/network_slicing", "api/network_gym_client/adapter/nqos_split", "api/network_gym_client/adapter/qos_steer", "api/network_gym_client/env", "api/network_gym_client/northbound_interface", "api/network_gym_env", "api/network_gym_server", "content/motivation", "content/overview", "content/quickstart", "environments/mx_network_slicing", "environments/mx_network_slicing/cellular_network_slicing", "environments/mx_traffic_management", "environments/mx_traffic_management/mx_qos_traffic_steering", "environments/mx_traffic_management/mx_traffic_splitting", "index", "old-index", "tutorials/handling_time_limits", "tutorials/training_agents"], "filenames": ["README.md", "api/network_gym_client.md", "api/network_gym_client/adapter.md", "api/network_gym_client/adapter/network_slicing.md", "api/network_gym_client/adapter/nqos_split.md", "api/network_gym_client/adapter/qos_steer.md", "api/network_gym_client/env.md", "api/network_gym_client/northbound_interface.md", "api/network_gym_env.md", "api/network_gym_server.md", "content/motivation.md", "content/overview.md", "content/quickstart.md", "environments/mx_network_slicing.md", "environments/mx_network_slicing/cellular_network_slicing.md", "environments/mx_traffic_management.md", "environments/mx_traffic_management/mx_qos_traffic_steering.md", "environments/mx_traffic_management/mx_traffic_splitting.md", "index.md", "old-index.rst", "tutorials/handling_time_limits.md", "tutorials/training_agents.md"], "titles": ["Build Documentation Website", "network_gym_client", "Adapter", "network_slicing Adapter", "nqos_split Adapter", "qos_steer Adapter", "Env", "NorthboundInterface", "network_gym_env", "network_gym_server", "Motivation", "Overview", "Quickstart", "Network Slicing", "Cellular Network Slicing", "Traffic Management", "Multi-Access QoS Traffic Steering", "Multi-Access Traffic Splitting", "NetworkGym: Democratizing Network AI Research and Development", "Welcome to Network Gym\u2019s documentation!", "Handling Time Limits", "Training Agents"], "terms": {"option": [0, 1, 6, 8, 12, 14, 16, 17, 20, 21], "creat": [0, 11, 12, 18, 21], "new": [0, 12], "virtual": 0, "python": [0, 12, 18], "environ": [0, 1, 2, 3, 5, 6, 7, 8, 9, 10, 13, 14, 15, 16, 17, 18, 20, 21], "python3": 0, "m": [0, 14], "venv": 0, "sphinx_venv": 0, "sourc": [0, 2, 3, 4, 5, 6, 7, 18, 21], "bin": 0, "activ": [0, 11, 12, 21], "requir": [0, 10, 16], "librari": 0, "pip": 0, "r": [0, 14, 17], "txt": 0, "In": [0, 11, 12, 13, 14, 15, 16, 17, 20], "doc": 0, "folder": 0, "run": [0, 6, 12, 14, 16, 17, 20], "follow": [0, 1, 6, 14, 16, 17], "command": 0, "export": 0, "static": 0, "_build": 0, "html": 0, "make": [0, 12], "clean": [0, 18, 21], "network_gym_cli": [0, 2, 3, 4, 5, 12, 14, 16, 17, 20], "The": [0, 1, 6, 8, 9, 11, 12, 14, 16, 17, 18, 20, 21], "contain": [0, 6, 12, 14], "modul": [0, 19], "rm": 0, "sphinx": 0, "apidoc": 0, "o": [0, 12, 18], "d": 0, "2": [0, 6, 14, 16, 17, 18], "markdown": 0, "e": [0, 1, 6, 10, 14, 16, 17, 18, 20], "g": [0, 1, 6, 10, 16, 17, 18], "eval": 0, "autoclass": 0, "env": [0, 1, 3, 4, 5, 7, 8, 12, 14, 16, 17, 20, 21], "networkgymcli": [1, 11, 14, 16, 17], "includ": [1, 8, 9, 12, 18, 21], "three": [1, 11, 14, 16, 17, 18, 20], "compon": [1, 8, 9], "custom": [1, 6, 11, 12, 14, 16, 17, 18, 20], "gymnasium": [1, 2, 6, 11, 12, 21], "adapt": [1, 6, 11, 14, 16, 17], "northbound": [1, 6, 7, 9, 11, 12], "interfac": [1, 6, 7, 8, 9, 12, 18], "inheret": 1, "class": [1, 3, 4, 5, 6, 7, 12], "commun": [1, 6, 8, 11, 12], "agent": [1, 3, 4, 5, 6, 11, 12, 13, 14, 15, 16, 17, 20], "us": [1, 6, 10, 12, 13, 14, 16, 17, 18, 20, 21], "standard": [1, 18], "exchang": [1, 12], "ob": [1, 2, 6, 12, 20], "reward": [1, 2, 6, 12, 20], "action": [1, 2, 3, 4, 5, 6, 13, 15, 18, 20, 21], "reset": [1, 6, 12, 20], "step": [1, 6, 14, 16, 17, 20], "function": [1, 3, 4, 5, 12, 14, 16, 17], "import": [1, 12, 14, 16, 17, 20], "gym": [1, 6, 7], "def": 1, "__init__": 1, "self": [1, 2, 3, 4, 5, 6, 7], "arg1": 1, "arg2": 1, "super": 1, "return": [1, 2, 3, 4, 5, 6, 7, 12], "observ": [1, 3, 4, 5, 6, 18, 21], "termin": [1, 6, 12, 14, 16, 17], "truncat": [1, 6, 12, 14, 16, 17], "info": [1, 6, 12, 20, 21], "seed": [1, 6], "none": [1, 6], "transform": [1, 2, 11], "data": [1, 2, 5, 7, 11, 12, 14, 16, 18, 21], "format": [1, 2, 11, 12], "from": [1, 2, 3, 4, 5, 6, 7, 12, 14, 20, 21], "network_gym": [1, 2, 6], "other": [1, 14, 16, 17, 20], "wai": [1, 6, 17], "around": [1, 18], "network": [1, 2, 3, 4, 5, 6, 7, 8, 11, 12, 16, 17], "stat": [1, 2, 3, 4, 5, 7, 8], "chang": [1, 20], "polici": [1, 2, 3, 4, 5, 7, 8, 12, 21], "connect": [1, 6, 7, 8, 9, 11, 12, 15, 16, 17, 18], "client": [1, 6, 7, 9, 12, 14, 16, 17, 18], "server": [1, 6, 7, 8, 9, 12, 14, 16, 17, 18], "configur": [1, 2, 3, 4, 5, 6, 7, 11, 12, 14, 16, 17, 18, 20], "paramet": [1, 2, 3, 4, 5, 6, 7, 8, 14, 16, 17, 18, 20], "between": [1, 2, 6, 8, 11, 17, 18], "envrion": [1, 9], "config_json": [2, 3, 4, 5, 6, 7, 12, 14, 16, 17, 20], "thi": [2, 3, 4, 5, 6, 11, 12, 13, 14, 15, 16, 17, 18, 20], "i": [2, 3, 4, 6, 7, 11, 12, 14, 16, 17, 18, 20, 21], "an": [2, 6, 9, 11, 12, 14, 16, 17, 18, 20, 21], "It": [2, 6, 11, 12, 14, 16, 17, 18, 20], "measur": [2, 3, 4, 5, 6, 7, 11, 12, 14], "json": [2, 3, 4, 5, 6, 7, 12, 14, 16, 17, 20], "space": [2, 3, 4, 5, 6, 18], "also": [2, 6, 10, 12, 14, 16, 17, 18, 20], "can": [2, 6, 11, 12, 14, 16, 17, 18, 20, 21], "appli": 2, "initi": [2, 3, 4, 5, 6, 7, 14], "file": [2, 3, 4, 5, 6, 7, 12, 14, 16, 17, 20], "wandb_log": 2, "send": [2, 6, 7, 12, 18, 21], "log": [2, 6, 17], "inform": [2, 6, 12, 16, 17, 18], "wandb": [2, 12], "df_to_dict": 2, "df": [2, 3, 4, 5], "descript": 2, "datatyp": 2, "panda": [2, 3, 5, 7], "datafram": [2, 3, 4, 5, 7], "dictionari": [2, 6, 12], "object": [2, 6, 14], "string": 2, "descritpt": 2, "dict": [2, 6, 12], "convert": 2, "base": [3, 4, 5, 11, 12, 17, 18], "get_observ": [3, 4, 5, 14, 16, 17], "prepar": [3, 4, 5], "should": [3, 4, 5, 6, 12, 14], "same": [3, 4, 5, 14], "number": [3, 4, 5, 6, 12, 14, 16, 17, 20], "featur": [3, 4, 5, 14, 16, 17, 18], "defin": [3, 4, 5, 6, 12, 14, 16, 17, 18, 21], "get_observation_spac": [3, 4, 5], "df_list": [3, 5, 7], "get_reward": [3, 4, 5, 14, 16, 17], "pd": [3, 4, 5], "get_polici": [3, 4, 5], "rl": [3, 4, 5, 6, 12, 18, 21], "No": [3, 12, 21], "yet": 3, "get_action_spac": [3, 4, 5], "get": [3, 4, 5, 6], "get_rbg_siz": 3, "bandwidth": [3, 16], "comput": [3, 12, 14, 16, 17, 18], "resourc": [3, 13, 14, 16, 17, 18], "block": [3, 14], "group": [3, 14], "size": 3, "bandwith": 3, "rb": [3, 14], "code": [3, 8, 18, 21], "coppi": 3, "ns3": [3, 10], "pf": 3, "type": [3, 6, 12, 18], "0": [3, 4, 6, 12, 14, 16, 17, 20], "alloc": [3, 13, 14, 18], "rbg": 3, "int": [3, 6, 7], "resouc": 3, "netowrk_util": [4, 17], "throughput": [4, 14, 16, 17], "delai": [4, 6, 14, 16, 17, 21], "alpha": 4, "5": [4, 14, 16, 17], "calcul": [4, 5, 16], "util": [4, 11, 14, 16, 17], "specifi": [4, 6, 12, 14, 16, 17, 20], "valu": [4, 6, 12, 14, 16, 17], "balanc": 4, "default": [4, 5, 6, 12, 14, 18], "float": 4, "repres": [4, 12, 14, 16, 17], "bit": 4, "per": [4, 5, 14, 16, 17, 20], "second": [4, 12, 16, 17, 20], "metric": [4, 12], "list": [5, 6, 7, 12], "calculate_wifi_qos_user_num": [5, 16], "qos_rat": [5, 16], "qo": [5, 12, 15, 18], "user": [5, 6, 12, 14, 15, 16, 17, 18, 21], "over": [5, 15, 16, 17], "wi": [5, 15, 16, 17], "fi": [5, 15, 16, 17], "rate": [5, 14, 16, 17], "doubl": 5, "id": [6, 7], "networkgym": [6, 7, 14, 16, 17, 20, 21], "initil": 6, "intern": [6, 12], "state": [6, 12, 21], "": [6, 12, 14, 16, 17, 20, 21], "prng": 6, "np_random": 6, "If": [6, 12, 20], "doe": [6, 18], "alreadi": [6, 16], "have": [6, 12, 14, 16, 17, 18, 20, 21], "pass": 6, "chosen": 6, "some": [6, 12], "entropi": 6, "timestamp": 6, "dev": 6, "urandom": 6, "howev": 6, "ha": [6, 12, 14, 16, 17, 20], "you": [6, 11, 12, 14, 16, 17, 20, 21], "integ": 6, "even": 6, "exist": 6, "usual": 6, "want": 6, "right": [6, 14], "after": [6, 12, 14, 16, 17, 20], "been": [6, 12, 21], "never": 6, "again": 6, "pleas": [6, 18], "refer": [6, 12, 18, 20, 21], "minim": [6, 17], "exampl": [6, 8, 12], "abov": [6, 18], "see": [6, 11, 12, 14, 16, 17], "paradigm": 6, "addit": [6, 12, 18], "how": [6, 11, 12, 16, 18, 20], "depend": 6, "specif": [6, 10, 11, 14, 18, 20, 21], "obstyp": 6, "auxiliari": 6, "complement": 6, "analog": 6, "one": [6, 12, 14, 17, 20], "timestep": [6, 12], "dynam": [6, 18], "gamsim": 6, "check": 6, "last": [6, 20], "episod": [6, 12, 20], "done": [6, 12], "acttyp": 6, "provid": [6, 8, 11, 12, 16, 17, 18, 20, 21], "updat": [6, 12, 15, 17], "element": [6, 11, 12], "observation_spac": [6, 12], "next": [6, 12], "due": 6, "supportsfloat": 6, "result": [6, 12, 14, 16, 17, 20], "take": [6, 12, 14, 16], "bool": [6, 7], "whether": 6, "reach": [6, 12, 16], "under": 6, "mdp": 6, "task": 6, "which": [6, 11, 12, 14, 16, 18], "posit": [6, 14, 16, 17], "neg": 6, "goal": [6, 16, 17], "move": 6, "lava": 6, "sutton": 6, "barton": 6, "gridworld": 6, "true": [6, 12], "need": [6, 10, 12, 14], "call": [6, 12], "condit": 6, "outsid": 6, "scope": 6, "satisfi": 6, "typic": 6, "timelimit": [6, 12], "could": [6, 12], "indic": [6, 7, 14, 16], "physic": [6, 18], "go": 6, "out": [6, 12], "bound": [6, 12], "end": [6, 12, 18, 20], "prematur": 6, "befor": 6, "diagnost": 6, "help": [6, 12], "debug": 6, "learn": [6, 10, 18, 21], "raw": 6, "flag": 6, "deprec": 6, "A": [6, 12, 14, 16, 17, 18, 20], "boolean": 6, "case": [6, 12, 14, 18], "further": 6, "undefin": 6, "wa": 6, "remov": 6, "openai": 6, "v26": 6, "favor": 6, "signal": [6, 12], "mai": [6, 12, 20], "emit": 6, "differ": [6, 11, 12, 16, 18, 20], "reason": [6, 18], "mayb": 6, "underli": 6, "solv": 6, "successfulli": 6, "certain": [6, 14, 16, 17, 20], "exceed": [6, 12], "simul": [6, 8, 10, 11, 18], "enter": 6, "invalid": 6, "action_spac": [6, 12, 20, 21], "correspond": 6, "valid": [6, 7, 12], "all": [6, 12, 13, 14, 16, 17], "For": [6, 12, 16, 18], "discret": [6, 12], "give": 6, "mean": [6, 18], "ar": [6, 12, 14, 16, 17, 18], "two": [6, 8, 9, 11, 20], "1": [6, 12, 14, 16, 17, 20], "box": [6, 12, 14, 16, 17], "3": [6, 11, 12, 14, 16, 17, 18, 20], "4028234663852886e": 6, "38": 6, "4": [6, 14], "float32": [6, 14, 16, 17], "shape": [6, 12, 14, 16, 17], "denot": [6, 14, 16, 17, 20], "arrai": [6, 12, 14, 21], "we": [6, 8, 9, 12, 14, 16, 17, 18, 20], "well": 6, "high": [6, 17, 18], "8000002e": 6, "00": 6, "4028235e": 6, "1887903e": 6, "01": 6, "dtype": 6, "low": [6, 17], "northbound_interface_cli": 6, "traslat": 6, "dataformat": 6, "sever": [7, 18], "repli": 7, "measurementreport": 7, "ok_flag": 7, "structur": 7, "store": [7, 12], "report": 7, "whehter": 7, "stst": 7, "recv": 7, "receiv": [7, 8, 12, 16, 17], "messag": 7, "process_measur": 7, "reply_json": 7, "networkgymenv": [8, 12, 14, 16, 17, 20], "southbound": [8, 9, 11], "At": [8, 9, 11, 14, 16, 17, 18, 20], "present": [8, 9, 11, 16], "onli": [8, 16], "support": [8, 11, 12, 16, 18], "emul": [8, 11], "testb": [8, 11], "releas": [8, 9], "futur": 8, "todo": 8, "networkgymserv": [9, 11, 12], "manag": [9, 16, 17, 18], "when": [9, 14, 16, 17, 18, 21], "select": [9, 11, 12, 14, 15, 16, 17], "idl": 9, "instanc": [9, 12, 16, 18], "add": [9, 12], "map": [9, 11, 12], "rout": [9, 11], "tabl": 9, "do": 9, "plan": [9, 12], "softwar": [9, 12, 18], "real": 10, "world": [10, 18], "dataset": 10, "control": [10, 16, 18], "oper": [10, 14, 16, 17, 18, 20], "difficult": 10, "acquir": 10, "align": 10, "usag": [10, 14], "itself": 10, "enough": 10, "train": [10, 14, 16, 17, 18], "test": [10, 16], "model": [10, 18], "reinforc": [10, 18, 21], "etc": [10, 18], "tool": [10, 18], "often": 10, "veri": 10, "complex": 10, "especi": 10, "research": [10, 11, 12], "lack": [10, 16], "common": 10, "simpl": [10, 12], "api": [10, 12, 21], "evalu": [10, 16], "benchmark": 10, "algorithm": [10, 12, 14, 18], "framework": [11, 18], "consist": [11, 12, 18], "essenti": [11, 18], "each": [11, 12, 13, 14, 15, 16, 17, 18, 20, 21], "plai": [11, 18], "crucial": [11, 18], "role": [11, 18, 21], "system": [11, 12, 14, 18], "addition": [11, 12, 18], "encompass": 11, "facilit": [11, 12, 18], "seamless": [11, 18], "interact": [11, 18, 20, 21], "within": [11, 12, 14], "graphic": 11, "represent": [11, 18], "below": [11, 12, 20, 21], "illustr": [11, 20], "architectur": [11, 18], "work": 11, "togeth": 11, "cohes": 11, "effici": [11, 14, 16, 18, 21], "servic": [11, 13, 14, 16, 18], "ai": [11, 12], "develop": [11, 12], "compris": 11, "main": 11, "establish": [11, 12], "enabl": [11, 12, 18, 21], "desir": [11, 16], "respons": 11, "allow": [11, 12, 18, 21], "compat": 11, "like": [11, 14, 18], "stabl": [11, 18], "baselines3": [11, 18], "cleanrl": [11, 18], "central": 11, "maintain": 11, "keep": 11, "track": 11, "its": [11, 14, 16, 18], "assign": [11, 14, 16, 17], "dure": [11, 16], "session": [11, 12, 14, 16, 17, 20], "either": [11, 12, 14, 16, 17], "through": [11, 12, 14, 16, 17, 18, 20], "n": [11, 12, 14, 16, 17, 18], "offer": [11, 18, 21], "distinct": [11, 14, 18], "furthermor": 11, "ten": 11, "statist": 11, "enhanc": [11, 12, 18], "capabl": [11, 12, 18], "machin": [12, 14, 16, 17, 18], "requst": 12, "collabor": 12, "mlwin": 12, "ring": 12, "univers": 12, "public": 12, "internet": 12, "intel": 12, "ani": [12, 16], "issu": 12, "feel": 12, "free": 12, "u": 12, "netaigym": 12, "gmail": 12, "com": 12, "launch": 12, "sim": 12, "aa": 12, "devcloud": 12, "avail": [12, 16, 21], "meanwhil": [12, 14], "networkgymsim": 12, "5g": 12, "ran": [12, 18], "distribut": [12, 14, 16, 17, 18], "energi": [12, 18], "save": [12, 18], "predict": 12, "more": [12, 14, 16, 17, 18], "onc": [12, 14, 16, 17, 20], "gain": 12, "begin": 12, "download": [12, 14, 16, 17], "your": [12, 18], "prompt": 12, "collect": [12, 18, 21], "them": [12, 14, 16], "forward": 12, "open": [12, 18], "non": 12, "relev": 12, "transmit": [12, 14, 16, 17], "continu": [12, 14, 16, 17, 20], "conveni": 12, "visual": 12, "websit": 12, "demonstr": [12, 20], "straightforward": 12, "packag": 12, "client_id": [12, 14, 16, 17, 20], "argument": 12, "classic": 12, "loop": [12, 18], "implement": [12, 14, 16], "load_config_fil": [12, 14, 16, 17, 20], "env_nam": [12, 20], "nqos_split": [12, 17, 20], "num_step": [12, 20], "1000": [12, 20], "rang": [12, 14, 20, 21], "sampl": [12, 20, 21], "exit": [12, 14, 16, 17, 20], "break": [12, 20], "epsiod": [12, 20], "up": [12, 20], "still": [12, 20], "anoth": [12, 20], "first": [12, 16, 17, 20], "load": [12, 14, 16, 17], "name": 12, "keyword": 12, "associ": 12, "dedic": [12, 13, 14], "worker": 12, "until": 12, "parallel": 12, "obtain": 12, "perform": [12, 13, 15, 16, 17, 18, 20], "As": 12, "along": 12, "One": 12, "multipl": [12, 14, 16, 17], "fix": 12, "restart": 12, "handl": [12, 14, 16, 17], "time": [12, 14, 16, 17], "limit": [12, 14, 16, 17], "detail": [12, 14, 16, 17, 18, 21], "everi": 12, "attribut": 12, "understand": 12, "expect": 12, "input": 12, "output": 12, "respect": [12, 14], "random": [12, 14, 16, 17], "instead": 12, "prefer": 12, "empti": [12, 18, 21], "shown": [12, 21], "np": [12, 21], "both": [12, 14, 15, 16, 17, 20], "inherit": 12, "major": 12, "possibl": [12, 16], "might": 12, "describ": 12, "dimension": 12, "where": [12, 14, 17], "upper": 12, "lower": 12, "our": [12, 14], "shift": 12, "tupl": 12, "multibinari": 12, "binari": 12, "multidiscret": [12, 16], "wrapper": [12, 14, 16, 17], "maximum": [12, 14, 16], "clipact": 12, "clip": 12, "li": [12, 18], "rescaleact": [12, 14, 16, 17], "rescal": [12, 14, 16, 17], "lie": 12, "interv": 12, "timeawareobserv": 12, "about": [12, 14, 16, 17, 18], "index": [12, 19], "ensur": [12, 14, 16, 17], "transit": 12, "markov": 12, "flattenobserv": 12, "flatten": 12, "normalizeobserv": [12, 14, 16, 17], "normal": [12, 14, 16, 17], "t": [12, 14], "coordin": 12, "center": [12, 18], "unit": 12, "varianc": 12, "wrap": 12, "arg": 12, "normal_obs_env": 12, "cellular": [13, 16, 17], "period": [13, 15, 16, 17], "among": [13, 14], "priorit": [13, 14, 16], "share": [13, 14], "while": [13, 14, 20], "strive": [13, 14], "meet": [13, 14, 16], "level": [13, 14, 16], "agreement": [13, 14], "entir": [14, 16, 17, 20], "partit": 14, "tailor": [14, 18], "serv": [14, 16, 18], "percept": 14, "possess": 14, "abil": 14, "great": 14, "design": [14, 18, 21], "belong": 14, "mac": 14, "schedul": [14, 16], "proport": 14, "fair": 14, "guarante": 14, "equit": 14, "imparti": 14, "inf": [14, 16, 17], "config": [14, 16, 17], "network_sl": 14, "pose": 14, "challeng": [14, 16], "compet": 14, "finit": 14, "primari": [14, 16], "strateg": [14, 17], "access": [14, 18], "vlab": [14, 16, 17], "multidimension": 14, "ndarrai": [14, 16, 17], "five": 14, "total": [14, 16], "mbp": [14, 16, 17], "fraction": 14, "express": 14, "percentag": 14, "sum": 14, "across": [14, 16], "equal": [14, 17], "averag": 14, "violat": [14, 16], "wise": 14, "record": 14, "experienc": 14, "moment": 14, "millisecond": 14, "min": [14, 16, 17], "max": [14, 16, 17], "note": [14, 16, 17, 20], "exce": 14, "surpass": 14, "appropri": 14, "scale": 14, "down": 14, "softmax": 14, "adher": 14, "constraint": [14, 16], "mathemat": 14, "sum_": 14, "a_i": 14, "quad": 14, "frac": 14, "j": 14, "a_j": 14, "num": [14, 17], "account": 14, "factor": 14, "cost": 14, "formul": 14, "left": 14, "text": 14, "_i": 14, "lambda": 14, "cdot": 14, "delay_violation_r": 14, "gamma": 14, "rb_usage_r": 14, "adjust": [14, 16], "emphasi": 14, "given": [14, 16, 17], "By": [14, 18], "consider": 14, "uniform": [14, 16, 17], "2d": [14, 16, 17], "plane": [14, 16, 17], "x": [14, 16, 17], "y": [14, 16, 17], "boundari": [14, 16, 17], "veloc": [14, 16, 17], "episodes_per_sess": [14, 16, 17, 20], "steps_per_episod": [14, 16, 17, 20], "l": [14, 16, 17, 20], "happen": [14, 16, 17], "length": [14, 16, 17, 20], "gener": [14, 16, 17, 20], "subsequ": [14, 16, 17, 20], "cannot": [14, 16, 17, 20], "reconfigur": [14, 16, 17, 20], "point": [14, 16, 17, 20], "program": [14, 16, 17, 18, 20], "mx": [15, 16, 17], "split": 15, "ratio": [15, 17], "consid": [15, 16, 17], "lte": [15, 16, 17], "steer": [15, 18], "link": [15, 16, 17], "part": [16, 17], "dimens": 16, "qos_steer": 16, "scenario": 16, "randomli": [16, 17], "intellig": 16, "direct": 16, "aim": [16, 17], "achiev": [16, 17], "best": 16, "qualiti": 16, "regard": 16, "halt": 16, "admiss": 16, "capac": [16, 17], "10": 16, "accommod": 16, "drop": 16, "remain": 16, "On": 16, "hand": 16, "contend": 16, "without": 16, "restrict": 16, "section": [16, 20], "qos_requir": 16, "test_duration_m": 16, "500": 16, "durat": 16, "delay_bound_m": 16, "100": 16, "flow": 16, "delay_violation_target": 16, "02": 16, "target": 16, "loss_target": 16, "001": 16, "loss": 16, "packet": 16, "otherwis": 16, "met": 16, "set": 16, "zero": 16, "estim": [16, 17], "channel": [16, 17, 18], "third": [16, 17], "combin": [16, 17], "mechan": 16, "maxim": [16, 17], "lead": 16, "essenc": 16, "optim": [16, 18, 21], "indirectli": 16, "overal": 16, "problem": 17, "station": 17, "closest": 17, "handov": 17, "disabl": 17, "latenc": 17, "transmitt": 17, "f": 17, "owd": 17, "innov": 18, "core": 18, "simplifi": 18, "known": 18, "integr": [18, 21], "variou": 18, "four": 18, "purpos": [18, 21], "offlin": [18, 21], "flexibl": [18, 21], "own": [18, 21], "special": [18, 21], "demo": [18, 21], "power": 18, "advanc": 18, "stabil": 18, "reliabl": 18, "tutori": [18, 21], "guidanc": 18, "instruct": [18, 21], "effect": 18, "remot": 18, "act": 18, "intermediari": 18, "leverag": 18, "traffic": 18, "slice": 18, "comprehens": 18, "realist": 18, "cut": 18, "edg": 18, "field": 18, "overview": 18, "page": [18, 19], "There": 18, "compel": 18, "languag": 18, "platform": 18, "separ": 18, "seamlessli": [18, 21], "c": 18, "divers": 18, "technologi": 18, "decoupl": 18, "deploy": 18, "thei": 18, "deploi": 18, "ml": 18, "workload": 18, "focus": 18, "layer": 18, "involv": 18, "abstract": 18, "aspect": 18, "endeavor": 18, "full": 18, "stack": 18, "cloud": 18, "fidel": 18, "proprietari": 18, "applic": 18, "xapp": 18, "rapp": 18, "ric": 18, "digit": 18, "twin": [18, 21], "search": 19, "outlin": 20, "env_config": 20, "start": 20, "showcas": 20, "process": 20, "fulfil": 21, "To": 21, "simpli": 21, "modifi": 21, "trigger": 21, "art": 21, "sota": 21, "These": 21, "popular": 21, "ones": 21, "ppo": 21, "proxim": 21, "ddpg": 21, "deep": 21, "determinist": 21, "gradient": 21, "sac": 21, "soft": 21, "actor": 21, "critic": 21, "td3": 21, "a2c": 21, "advantag": 21, "moreov": 21}, "objects": {"network_gym_client": [[2, 0, 1, "", "Adapter"], [6, 0, 1, "", "Env"], [7, 0, 1, "", "MeasurementReport"], [7, 0, 1, "", "NorthboundInterface"]], "network_gym_client.Adapter": [[2, 1, 1, "", "df_to_dict"], [2, 1, 1, "", "wandb_log"]], "network_gym_client.Env": [[6, 2, 1, "", "action_space"], [6, 2, 1, "", "adapter"], [6, 2, 1, "", "northbound_interface_client"], [6, 2, 1, "", "observation_space"], [6, 1, 1, "", "reset"], [6, 1, 1, "", "step"]], "network_gym_client.NorthboundInterface": [[7, 1, 1, "", "connect"], [7, 1, 1, "", "process_measurement"], [7, 1, 1, "", "recv"], [7, 1, 1, "", "send"]], "network_gym_client.envs.network_slicing": [[3, 0, 1, "", "Adapter"]], "network_gym_client.envs.network_slicing.Adapter": [[3, 1, 1, "", "get_action_space"], [3, 1, 1, "", "get_observation"], [3, 1, 1, "", "get_observation_space"], [3, 1, 1, "", "get_policy"], [3, 1, 1, "", "get_rbg_size"], [3, 1, 1, "", "get_reward"]], "network_gym_client.envs.nqos_split": [[4, 0, 1, "", "Adapter"]], "network_gym_client.envs.nqos_split.Adapter": [[4, 1, 1, "", "get_action_space"], [4, 1, 1, "", "get_observation"], [4, 1, 1, "", "get_observation_space"], [4, 1, 1, "", "get_policy"], [4, 1, 1, "", "get_reward"], [4, 1, 1, "", "netowrk_util"]], "network_gym_client.envs.qos_steer": [[5, 0, 1, "", "Adapter"]], "network_gym_client.envs.qos_steer.Adapter": [[5, 1, 1, "", "calculate_wifi_qos_user_num"], [5, 1, 1, "", "get_action_space"], [5, 1, 1, "", "get_observation"], [5, 1, 1, "", "get_observation_space"], [5, 1, 1, "", "get_policy"], [5, 1, 1, "", "get_reward"]]}, "objtypes": {"0": "py:class", "1": "py:function", "2": "py:attribute"}, "objnames": {"0": ["py", "class", "Python class"], "1": ["py", "function", "Python function"], "2": ["py", "attribute", "Python attribute"]}, "titleterms": {"build": 0, "document": [0, 19], "websit": 0, "instal": 0, "depend": 0, "us": 0, "tip": 0, "auto": 0, "gener": 0, "rst": 0, "file": 0, "add": 0, "class": [0, 2, 8, 11], "function": 0, "md": 0, "network_gym_cli": [1, 6, 7], "adapt": [2, 3, 4, 5], "The": 2, "base": 2, "method": [2, 3, 4, 5, 6, 7], "addit": [2, 3, 4, 5, 7], "network_sl": 3, "reward": [3, 4, 5, 14, 16, 17], "nqos_split": 4, "qos_steer": 5, "env": 6, "attribut": 6, "northboundinterfac": 7, "network_gym_env": 8, "networkgym": [8, 10, 11, 12, 18], "uml": 8, "diagram": [8, 11], "network_gym_serv": 9, "motiv": 10, "where": 10, "i": 10, "data": 10, "network": [10, 13, 14, 18, 19, 21], "ai": [10, 18], "develop": [10, 18], "challeng": 10, "why": [10, 18], "overview": 11, "compon": [11, 18], "interfac": 11, "client": 11, "server": 11, "environ": [11, 12], "quickstart": 12, "access": [12, 16, 17], "testb": 12, "via": 12, "vlab": 12, "basic": 12, "usag": 12, "upon": 12, "start": [12, 14, 16, 17], "networkgymcli": 12, "follow": 12, "seri": 12, "step": 12, "occur": 12, "dure": 12, "simul": 12, "process": 12, "repeat": 12, "when": 12, "conclud": 12, "initi": 12, "interact": 12, "explain": 12, "code": [12, 20], "action": [12, 14, 16, 17], "observ": [12, 14, 16, 17], "space": [12, 14, 16, 17], "modifi": 12, "slice": [13, 14], "cellular": 14, "descript": [14, 16, 17], "prerequisit": [14, 16, 17], "argument": [14, 16, 17], "state": [14, 16, 17], "episod": [14, 16, 17], "end": [14, 16, 17], "traffic": [15, 16, 17], "manag": 15, "multi": [16, 17], "qo": 16, "steer": 16, "split": 17, "transit": 17, "dynam": 17, "democrat": 18, "research": 18, "agent": [18, 21], "option": 18, "gymnasium": 18, "api": 18, "kei": 18, "divid": 18, "scope": 18, "limit": [18, 20], "In": 18, "out": 18, "welcom": 19, "gym": 19, "": 19, "indic": 19, "tabl": 19, "handl": 20, "time": 20, "truncat": 20, "termin": 20, "sequenti": 20, "train": [20, 21], "exampl": 20, "python": 20, "system": 21, "default": 21, "custom": 21, "algorithm": 21, "stabl": 21, "baselines3": 21, "cleanrl": 21}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 8, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.viewcode": 1, "sphinx": 57}, "alltitles": {"Build Documentation Website": [[0, "build-documentation-website"]], "Install Dependencies": [[0, "install-dependencies"]], "Build Website": [[0, "build-website"]], "Useful Tips": [[0, "useful-tips"]], "Auto Generate rst Files": [[0, "auto-generate-rst-files"]], "Add Classes and Functions to md Files": [[0, "add-classes-and-functions-to-md-files"]], "network_gym_client": [[1, "network-gym-client"]], "Adapter": [[2, "adapter"]], "The Base Class": [[2, "the-base-class"]], "Methods": [[2, "methods"], [3, "methods"], [4, "methods"], [5, "methods"], [6, "methods"], [7, "methods"]], "Additional Methods": [[2, "additional-methods"], [3, "additional-methods"], [4, "additional-methods"], [5, "additional-methods"], [7, "additional-methods"]], "network_slicing Adapter": [[3, "network-slicing-adapter"]], "": [[3, "id1"], [4, "id1"], [5, "id1"]], "Reward Methods": [[3, "reward-methods"], [5, "reward-methods"]], "nqos_split Adapter": [[4, "nqos-split-adapter"]], "Reward Method": [[4, "reward-method"]], "qos_steer Adapter": [[5, "qos-steer-adapter"]], "Env": [[6, "env"]], "network_gym_client.Env": [[6, "network-gym-client-env"]], "Attributes": [[6, "attributes"]], "NorthboundInterface": [[7, "northboundinterface"]], "network_gym_client.NorthboundInterface": [[7, "network-gym-client-northboundinterface"]], "network_gym_env": [[8, "network-gym-env"]], "NetworkGym UML Class Diagram": [[8, "networkgym-uml-class-diagram"]], "network_gym_server": [[9, "network-gym-server"]], "Motivation": [[10, "motivation"]], "Where is Data?": [[10, "where-is-data"]], "Network AI Developer Challenges (Why NetworkGym?)": [[10, "network-ai-developer-challenges-why-networkgym"]], "Overview": [[11, "overview"]], "NetworkGym: Components and Interfaces": [[11, "networkgym-components-and-interfaces"]], "Client Component": [[11, "client-component"]], "Server Component": [[11, "server-component"]], "Environment Component": [[11, "environment-component"]], "Class Diagram": [[11, "class-diagram"]], "Quickstart": [[12, "quickstart"]], "Accessing the NetworkGym Testbed via vLab": [[12, "accessing-the-networkgym-testbed-via-vlab"]], "Basic Usage": [[12, "basic-usage"]], "\u25b6\ufe0f Upon starting the NetworkGymClient, the following series of steps occur:": [[12, null]], "\ud83d\udd01 During the simulation, the process repeats as follows:": [[12, null]], "\u23f9\ufe0f When the NetworkGym or the simulation concludes:": [[12, null]], "Initializing Environments": [[12, "initializing-environments"]], "Interacting with the Environment": [[12, "interacting-with-the-environment"]], "Explaining the code": [[12, "explaining-the-code"]], "Action and observation spaces": [[12, "action-and-observation-spaces"]], "Modifying the environment": [[12, "modifying-the-environment"]], "Network Slicing": [[13, "network-slicing"]], "Cellular Network Slicing": [[14, "cellular-network-slicing"]], "Description": [[14, "description"], [16, "description"], [17, "description"]], "Prerequisite": [[14, "prerequisite"], [16, "prerequisite"], [17, "prerequisite"]], "Observation Space": [[14, "observation-space"], [16, "observation-space"], [17, "observation-space"]], "Action Space": [[14, "action-space"], [16, "action-space"], [17, "action-space"]], "Reward": [[14, "reward"], [16, "reward"], [17, "reward"]], "Arguments": [[14, "arguments"], [16, "arguments"], [17, "arguments"]], "Starting State": [[14, "starting-state"], [16, "starting-state"], [17, "starting-state"]], "Episode End": [[14, "episode-end"], [16, "episode-end"], [17, "episode-end"]], "Traffic Management": [[15, "traffic-management"]], "Multi-Access QoS Traffic Steering": [[16, "multi-access-qos-traffic-steering"]], "Multi-Access Traffic Splitting": [[17, "multi-access-traffic-splitting"]], "Transition Dynamics": [[17, "transition-dynamics"]], "NetworkGym: Democratizing Network AI Research and Development": [[18, "networkgym-democratizing-network-ai-research-and-development"]], "Network Agent Options": [[18, "network-agent-options"]], "Gymnasium API": [[18, "gymnasium-api"]], "Key Components of NetworkGym": [[18, "key-components-of-networkgym"]], "NetworkGym: Why Divide?": [[18, "networkgym-why-divide"]], "NetworkGym: Scope and Limitations": [[18, "networkgym-scope-and-limitations"]], "\u2714\ufe0f In-Scope": [[18, "in-scope"]], "\u274c Out-of-Scope": [[18, "out-of-scope"]], "Welcome to Network Gym\u2019s documentation!": [[19, "welcome-to-network-gym-s-documentation"]], "Indices and tables": [[19, "indices-and-tables"]], "Handling Time Limits": [[20, "handling-time-limits"]], "Truncation": [[20, "truncation"]], "Termination": [[20, "termination"]], "Sequential Training Example": [[20, "sequential-training-example"]], "Python code:": [[20, "python-code"]], "Training Agents": [[21, "training-agents"]], "System Default Agent": [[21, "system-default-agent"]], "Custom Algorithm Agent": [[21, "custom-algorithm-agent"]], "Stable-Baselines3 Network Agent": [[21, "stable-baselines3-network-agent"]], "CleanRL Network Agent": [[21, "cleanrl-network-agent"]]}, "indexentries": {"adapter (class in network_gym_client)": [[2, "network_gym_client.Adapter"]], "df_to_dict() (in module network_gym_client.adapter)": [[2, "network_gym_client.Adapter.df_to_dict"]], "wandb_log() (in module network_gym_client.adapter)": [[2, "network_gym_client.Adapter.wandb_log"]], "adapter (class in network_gym_client.envs.network_slicing)": [[3, "network_gym_client.envs.network_slicing.Adapter"]], "get_action_space() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_policy"]], "get_rbg_size() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_rbg_size"]], "get_reward() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_reward"]], "adapter (class in network_gym_client.envs.nqos_split)": [[4, "network_gym_client.envs.nqos_split.Adapter"]], "get_action_space() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_policy"]], "get_reward() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_reward"]], "netowrk_util() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.netowrk_util"]], "adapter (class in network_gym_client.envs.qos_steer)": [[5, "network_gym_client.envs.qos_steer.Adapter"]], "calculate_wifi_qos_user_num() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.calculate_wifi_qos_user_num"]], "get_action_space() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_policy"]], "get_reward() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_reward"]], "env (class in network_gym_client)": [[6, "network_gym_client.Env"]], "action_space (network_gym_client.env attribute)": [[6, "network_gym_client.Env.action_space"]], "adapter (network_gym_client.env attribute)": [[6, "network_gym_client.Env.adapter"]], "northbound_interface_client (network_gym_client.env attribute)": [[6, "network_gym_client.Env.northbound_interface_client"]], "observation_space (network_gym_client.env attribute)": [[6, "network_gym_client.Env.observation_space"]], "reset() (in module network_gym_client.env)": [[6, "network_gym_client.Env.reset"]], "step() (in module network_gym_client.env)": [[6, "network_gym_client.Env.step"]], "measurementreport (class in network_gym_client)": [[7, "network_gym_client.MeasurementReport"]], "northboundinterface (class in network_gym_client)": [[7, "network_gym_client.NorthboundInterface"]], "connect() (in module network_gym_client.northboundinterface)": [[7, "network_gym_client.NorthboundInterface.connect"]], "process_measurement() (in module network_gym_client.northboundinterface)": [[7, "network_gym_client.NorthboundInterface.process_measurement"]], "recv() (in module network_gym_client.northboundinterface)": [[7, "network_gym_client.NorthboundInterface.recv"]], "send() (in module network_gym_client.northboundinterface)": [[7, "network_gym_client.NorthboundInterface.send"]]}}) \ No newline at end of file +Search.setIndex({"docnames": ["README", "api/network_gym_client", "api/network_gym_client/adapter", "api/network_gym_client/adapter/network_slicing", "api/network_gym_client/adapter/nqos_split", "api/network_gym_client/adapter/qos_steer", "api/network_gym_client/env", "api/network_gym_client/northbound_interface", "api/network_gym_env", "api/network_gym_env/custom_env", "api/network_gym_env/dummy_sim", "api/network_gym_server", "content/motivation", "content/overview", "content/quickstart", "environments/mx_network_slicing", "environments/mx_network_slicing/cellular_network_slicing", "environments/mx_traffic_management", "environments/mx_traffic_management/mx_qos_traffic_steering", "environments/mx_traffic_management/mx_traffic_splitting", "index", "old-index", "tutorials/customizing_observation_and_reward", "tutorials/handling_time_limits", "tutorials/implementing_custom_environment", "tutorials/training_agents"], "filenames": ["README.md", "api/network_gym_client.md", "api/network_gym_client/adapter.md", "api/network_gym_client/adapter/network_slicing.md", "api/network_gym_client/adapter/nqos_split.md", "api/network_gym_client/adapter/qos_steer.md", "api/network_gym_client/env.md", "api/network_gym_client/northbound_interface.md", "api/network_gym_env.md", "api/network_gym_env/custom_env.md", "api/network_gym_env/dummy_sim.md", "api/network_gym_server.md", "content/motivation.md", "content/overview.md", "content/quickstart.md", "environments/mx_network_slicing.md", "environments/mx_network_slicing/cellular_network_slicing.md", "environments/mx_traffic_management.md", "environments/mx_traffic_management/mx_qos_traffic_steering.md", "environments/mx_traffic_management/mx_traffic_splitting.md", "index.md", "old-index.rst", "tutorials/customizing_observation_and_reward.md", "tutorials/handling_time_limits.md", "tutorials/implementing_custom_environment.md", "tutorials/training_agents.md"], "titles": ["Build Documentation Website", "network_gym_client", "Adapter", "network_slicing Adapter", "nqos_split Adapter", "qos_steer Adapter", "Env", "NorthBoundClient", "network_gym_env", "Custom Environment", "Dummy Simulator", "network_gym_server", "Motivation", "Overview", "Quickstart", "Network Slicing", "Cellular Network Slicing", "Traffic Management", "Multi-Access QoS Traffic Steering", "Multi-Access Traffic Splitting", "NetworkGym: Democratizing Network AI Research and Development", "Welcome to Network Gym\u2019s documentation!", "Customizing Observation and Reward", "Handling Time Limits", "Implementing Custom Environment", "Training Agents"], "terms": {"option": [0, 1, 6, 14, 16, 18, 19, 23, 25], "creat": [0, 10, 13, 14, 20, 25], "new": [0, 10, 14, 22], "virtual": 0, "python": [0, 14, 20], "environ": [0, 1, 2, 3, 5, 6, 7, 8, 10, 11, 12, 15, 16, 17, 18, 19, 20, 22, 23, 25], "python3": 0, "m": [0, 16], "venv": 0, "sphinx_venv": 0, "sourc": [0, 2, 3, 4, 5, 6, 7, 9, 10, 20, 25], "bin": 0, "activ": [0, 8, 13, 14, 25], "requir": [0, 12, 18, 22], "librari": 0, "pip": 0, "r": [0, 16, 19], "txt": 0, "In": [0, 8, 13, 14, 15, 16, 17, 18, 19, 23], "doc": 0, "folder": 0, "run": [0, 6, 9, 10, 14, 16, 18, 19, 23], "follow": [0, 1, 6, 16, 18, 19, 22], "command": 0, "export": 0, "static": 0, "_build": 0, "html": 0, "make": [0, 14], "clean": [0, 20, 25], "network_gym_cli": [0, 2, 3, 4, 5, 14, 16, 18, 19, 23], "The": [0, 1, 6, 8, 10, 11, 13, 14, 16, 18, 19, 20, 22, 23, 25], "contain": [0, 6, 14, 16, 22], "modul": [0, 8, 21], "rm": 0, "sphinx": 0, "apidoc": 0, "o": [0, 14, 20], "d": 0, "2": [0, 6, 16, 18, 19, 20, 22], "markdown": 0, "e": [0, 1, 6, 10, 12, 16, 18, 19, 20, 23], "g": [0, 1, 6, 10, 12, 18, 19, 20], "eval": 0, "autoclass": 0, "env": [0, 1, 3, 4, 5, 7, 8, 10, 14, 16, 18, 19, 23, 25], "networkgymcli": [1, 13, 16, 18, 19], "includ": [1, 11, 14, 20, 22, 25], "three": [1, 13, 16, 18, 19, 20, 23], "compon": [1, 8, 11], "custom": [1, 6, 13, 14, 16, 18, 19, 20, 23], "gymnasium": [1, 2, 6, 13, 14, 25], "adapt": [1, 6, 13, 16, 18, 19], "northbound": [1, 6, 7, 11, 13, 14], "interfac": [1, 6, 7, 8, 11, 14, 20, 24], "inheret": 1, "class": [1, 3, 4, 5, 6, 7, 9, 10, 14], "commun": [1, 6, 8, 13, 14], "agent": [1, 3, 4, 5, 6, 13, 14, 15, 16, 17, 18, 19, 22, 23], "us": [1, 6, 10, 12, 14, 15, 16, 18, 19, 20, 22, 23, 25], "standard": [1, 20], "exchang": [1, 8, 14], "ob": [1, 2, 6, 14, 23], "reward": [1, 2, 6, 14, 23], "action": [1, 2, 3, 4, 5, 6, 8, 10, 15, 17, 20, 22, 23, 25], "reset": [1, 6, 14, 23], "step": [1, 6, 16, 18, 19, 23], "function": [1, 3, 4, 5, 14, 16, 18, 19, 22], "import": [1, 14, 16, 18, 19, 23], "gym": [1, 6, 7], "def": 1, "__init__": 1, "self": [1, 2, 3, 4, 5, 6, 7, 9, 10], "arg1": 1, "arg2": 1, "super": 1, "return": [1, 2, 3, 4, 5, 6, 7, 10, 14], "observ": [1, 3, 4, 5, 6, 20, 25], "termin": [1, 6, 8, 10, 14, 16, 18, 19], "truncat": [1, 6, 14, 16, 18, 19], "info": [1, 6, 14, 23, 25], "seed": [1, 6], "none": [1, 6], "transform": [1, 2, 13, 22], "data": [1, 2, 5, 8, 13, 14, 16, 18, 20, 22, 25], "format": [1, 2, 13, 14], "from": [1, 2, 3, 4, 5, 6, 7, 8, 14, 16, 22, 23, 25], "network_gym": [1, 2, 6], "other": [1, 16, 18, 19, 23], "wai": [1, 6, 19, 22], "around": [1, 20], "network": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 13, 14, 18, 19, 22, 24], "stat": [1, 2, 3, 4, 5, 7, 10], "chang": [1, 23], "polici": [1, 2, 3, 4, 5, 7, 8, 14, 25], "connect": [1, 6, 7, 8, 9, 10, 11, 13, 14, 17, 18, 19, 20, 22], "client": [1, 6, 7, 8, 10, 11, 14, 16, 18, 19, 20], "server": [1, 6, 7, 8, 10, 11, 14, 16, 18, 19, 20], "configur": [1, 2, 3, 4, 5, 6, 7, 8, 10, 13, 14, 16, 18, 19, 20, 22, 23], "paramet": [1, 2, 3, 4, 5, 6, 7, 9, 10, 16, 18, 19, 20, 23], "between": [1, 2, 6, 8, 13, 19, 20], "envrion": [1, 11], "config_json": [2, 3, 4, 5, 6, 7, 10, 14, 16, 18, 19, 23], "thi": [2, 3, 4, 5, 6, 8, 13, 14, 15, 16, 17, 18, 19, 20, 22, 23], "i": [2, 3, 4, 6, 7, 10, 13, 14, 16, 18, 19, 20, 22, 23, 25], "an": [2, 6, 8, 11, 13, 14, 16, 18, 19, 20, 22, 23, 25], "It": [2, 6, 8, 13, 14, 16, 18, 19, 20, 23], "measur": [2, 3, 4, 5, 6, 7, 8, 9, 10, 13, 14, 16, 22], "json": [2, 3, 4, 5, 6, 7, 10, 14, 16, 18, 19, 22, 23], "space": [2, 3, 4, 5, 6, 20], "also": [2, 6, 12, 14, 16, 18, 19, 20, 23, 24], "can": [2, 6, 8, 13, 14, 16, 18, 19, 20, 22, 23, 25], "appli": [2, 22], "initi": [2, 3, 4, 5, 6, 7, 8, 9, 16], "file": [2, 3, 4, 5, 6, 7, 10, 14, 16, 18, 19, 22, 23], "wandb_log": 2, "send": [2, 6, 7, 8, 9, 10, 14, 20, 22, 25], "log": [2, 6, 19], "inform": [2, 6, 14, 18, 19, 20, 22], "wandb": [2, 14], "df_to_dict": 2, "df": [2, 3, 4, 5, 22], "descript": [2, 22], "datatyp": 2, "panda": [2, 5], "datafram": [2, 3, 4, 5, 7], "dictionari": [2, 6, 14], "object": [2, 6, 16], "string": 2, "descritpt": 2, "dict": [2, 6, 10, 14], "convert": 2, "base": [3, 4, 5, 8, 13, 14, 19, 20], "get_observ": [3, 4, 5, 16, 18, 19], "prepar": [3, 4, 5], "should": [3, 4, 5, 6, 14, 16, 22], "same": [3, 4, 5, 16, 22], "number": [3, 4, 5, 6, 10, 14, 16, 18, 19, 23], "featur": [3, 4, 5, 16, 18, 19, 20], "defin": [3, 4, 5, 6, 14, 16, 18, 19, 20, 25], "get_observation_spac": [3, 4, 5], "pd": [3, 4, 5, 7], "get_reward": [3, 4, 5, 16, 18, 19], "get_polici": [3, 4, 5], "rl": [3, 4, 5, 6, 14, 20, 22, 25], "No": [3, 14, 25], "yet": 3, "get_action_spac": [3, 4, 5], "get": [3, 4, 5, 6, 22], "get_rbg_siz": 3, "bandwidth": [3, 18], "comput": [3, 14, 16, 18, 19, 20, 22], "resourc": [3, 15, 16, 18, 19, 20], "block": [3, 16], "group": [3, 16, 22], "size": [3, 22], "bandwith": 3, "rb": [3, 16], "code": [3, 20, 22, 25], "coppi": 3, "ns3": [3, 12], "pf": 3, "type": [3, 6, 14, 20], "0": [3, 4, 6, 14, 16, 18, 19, 22, 23], "alloc": [3, 15, 16, 20], "rbg": 3, "int": [3, 6, 7, 9, 10], "resouc": 3, "netowrk_util": [4, 19], "throughput": [4, 16, 18, 19, 22], "delai": [4, 6, 16, 18, 19, 22, 25], "alpha": 4, "5": [4, 16, 18, 19, 22], "calcul": [4, 5, 18], "util": [4, 13, 16, 18, 19, 22], "specifi": [4, 6, 14, 16, 18, 19, 23], "valu": [4, 6, 14, 16, 18, 19, 22], "balanc": 4, "default": [4, 5, 6, 14, 16, 20, 22], "float": 4, "repres": [4, 14, 16, 18, 19], "bit": 4, "per": [4, 5, 16, 18, 19, 23], "second": [4, 14, 18, 19, 23], "metric": [4, 9, 14], "calculate_wifi_qos_user_num": [5, 18], "qos_rat": [5, 18, 22], "qo": [5, 14, 17, 20, 22], "user": [5, 6, 10, 14, 16, 17, 18, 19, 20, 22, 25], "over": [5, 17, 18, 19, 22], "wi": [5, 17, 18, 19, 22], "fi": [5, 17, 18, 19, 22], "rate": [5, 16, 18, 19, 22], "doubl": 5, "id": [6, 7, 22], "networkgym": [6, 7, 16, 18, 19, 23, 24, 25], "initil": [6, 10], "intern": [6, 14], "state": [6, 14, 25], "": [6, 8, 14, 16, 18, 19, 22, 23, 25], "prng": 6, "np_random": 6, "If": [6, 14, 23], "doe": [6, 20], "alreadi": [6, 18], "have": [6, 14, 16, 18, 19, 20, 22, 23, 25], "pass": 6, "chosen": 6, "some": [6, 14], "entropi": 6, "timestamp": [6, 10, 22], "dev": 6, "urandom": 6, "howev": 6, "ha": [6, 14, 16, 18, 19, 23], "you": [6, 13, 14, 16, 18, 19, 22, 23, 25], "integ": 6, "even": 6, "exist": [6, 22], "usual": 6, "want": 6, "right": [6, 16], "after": [6, 14, 16, 18, 19, 22, 23], "been": [6, 14, 25], "never": 6, "again": 6, "pleas": [6, 20], "refer": [6, 14, 20, 23, 25], "minim": [6, 19], "exampl": [6, 14, 22], "abov": [6, 20], "see": [6, 13, 14, 16, 18, 19], "paradigm": 6, "addit": [6, 14, 20], "how": [6, 13, 14, 18, 20, 23], "depend": 6, "specif": [6, 12, 13, 16, 20, 22, 23, 25], "obstyp": 6, "auxiliari": 6, "complement": 6, "analog": 6, "one": [6, 8, 10, 14, 16, 19, 22, 23], "timestep": [6, 14], "dynam": [6, 20], "list": [6, 9, 10, 14, 22], "gamsim": 6, "check": 6, "last": [6, 23], "episod": [6, 14, 23], "done": [6, 14], "acttyp": 6, "provid": [6, 13, 14, 18, 19, 20, 22, 23, 24, 25], "updat": [6, 14, 17, 19], "element": [6, 13, 14], "observation_spac": [6, 14], "next": [6, 14], "due": 6, "supportsfloat": 6, "result": [6, 10, 14, 16, 18, 19, 23], "take": [6, 14, 16, 18, 22], "bool": 6, "whether": 6, "reach": [6, 14, 18], "under": 6, "mdp": 6, "task": 6, "which": [6, 13, 14, 16, 18, 20], "posit": [6, 16, 18, 19], "neg": 6, "goal": [6, 18, 19, 22], "move": 6, "lava": 6, "sutton": 6, "barton": 6, "gridworld": 6, "true": [6, 14, 22], "need": [6, 12, 14, 16], "call": [6, 14], "condit": 6, "outsid": 6, "scope": 6, "satisfi": 6, "typic": 6, "timelimit": [6, 14], "could": [6, 14], "indic": [6, 16, 18], "physic": [6, 20], "go": 6, "out": [6, 14, 22], "bound": [6, 14, 22], "end": [6, 8, 14, 20, 22, 23], "prematur": 6, "befor": 6, "diagnost": 6, "help": [6, 14], "debug": 6, "learn": [6, 12, 20, 25], "raw": 6, "flag": 6, "deprec": 6, "A": [6, 10, 14, 16, 18, 19, 20, 22, 23], "boolean": 6, "case": [6, 14, 16, 20, 22], "further": 6, "undefin": 6, "wa": 6, "remov": 6, "openai": 6, "v26": 6, "favor": 6, "signal": [6, 8, 14], "mai": [6, 14, 23], "emit": 6, "differ": [6, 13, 14, 18, 20, 23], "reason": [6, 20], "mayb": 6, "underli": 6, "solv": 6, "successfulli": 6, "certain": [6, 16, 18, 19, 23], "exceed": [6, 14], "simul": [6, 8, 9, 12, 13, 20, 22], "enter": 6, "invalid": 6, "action_spac": [6, 14, 23, 25], "correspond": 6, "valid": [6, 14, 22], "all": [6, 14, 15, 16, 18, 19, 22], "For": [6, 14, 18, 20], "discret": [6, 14], "give": 6, "mean": [6, 20], "ar": [6, 8, 14, 16, 18, 19, 20], "two": [6, 8, 11, 13, 23], "1": [6, 14, 16, 18, 19, 22, 23], "box": [6, 14, 16, 18, 19], "3": [6, 8, 13, 14, 16, 18, 19, 20, 22, 23], "4028234663852886e": 6, "38": [6, 22], "4": [6, 16, 22], "float32": [6, 16, 18, 19], "shape": [6, 14, 16, 18, 19], "denot": [6, 16, 18, 19, 23], "arrai": [6, 14, 16, 25], "we": [6, 11, 14, 16, 18, 19, 20, 22, 23], "well": 6, "high": [6, 19, 20], "8000002e": 6, "00": 6, "4028235e": 6, "1887903e": 6, "01": 6, "dtype": 6, "low": [6, 19], "northbound_interface_cli": 6, "traslat": 6, "dataformat": 6, "sever": [7, 20], "repli": 7, "start": [7, 8, 10, 22, 23], "request": [7, 8], "where": [7, 14, 16, 19, 22], "load": [7, 14, 16, 18, 19, 22], "recv": 7, "receiv": [7, 8, 10, 14, 18, 19], "messag": [7, 8], "process_measur": 7, "reply_json": 7, "networkgymenv": [8, 14, 16, 18, 19, 23], "compris": [8, 13, 22], "core": [8, 20], "These": [8, 25], "establish": [8, 13, 14], "via": [8, 24], "design": [8, 16, 20, 25], "southbound": [8, 10, 11, 13, 24], "serv": [8, 16, 18, 20], "bridg": 8, "facilit": [8, 13, 14, 20], "statist": [8, 13, 22], "keep": [8, 13, 22], "aliv": 8, "period": [8, 15, 17, 18, 19], "dispatch": 8, "hello": 8, "upon": 8, "disengag": 8, "n": [8, 13, 14, 16, 18, 19, 20], "instig": 8, "plan": [8, 11, 14], "incorpor": 8, "emul": [8, 13], "testb": [8, 13, 22], "altern": 8, "earmark": 8, "futur": 8, "releas": [8, 11], "depict": 8, "flow": [8, 18, 22], "complet": 8, "cycl": 8, "within": [8, 13, 14, 16, 22], "session": [8, 13, 14, 16, 18, 19, 23], "process": [8, 23], "launch": [8, 14], "env_config": [8, 10, 23], "trigger": [8, 25], "gener": [8, 10, 16, 18, 19, 23], "incom": 8, "onc": [8, 14, 16, 18, 19, 23], "conclud": 8, "map": [8, 11, 13, 14], "importantli": 8, "onli": [8, 18, 22], "env_sim": [8, 10], "ani": [8, 14, 18], "given": [8, 16, 18, 19], "moment": [8, 16], "cannot": [8, 16, 18, 19, 23], "occur": 8, "simultan": 8, "standpoint": 8, "treat": 8, "both": [8, 14, 16, 17, 18, 19, 22, 23], "equival": 8, "identifi": 8, "them": [8, 14, 16, 18], "entiti": 8, "ident": [9, 10], "networkgymsim": [9, 14], "env_list": 9, "port_num": 9, "8091": 9, "random": [9, 10, 14, 16, 18, 19], "str": [9, 10], "support": [9, 13, 14, 18, 20, 24], "port": [9, 10], "env_ident": 10, "env_port": 10, "client_ident": 10, "sampl": [10, 14, 23, 25], "when": [10, 11, 16, 18, 19, 20, 25], "resum": 10, "socket": 10, "env_idennt": 10, "first": [10, 14, 18, 19, 23], "who": 10, "generate_dummy_measur": 10, "name": [10, 14, 22], "tag": 10, "num_us": 10, "ad": 10, "start_simul": 10, "api": [10, 12, 14, 25], "report": 10, "run_one_interv": 10, "interv": [10, 14], "networkgymserv": [11, 13, 14], "manag": [11, 18, 19, 20], "select": [11, 13, 14, 16, 17, 18, 19], "idl": 11, "instanc": [11, 14, 18, 20], "add": [11, 14], "rout": [11, 13], "tabl": 11, "At": [11, 13, 16, 18, 19, 20, 23], "present": [11, 13, 18, 22], "do": 11, "softwar": [11, 14, 20], "real": 12, "world": [12, 20], "dataset": 12, "control": [12, 18, 20], "oper": [12, 16, 18, 19, 20, 23], "difficult": 12, "acquir": 12, "align": 12, "usag": [12, 16], "itself": 12, "enough": 12, "train": [12, 16, 18, 19, 20], "test": [12, 18], "model": [12, 20], "reinforc": [12, 20, 25], "etc": [12, 20], "tool": [12, 20], "often": 12, "veri": 12, "complex": 12, "especi": 12, "research": [12, 13, 14], "lack": [12, 18], "common": 12, "simpl": [12, 14], "evalu": [12, 18], "benchmark": 12, "algorithm": [12, 14, 16, 20], "framework": [13, 20], "consist": [13, 14, 20], "essenti": [13, 20], "each": [13, 14, 15, 16, 17, 18, 19, 20, 22, 23, 25], "plai": [13, 20], "crucial": [13, 20], "role": [13, 20, 25], "system": [13, 14, 16, 20, 22], "addition": [13, 14, 20], "encompass": [13, 22], "seamless": [13, 20], "interact": [13, 20, 23, 25], "graphic": 13, "represent": [13, 20], "below": [13, 14, 22, 23, 25], "illustr": [13, 22, 23], "architectur": [13, 20], "work": 13, "togeth": 13, "cohes": 13, "effici": [13, 16, 18, 20, 25], "servic": [13, 15, 16, 18, 20], "ai": [13, 14], "develop": [13, 14], "main": 13, "enabl": [13, 14, 20, 25], "desir": [13, 18], "respons": 13, "allow": [13, 14, 20, 25], "compat": 13, "like": [13, 16, 20], "stabl": [13, 20], "baselines3": [13, 20], "cleanrl": [13, 20], "central": 13, "maintain": 13, "track": 13, "its": [13, 16, 18, 20, 22], "assign": [13, 16, 18, 19], "dure": [13, 18], "either": [13, 14, 16, 18, 19], "through": [13, 14, 16, 18, 19, 20, 23], "offer": [13, 20, 22, 25], "distinct": [13, 16, 20], "furthermor": 13, "ten": 13, "enhanc": [13, 14, 20], "capabl": [13, 14, 20], "machin": [14, 16, 18, 19, 20], "requst": 14, "collabor": 14, "mlwin": 14, "ring": 14, "univers": 14, "public": 14, "internet": 14, "intel": 14, "issu": [14, 22], "feel": 14, "free": 14, "u": 14, "netaigym": 14, "gmail": 14, "com": 14, "sim": 14, "aa": 14, "devcloud": 14, "avail": [14, 18, 25], "meanwhil": [14, 16], "5g": 14, "ran": [14, 20], "distribut": [14, 16, 18, 19, 20], "energi": [14, 20], "save": [14, 20], "predict": 14, "more": [14, 16, 18, 19, 20], "gain": 14, "begin": 14, "download": [14, 16, 18, 19], "your": [14, 20, 22], "prompt": 14, "collect": [14, 20, 25], "forward": 14, "open": [14, 20], "non": [14, 22], "store": 14, "relev": 14, "transmit": [14, 16, 18, 19], "continu": [14, 16, 18, 19, 23], "conveni": 14, "visual": 14, "websit": 14, "demonstr": [14, 23], "straightforward": 14, "packag": 14, "client_id": [14, 16, 18, 19, 23], "argument": 14, "classic": 14, "loop": [14, 20], "implement": [14, 16, 18], "load_config_fil": [14, 16, 18, 19, 23], "env_nam": [14, 23], "nqos_split": [14, 19, 23], "num_step": [14, 23], "1000": [14, 23], "rang": [14, 16, 22, 23, 24, 25], "exit": [14, 16, 18, 19, 23], "break": [14, 23], "epsiod": [14, 23], "up": [14, 23], "still": [14, 23], "anoth": [14, 23], "keyword": 14, "associ": 14, "dedic": [14, 15, 16], "worker": 14, "until": 14, "parallel": 14, "obtain": 14, "perform": [14, 15, 17, 18, 19, 20, 23], "As": 14, "along": 14, "One": [14, 22], "multipl": [14, 16, 18, 19], "fix": 14, "restart": 14, "handl": [14, 16, 18, 19], "time": [14, 16, 18, 19, 22], "limit": [14, 16, 18, 19], "detail": [14, 16, 18, 19, 20, 25], "everi": 14, "attribut": 14, "understand": 14, "expect": 14, "input": [14, 22], "output": [14, 22], "respect": [14, 16], "instead": 14, "prefer": 14, "empti": [14, 20, 25], "shown": [14, 25], "np": [14, 25], "inherit": 14, "major": 14, "possibl": [14, 18], "might": [14, 22], "describ": 14, "dimension": 14, "upper": 14, "lower": 14, "our": [14, 16], "shift": 14, "tupl": 14, "multibinari": 14, "binari": 14, "multidiscret": [14, 18], "wrapper": [14, 16, 18, 19], "maximum": [14, 16, 18, 22], "clipact": 14, "clip": 14, "li": [14, 20], "rescaleact": [14, 16, 18, 19], "rescal": [14, 16, 18, 19], "lie": 14, "timeawareobserv": 14, "about": [14, 16, 18, 19, 20], "index": [14, 21], "ensur": [14, 16, 18, 19], "transit": 14, "markov": 14, "flattenobserv": 14, "flatten": 14, "normalizeobserv": [14, 16, 18, 19], "normal": [14, 16, 18, 19], "t": [14, 16, 22], "coordin": 14, "center": [14, 20], "unit": 14, "varianc": 14, "wrap": 14, "arg": 14, "normal_obs_env": 14, "cellular": [15, 18, 19], "among": [15, 16], "priorit": [15, 16, 18], "share": [15, 16], "while": [15, 16, 23], "strive": [15, 16], "meet": [15, 16, 18, 22], "level": [15, 16, 18], "agreement": [15, 16], "entir": [16, 18, 19, 23], "partit": 16, "tailor": [16, 20, 22], "percept": 16, "possess": 16, "abil": 16, "great": 16, "belong": 16, "mac": 16, "schedul": [16, 18], "proport": 16, "fair": 16, "guarante": 16, "equit": 16, "imparti": 16, "inf": [16, 18, 19], "config": [16, 18, 19], "network_sl": 16, "pose": 16, "challeng": [16, 18], "compet": 16, "finit": 16, "primari": [16, 18], "strateg": [16, 19], "access": [16, 20, 22], "vlab": [16, 18, 19], "multidimension": 16, "ndarrai": [16, 18, 19], "five": 16, "total": [16, 18], "mbp": [16, 18, 19], "fraction": 16, "express": 16, "percentag": [16, 22], "sum": 16, "across": [16, 18], "equal": [16, 19, 22], "averag": 16, "violat": [16, 18, 22], "wise": 16, "record": 16, "experienc": 16, "millisecond": 16, "min": [16, 18, 19], "max": [16, 18, 19], "note": [16, 18, 19, 22, 23], "exce": 16, "surpass": 16, "appropri": 16, "scale": 16, "down": 16, "softmax": 16, "adher": 16, "constraint": [16, 18], "mathemat": 16, "sum_": 16, "a_i": 16, "quad": 16, "frac": 16, "j": 16, "a_j": 16, "num": [16, 19], "account": 16, "factor": 16, "cost": 16, "formul": 16, "left": 16, "text": 16, "_i": 16, "lambda": 16, "cdot": 16, "delay_violation_r": 16, "gamma": 16, "rb_usage_r": 16, "adjust": [16, 18], "emphasi": 16, "By": [16, 20], "consider": 16, "uniform": [16, 18, 19], "2d": [16, 18, 19], "plane": [16, 18, 19], "x": [16, 18, 19], "y": [16, 18, 19], "boundari": [16, 18, 19], "veloc": [16, 18, 19], "episodes_per_sess": [16, 18, 19, 23], "steps_per_episod": [16, 18, 19, 23], "l": [16, 18, 19, 23], "happen": [16, 18, 19], "length": [16, 18, 19, 23], "subsequ": [16, 18, 19, 22, 23], "reconfigur": [16, 18, 19, 23], "point": [16, 18, 19, 22, 23], "program": [16, 18, 19, 20, 23], "mx": [17, 18, 19], "split": [17, 22], "ratio": [17, 19, 22], "consid": [17, 18, 19], "lte": [17, 18, 19, 22], "steer": [17, 20], "link": [17, 18, 19, 22], "part": [18, 19], "dimens": 18, "qos_steer": 18, "scenario": 18, "randomli": [18, 19], "intellig": 18, "direct": [18, 22], "aim": [18, 19], "achiev": [18, 19], "best": 18, "qualiti": 18, "regard": 18, "halt": 18, "admiss": 18, "capac": [18, 19, 22], "10": [18, 22], "accommod": 18, "drop": [18, 22], "remain": 18, "On": 18, "hand": 18, "contend": 18, "without": 18, "restrict": 18, "section": [18, 23], "qos_requir": 18, "test_duration_m": 18, "500": 18, "durat": 18, "delay_bound_m": [18, 22], "100": [18, 22], "delay_violation_target": 18, "02": 18, "target": 18, "loss_target": 18, "001": 18, "loss": 18, "packet": [18, 22], "otherwis": 18, "met": 18, "set": 18, "zero": 18, "estim": [18, 19], "channel": [18, 19, 20], "third": [18, 19], "combin": [18, 19, 22], "mechan": 18, "maxim": [18, 19], "lead": 18, "essenc": 18, "optim": [18, 20, 25], "indirectli": 18, "overal": 18, "problem": 19, "station": 19, "closest": 19, "handov": 19, "disabl": 19, "latenc": 19, "transmitt": 19, "f": 19, "owd": [19, 22], "innov": 20, "simplifi": 20, "known": 20, "integr": [20, 25], "variou": 20, "four": 20, "purpos": [20, 25], "offlin": [20, 25], "flexibl": [20, 22, 25], "own": [20, 25], "special": [20, 25], "demo": [20, 25], "power": 20, "advanc": 20, "stabil": 20, "reliabl": 20, "tutori": [20, 25], "guidanc": 20, "instruct": [20, 25], "effect": 20, "remot": 20, "act": 20, "intermediari": 20, "leverag": 20, "traffic": [20, 22], "slice": 20, "comprehens": [20, 22], "realist": 20, "cut": 20, "edg": 20, "field": [20, 22], "overview": 20, "page": [20, 21], "There": 20, "compel": 20, "languag": 20, "platform": 20, "separ": 20, "seamlessli": [20, 25], "c": 20, "divers": 20, "technologi": 20, "decoupl": 20, "deploy": 20, "thei": 20, "deploi": 20, "ml": 20, "workload": 20, "focus": 20, "layer": [20, 22], "involv": [20, 22], "abstract": 20, "aspect": 20, "endeavor": 20, "full": 20, "stack": 20, "cloud": 20, "fidel": 20, "proprietari": 20, "applic": [20, 22], "xapp": 20, "rapp": 20, "ric": 20, "digit": 20, "twin": [20, 25], "search": 21, "suit": 22, "experiment": 22, "To": [22, 25], "construct": 22, "cid": 22, "end_t": 22, "start_t": 22, "dl": 22, "6000": 22, "gma": 22, "ap_id": 22, "5900": 22, "255": 22, "delay_viol": 22, "59": 22, "64912280701754": 22, "max_owd": 22, "90": 22, "341": 22, "153": 22, "379": 22, "11": 22, "6": 22, "phy": 22, "max_rat": 22, "35": 22, "29": 22, "7": 22, "78": 22, "8": 22, "measurement_ok": 22, "9": 22, "missed_act": 22, "37": 22, "296": 22, "128": 22, "353": 22, "292": 22, "127": 22, "359": 22, "12": 22, "13": 22, "qos_mark": 22, "14": 22, "15": 22, "16": 22, "17": 22, "18": 22, "24": 22, "464": 22, "968": 22, "19": 22, "28": 22, "776": 22, "272": 22, "20": 22, "76": 22, "856": 22, "21": 22, "split_ratio": 22, "26": 22, "22": 22, "23": 22, "traffic_ratio": 22, "69": 22, "30": 22, "77": 22, "52": 22, "31": 22, "70": 22, "48": 22, "25": 22, "tx_rate": 22, "09": 22, "04": 22, "434": 22, "004": 22, "column": 22, "explan": 22, "signifi": 22, "downlink": 22, "ul": 22, "uplink": 22, "converg": 22, "now": 22, "elabor": 22, "row": 22, "cell": 22, "isn": 22, "reorder": 22, "order": 22, "count": 22, "miss": 22, "too": 22, "long": 22, "wait": 22, "therefor": 22, "alwai": 22, "deliveri": 22, "32": 22, "cover": 22, "34": 22, "sinc": 22, "sine": 22, "retriev": 22, "df_rate": 22, "reset_index": 22, "isol": 22, "execut": 22, "lastli": 22, "emploi": 22, "explod": 22, "nest": 22, "frame": 22, "origin": 22, "structur": 22, "here": 22, "final": 22, "outlin": 23, "showcas": 23, "sb": 24, "fulfil": 25, "simpli": 25, "modifi": 25, "art": 25, "sota": 25, "popular": 25, "ones": 25, "ppo": 25, "proxim": 25, "ddpg": 25, "deep": 25, "determinist": 25, "gradient": 25, "sac": 25, "soft": 25, "actor": 25, "critic": 25, "td3": 25, "a2c": 25, "advantag": 25, "moreov": 25}, "objects": {"network_gym_client": [[2, 0, 1, "", "Adapter"], [6, 0, 1, "", "Env"], [7, 0, 1, "", "NorthBoundClient"]], "network_gym_client.Adapter": [[2, 1, 1, "", "df_to_dict"], [2, 1, 1, "", "wandb_log"]], "network_gym_client.Env": [[6, 2, 1, "", "action_space"], [6, 2, 1, "", "adapter"], [6, 2, 1, "", "northbound_interface_client"], [6, 2, 1, "", "observation_space"], [6, 1, 1, "", "reset"], [6, 1, 1, "", "step"]], "network_gym_client.NorthBoundClient": [[7, 1, 1, "", "connect"], [7, 1, 1, "", "process_measurement"], [7, 1, 1, "", "recv"], [7, 1, 1, "", "send"]], "network_gym_client.envs.network_slicing": [[3, 0, 1, "", "Adapter"]], "network_gym_client.envs.network_slicing.Adapter": [[3, 1, 1, "", "get_action_space"], [3, 1, 1, "", "get_observation"], [3, 1, 1, "", "get_observation_space"], [3, 1, 1, "", "get_policy"], [3, 1, 1, "", "get_rbg_size"], [3, 1, 1, "", "get_reward"]], "network_gym_client.envs.nqos_split": [[4, 0, 1, "", "Adapter"]], "network_gym_client.envs.nqos_split.Adapter": [[4, 1, 1, "", "get_action_space"], [4, 1, 1, "", "get_observation"], [4, 1, 1, "", "get_observation_space"], [4, 1, 1, "", "get_policy"], [4, 1, 1, "", "get_reward"], [4, 1, 1, "", "netowrk_util"]], "network_gym_client.envs.qos_steer": [[5, 0, 1, "", "Adapter"]], "network_gym_client.envs.qos_steer.Adapter": [[5, 1, 1, "", "calculate_wifi_qos_user_num"], [5, 1, 1, "", "get_action_space"], [5, 1, 1, "", "get_observation"], [5, 1, 1, "", "get_observation_space"], [5, 1, 1, "", "get_policy"], [5, 1, 1, "", "get_reward"]], "network_gym_env": [[9, 0, 1, "", "CustomEnv"], [10, 0, 1, "", "DummySim"]], "network_gym_env.CustomEnv": [[9, 1, 1, "", "run"]], "network_gym_env.DummySim": [[10, 1, 1, "", "generate_dummy_measurement"], [10, 1, 1, "", "run_one_interval"], [10, 1, 1, "", "start_simulation"]]}, "objtypes": {"0": "py:class", "1": "py:function", "2": "py:attribute"}, "objnames": {"0": ["py", "class", "Python class"], "1": ["py", "function", "Python function"], "2": ["py", "attribute", "Python attribute"]}, "titleterms": {"build": 0, "document": [0, 21], "websit": 0, "instal": 0, "depend": 0, "us": 0, "tip": 0, "auto": 0, "gener": 0, "rst": 0, "file": 0, "add": 0, "class": [0, 2, 13], "function": 0, "md": 0, "network_gym_cli": [1, 6, 7], "adapt": [2, 3, 4, 5], "The": 2, "base": 2, "method": [2, 3, 4, 5, 6, 7, 9, 10], "addit": [2, 3, 4, 5, 7], "network_sl": 3, "reward": [3, 4, 5, 16, 18, 19, 22], "nqos_split": 4, "qos_steer": 5, "env": 6, "attribut": 6, "northboundcli": 7, "network_gym_env": [8, 9, 10], "networkgym": [8, 12, 13, 14, 20], "uml": 8, "diagram": [8, 13], "custom": [9, 22, 24, 25], "environ": [9, 13, 14, 24], "customenv": 9, "dummi": 10, "simul": [10, 14], "dummysim": 10, "network_gym_serv": 11, "motiv": 12, "where": 12, "i": 12, "data": 12, "network": [12, 15, 16, 20, 21, 25], "ai": [12, 20], "develop": [12, 20], "challeng": 12, "why": [12, 20], "overview": 13, "compon": [13, 20], "interfac": 13, "client": 13, "server": 13, "quickstart": 14, "access": [14, 18, 19], "testb": 14, "via": 14, "vlab": 14, "basic": 14, "usag": 14, "upon": 14, "start": [14, 16, 18, 19], "networkgymcli": 14, "follow": 14, "seri": 14, "step": 14, "occur": 14, "dure": 14, "process": 14, "repeat": 14, "when": 14, "conclud": 14, "initi": 14, "interact": 14, "explain": 14, "code": [14, 23], "action": [14, 16, 18, 19], "observ": [14, 16, 18, 19, 22], "space": [14, 16, 18, 19], "modifi": 14, "slice": [15, 16], "cellular": 16, "descript": [16, 18, 19], "prerequisit": [16, 18, 19], "argument": [16, 18, 19], "state": [16, 18, 19], "episod": [16, 18, 19], "end": [16, 18, 19], "traffic": [17, 18, 19], "manag": 17, "multi": [18, 19], "qo": 18, "steer": 18, "split": 19, "transit": 19, "dynam": 19, "democrat": 20, "research": 20, "agent": [20, 25], "option": 20, "gymnasium": 20, "api": 20, "kei": 20, "divid": 20, "scope": 20, "limit": [20, 23], "In": 20, "out": 20, "welcom": 21, "gym": 21, "": 21, "indic": 21, "tabl": 21, "work": 22, "datafram": 22, "format": 22, "handl": 23, "time": 23, "truncat": 23, "termin": 23, "sequenti": 23, "train": [23, 25], "exampl": 23, "python": 23, "implement": 24, "system": 25, "default": 25, "algorithm": 25, "stabl": 25, "baselines3": 25, "cleanrl": 25}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 8, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.viewcode": 1, "sphinx": 57}, "alltitles": {"Build Documentation Website": [[0, "build-documentation-website"]], "Install Dependencies": [[0, "install-dependencies"]], "Build Website": [[0, "build-website"]], "Useful Tips": [[0, "useful-tips"]], "Auto Generate rst Files": [[0, "auto-generate-rst-files"]], "Add Classes and Functions to md Files": [[0, "add-classes-and-functions-to-md-files"]], "network_gym_client": [[1, "network-gym-client"]], "Adapter": [[2, "adapter"]], "The Base Class": [[2, "the-base-class"]], "Methods": [[2, "methods"], [3, "methods"], [4, "methods"], [5, "methods"], [6, "methods"], [7, "methods"], [9, "methods"], [10, "methods"]], "Additional Methods": [[2, "additional-methods"], [3, "additional-methods"], [4, "additional-methods"], [5, "additional-methods"], [7, "additional-methods"]], "network_slicing Adapter": [[3, "network-slicing-adapter"]], "": [[3, "id1"], [4, "id1"], [5, "id1"]], "Reward Methods": [[3, "reward-methods"], [5, "reward-methods"]], "nqos_split Adapter": [[4, "nqos-split-adapter"]], "Reward Method": [[4, "reward-method"]], "qos_steer Adapter": [[5, "qos-steer-adapter"]], "Env": [[6, "env"]], "network_gym_client.Env": [[6, "network-gym-client-env"]], "Attributes": [[6, "attributes"]], "NorthBoundClient": [[7, "northboundclient"]], "network_gym_client.NorthBoundClient": [[7, "network-gym-client-northboundclient"]], "network_gym_env": [[8, "network-gym-env"]], "NetworkGym UML Diagram": [[8, "networkgym-uml-diagram"]], "Custom Environment": [[9, "custom-environment"]], "network_gym_env.CustomEnv": [[9, "network-gym-env-customenv"]], "Dummy Simulator": [[10, "dummy-simulator"]], "network_gym_env.DummySim": [[10, "network-gym-env-dummysim"]], "network_gym_server": [[11, "network-gym-server"]], "Motivation": [[12, "motivation"]], "Where is Data?": [[12, "where-is-data"]], "Network AI Developer Challenges (Why NetworkGym?)": [[12, "network-ai-developer-challenges-why-networkgym"]], "Overview": [[13, "overview"]], "NetworkGym: Components and Interfaces": [[13, "networkgym-components-and-interfaces"]], "Client Component": [[13, "client-component"]], "Server Component": [[13, "server-component"]], "Environment Component": [[13, "environment-component"]], "Class Diagram": [[13, "class-diagram"]], "Quickstart": [[14, "quickstart"]], "Accessing the NetworkGym Testbed via vLab": [[14, "accessing-the-networkgym-testbed-via-vlab"]], "Basic Usage": [[14, "basic-usage"]], "\u25b6\ufe0f Upon starting the NetworkGymClient, the following series of steps occur:": [[14, null]], "\ud83d\udd01 During the simulation, the process repeats as follows:": [[14, null]], "\u23f9\ufe0f When the NetworkGym or the simulation concludes:": [[14, null]], "Initializing Environments": [[14, "initializing-environments"]], "Interacting with the Environment": [[14, "interacting-with-the-environment"]], "Explaining the code": [[14, "explaining-the-code"]], "Action and observation spaces": [[14, "action-and-observation-spaces"]], "Modifying the environment": [[14, "modifying-the-environment"]], "Network Slicing": [[15, "network-slicing"]], "Cellular Network Slicing": [[16, "cellular-network-slicing"]], "Description": [[16, "description"], [18, "description"], [19, "description"]], "Prerequisite": [[16, "prerequisite"], [18, "prerequisite"], [19, "prerequisite"]], "Observation Space": [[16, "observation-space"], [18, "observation-space"], [19, "observation-space"]], "Action Space": [[16, "action-space"], [18, "action-space"], [19, "action-space"]], "Reward": [[16, "reward"], [18, "reward"], [19, "reward"]], "Arguments": [[16, "arguments"], [18, "arguments"], [19, "arguments"]], "Starting State": [[16, "starting-state"], [18, "starting-state"], [19, "starting-state"]], "Episode End": [[16, "episode-end"], [18, "episode-end"], [19, "episode-end"]], "Traffic Management": [[17, "traffic-management"]], "Multi-Access QoS Traffic Steering": [[18, "multi-access-qos-traffic-steering"]], "Multi-Access Traffic Splitting": [[19, "multi-access-traffic-splitting"]], "Transition Dynamics": [[19, "transition-dynamics"]], "NetworkGym: Democratizing Network AI Research and Development": [[20, "networkgym-democratizing-network-ai-research-and-development"]], "Network Agent Options": [[20, "network-agent-options"]], "Gymnasium API": [[20, "gymnasium-api"]], "Key Components of NetworkGym": [[20, "key-components-of-networkgym"]], "NetworkGym: Why Divide?": [[20, "networkgym-why-divide"]], "NetworkGym: Scope and Limitations": [[20, "networkgym-scope-and-limitations"]], "\u2714\ufe0f In-Scope": [[20, "in-scope"]], "\u274c Out-of-Scope": [[20, "out-of-scope"]], "Welcome to Network Gym\u2019s documentation!": [[21, "welcome-to-network-gym-s-documentation"]], "Indices and tables": [[21, "indices-and-tables"]], "Customizing Observation and Reward": [[22, "customizing-observation-and-reward"]], "Working with the DataFrame Format": [[22, "working-with-the-dataframe-format"]], "Customizing Observation": [[22, "customizing-observation"]], "Customizing Reward": [[22, "customizing-reward"]], "Handling Time Limits": [[23, "handling-time-limits"]], "Truncation": [[23, "truncation"]], "Termination": [[23, "termination"]], "Sequential Training Example": [[23, "sequential-training-example"]], "Python code:": [[23, "python-code"]], "Implementing Custom Environment": [[24, "implementing-custom-environment"]], "Training Agents": [[25, "training-agents"]], "System Default Agent": [[25, "system-default-agent"]], "Custom Algorithm Agent": [[25, "custom-algorithm-agent"]], "Stable-Baselines3 Network Agent": [[25, "stable-baselines3-network-agent"]], "CleanRL Network Agent": [[25, "cleanrl-network-agent"]]}, "indexentries": {"adapter (class in network_gym_client)": [[2, "network_gym_client.Adapter"]], "df_to_dict() (in module network_gym_client.adapter)": [[2, "network_gym_client.Adapter.df_to_dict"]], "wandb_log() (in module network_gym_client.adapter)": [[2, "network_gym_client.Adapter.wandb_log"]], "adapter (class in network_gym_client.envs.network_slicing)": [[3, "network_gym_client.envs.network_slicing.Adapter"]], "get_action_space() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_policy"]], "get_rbg_size() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_rbg_size"]], "get_reward() (in module network_gym_client.envs.network_slicing.adapter)": [[3, "network_gym_client.envs.network_slicing.Adapter.get_reward"]], "adapter (class in network_gym_client.envs.nqos_split)": [[4, "network_gym_client.envs.nqos_split.Adapter"]], "get_action_space() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_policy"]], "get_reward() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.get_reward"]], "netowrk_util() (in module network_gym_client.envs.nqos_split.adapter)": [[4, "network_gym_client.envs.nqos_split.Adapter.netowrk_util"]], "adapter (class in network_gym_client.envs.qos_steer)": [[5, "network_gym_client.envs.qos_steer.Adapter"]], "calculate_wifi_qos_user_num() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.calculate_wifi_qos_user_num"]], "get_action_space() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_action_space"]], "get_observation() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_observation"]], "get_observation_space() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_observation_space"]], "get_policy() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_policy"]], "get_reward() (in module network_gym_client.envs.qos_steer.adapter)": [[5, "network_gym_client.envs.qos_steer.Adapter.get_reward"]], "env (class in network_gym_client)": [[6, "network_gym_client.Env"]], "action_space (network_gym_client.env attribute)": [[6, "network_gym_client.Env.action_space"]], "adapter (network_gym_client.env attribute)": [[6, "network_gym_client.Env.adapter"]], "northbound_interface_client (network_gym_client.env attribute)": [[6, "network_gym_client.Env.northbound_interface_client"]], "observation_space (network_gym_client.env attribute)": [[6, "network_gym_client.Env.observation_space"]], "reset() (in module network_gym_client.env)": [[6, "network_gym_client.Env.reset"]], "step() (in module network_gym_client.env)": [[6, "network_gym_client.Env.step"]], "northboundclient (class in network_gym_client)": [[7, "network_gym_client.NorthBoundClient"]], "connect() (in module network_gym_client.northboundclient)": [[7, "network_gym_client.NorthBoundClient.connect"]], "process_measurement() (in module network_gym_client.northboundclient)": [[7, "network_gym_client.NorthBoundClient.process_measurement"]], "recv() (in module network_gym_client.northboundclient)": [[7, "network_gym_client.NorthBoundClient.recv"]], "send() (in module network_gym_client.northboundclient)": [[7, "network_gym_client.NorthBoundClient.send"]], "customenv (class in network_gym_env)": [[9, "network_gym_env.CustomEnv"]], "run() (in module network_gym_env.customenv)": [[9, "network_gym_env.CustomEnv.run"]], "dummysim (class in network_gym_env)": [[10, "network_gym_env.DummySim"]], "generate_dummy_measurement() (in module network_gym_env.dummysim)": [[10, "network_gym_env.DummySim.generate_dummy_measurement"]], "run_one_interval() (in module network_gym_env.dummysim)": [[10, "network_gym_env.DummySim.run_one_interval"]], "start_simulation() (in module network_gym_env.dummysim)": [[10, "network_gym_env.DummySim.start_simulation"]]}}) \ No newline at end of file diff --git a/tutorials/customizing_observation_and_reward.html b/tutorials/customizing_observation_and_reward.html new file mode 100644 index 0000000..8cfa6c4 --- /dev/null +++ b/tutorials/customizing_observation_and_reward.html @@ -0,0 +1,608 @@ + + + + + + + + + + + + Customizing Observation and Reward — NetworkGym Documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
              +
              +
              +
              + + + Ctrl+K +
              +
              + + + +
              +
              + +
              + + + +
              + + + + +
              + +
              +
              + + +
              +
              +
              + + +
              +
              + + + + +
              + +
              + + + +
              + +
              +
              + +
              +
              + +
              + +
              + +
              + + +
              + +
              + +
              + + + + + +
              + + +
              + + + + + + + + + + + +
              + +
              + +
              +
              + + + +
              +

              Customizing Observation and Reward

              + +
              + +
              +
              + + + + +
              + +
              +

              Customizing Observation and Reward#

              +

              You have the flexibility to tailor the observation and reward functions to suit your experimentation goals within the environments we offer. To illustrate, we present a comprehensive example of network statistics measurement. The observation and reward functions can be constructed utilizing this information.

              +
                    cid direction end_ts group             name start_ts          user                                              value
+0     LTE        DL   6000   GMA            ap_id     5900  [0, 1, 2, 3]                       [255.0, 255.0, 255.0, 255.0]
+1   Wi-Fi        DL   6000   GMA            ap_id     5900  [0, 1, 2, 3]                               [0.0, 0.0, 1.0, 1.0]
+2     All        DL   6000   GMA  delay_violation     5900  [0, 1, 2, 3]           [59.64912280701754, 100.0, 100.0, 100.0]
+3     All        DL   6000   GMA          max_owd     5900  [0, 1, 2, 3]                        [90.0, 341.0, 153.0, 379.0]
+4     LTE        DL   6000   GMA          max_owd     5900  [0, 1, 2, 3]                        [90.0, 341.0, 153.0, 379.0]
+5   Wi-Fi        DL   6000   GMA          max_owd     5900  [0, 1, 2, 3]                             [11.0, 11.0, 1.0, 2.0]
+6     LTE        DL   6000   PHY         max_rate     5900  [0, 1, 2, 3]                           [35.0, 35.0, 29.0, 35.0]
+7   Wi-Fi        DL   6000   PHY         max_rate     5900  [0, 1, 2, 3]                           [78.0, 78.0, 78.0, 78.0]
+8     All        DL   6000   GMA   measurement_ok     5900  [0, 1, 2, 3]                               [1.0, 1.0, 1.0, 1.0]
+9     All        DL   6000   GMA    missed_action     5900  [0, 1, 2, 3]                               [0.0, 0.0, 0.0, 0.0]
+10    All        DL   6000   GMA              owd     5900  [0, 1, 2, 3]                        [37.0, 296.0, 128.0, 353.0]
+11    LTE        DL   6000   GMA              owd     5900  [0, 1, 2, 3]                        [38.0, 292.0, 127.0, 359.0]
+12  Wi-Fi        DL   6000   GMA              owd     5900  [0, 1, 2, 3]                               [5.0, 2.0, 0.0, 0.0]
+13    LTE        DL   6000   GMA      qos_marking     5900  [0, 1, 2, 3]                               [1.0, 1.0, 1.0, 1.0]
+14  Wi-Fi        DL   6000   GMA      qos_marking     5900  [0, 1, 2, 3]                               [1.0, 1.0, 0.0, 0.0]
+15    All        DL   6000   GMA         qos_rate     5900  [0, 1, 2, 3]                               [0.0, 0.0, 0.0, 0.0]
+16    LTE        DL   6000   GMA         qos_rate     5900  [0, 1, 2, 3]                               [0.0, 0.0, 0.0, 0.0]
+17  Wi-Fi        DL   6000   GMA         qos_rate     5900  [0, 1, 2, 3]                               [0.0, 0.0, 0.0, 0.0]
+18    All        DL   6000   GMA             rate     5900  [0, 1, 2, 3]                      [13.24, 18.464, 6.968, 19.28]
+19    LTE        DL   6000   GMA             rate     5900  [0, 1, 2, 3]                        [10.8, 7.776, 6.272, 7.776]
+20  Wi-Fi        DL   6000   GMA             rate     5900  [0, 1, 2, 3]                         [4.76, 18.464, 1.856, 7.2]
+21    LTE        DL   6000   GMA      split_ratio     5900  [0, 1, 2, 3]                           [26.0, 15.0, 24.0, 20.0]
+22  Wi-Fi        DL   6000   GMA      split_ratio     5900  [0, 1, 2, 3]                             [6.0, 17.0, 8.0, 12.0]
+23    LTE        DL   6000   GMA    traffic_ratio     5900  [0, 1, 2, 3]                           [69.0, 30.0, 77.0, 52.0]
+24  Wi-Fi        DL   6000   GMA    traffic_ratio     5900  [0, 1, 2, 3]                           [31.0, 70.0, 23.0, 48.0]
+25    All        DL   6000   GMA          tx_rate     5900  [0, 1, 2, 3]                      [25.09, 29.04, 7.434, 18.004]
+
              +
              +

              The network statistics measurement contains 8 columns, with each field’s explanation provided below:

              +
        + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

        Column

        Description

        cid

        Connection ID: Wi-Fi, LTE, or All, where All signifies the combination of all links.

        direction

        DL: downlink; UL: uplink.

        end_ts

        The ending timestamp of this measurement.

        group

        GMA: GMA convergence layer measurement; PHY: measurement from the PHY layer.

        name

        Name of the measurement.

        start_ts

        The starting timestamp of this measurement.

        user

        A list of user IDs for this measurement.

        value

        A list of values for this measurement (user and value should have the same size).

        +

        Now, we elaborate on each row:

        + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

        Row

        Name

        Description

        0,1

        ap_id

        LTE cell ID and Wi-Fi access point ID as measured by each user, encompassing LTE and Wi-Fi links. (NOTE: LTE isn’t measured, 255 is the default value.)

        2

        delay_violation

        One-way delay violation percentage (%) as measured by each user; delay bound (delay_bound_ms) can be configured in the JSON file. (NOTE: Only for QoS use case)

        3,4,5

        max_owd

        Maximum one-way delay measured by each user, involving LTE link, Wi-Fi link, and ALL. (ALL: after reordering out-of-order packets from both links).

        6,7

        max_rate

        LTE/Wi-Fi link capacity as measured by each user.

        8

        measurement_ok

        1: Measurement is valid; 0: An issue exists with the measurement.

        9

        missed_action

        Count of missed actions by the environment. (NOTE: Applicable in testbed environment where RL agent might take too long to compute an action. Simulator waits for action by default in simulations, therefore should always be 0.)

        10,11,12

        owd

        One-way delay measured by each user, involving LTE link, Wi-Fi link, and ALL. (ALL: after reordering out-of-order packets from both links).

        13,14

        qos_marking

        1: QoS user; 0: Non-QoS user.

        15,16,17

        qos_rate

        QoS delivery rate (output traffic throughput meeting QoS requirement) as measured by each user, comprising traffic over LTE link, Wi-Fi link, and ALL (combined).

        18,19,20

        rate

        Delivery rate (output traffic throughput) measured by each user, including traffic over LTE link, Wi-Fi link, and ALL (combined).

        21,22

        split_ratio

        Traffic split ratio (in range of [0, 32]) as measured by each user, covering LTE link, Wi-Fi link. LTE split ratio + Wi-Fi split ratio equals 32.

        23,34

        traffic_ratio

        Traffic ratio (%) sending to each link measured by each user, involving LTE link, Wi-Fi link. traffic_ratio might not equal split_ratio, since the system takes sine time to apply the new action.

        25

        tx_rate

        Load (input traffic throughput) measured by each user.

        +
        +

        Working with the DataFrame Format#

        +

        To retrieve a specific network statistic, such as `rate``, you can utilize the following code:

        +
        df_rate = df[df['name'] == 'rate'].reset_index(drop=True) # get the rate
+
        +
        +

        To isolate the link for the rate measurement, such as All, you can execute the subsequent code:

        +
        df_rate = df_rate[df_rate['cid'] == 'All'].reset_index(drop=True) # get the rate for the all link combined
+
        +
        +

        Lastly, employ the explode function to transform the nested data frame into its original structure:

        +
        df_rate = df_rate.explode(column=['user', 'value']) #keep the flow rate.
+
        +
        +

        Here is the final output:

        +
           cid direction end_ts group  name start_ts user   value
+0  All        DL   6000   GMA  rate     5900    0   13.24
+0  All        DL   6000   GMA  rate     5900    1  18.464
+0  All        DL   6000   GMA  rate     5900    2   6.968
+0  All        DL   6000   GMA  rate     5900    3   19.28
+
        +
        +
        +
        +

        Customizing Observation#

        +
        +
        +

        Customizing Reward#

        +
        + + + + + + + + + + + + + + +
        + +
        +
        + Contents +
        +
        + +
        + + + +
        + +
        + +
        + +

        +By Menglei Zhang +

        + +
        + +
        + +

        + + © Copyright 2023, Intel Labs Multi-Access Intelligence. +
        + +

        + +
        + +
        + +
        + +
        + +
        + +
        +
        + + + + + + + + + + +
        +
        + + \ No newline at end of file diff --git a/tutorials/handling_time_limits.html b/tutorials/handling_time_limits.html index 043f0d0..85745f9 100644 --- a/tutorials/handling_time_limits.html +++ b/tutorials/handling_time_limits.html @@ -147,20 +147,24 @@
      • network_slicing Adapter
      • -
    • NorthboundInterface
      • +
    • NorthBoundClient
  • network_gym_server
    • -
  • network_gym_env
    • +
  • network_gym_env +

    • Environments

    Environments