gtp and s1c Bind Addresses in srsEnb

[janolestein]

Hello everybody,

i have a question regarding the Configuration of the eNodeB, specifically the gtp and s1c bind addresses.

Let me quickly preface this with what i am doing: (Its for a Uni Project if your wondering)
I want to run two eNodeB on two different PCs with one running open5GS as the EPC.
Both PCs have a B210 connected to them and i have them configured to be able to do S1-Handovers.

The thing is: Its working, i found the right configuration for the eNodeBs to talk to the EPC over the Network. But my Problem is that i don't fully understand why it works and why the values for the bind addresses are what they are to be working.

My Problem is specifically with the "gtp_bind_addr" and the "s1c_bind_addr".
All the Documentation i can find is to vague for me to fully understand how they are operating behind the scenes and what they actually do.

And that is my actual question, does anybody have links to Documentation that makes it clearer or maybe just an explanation?
Any help would be very much appreciated.

To maybe make things a little clearer here is the Configuration i am working with:

PC1 (Local Network Address: 10.10.10.33)
This one is running the EPC and one of the eNodeBs.

Only things changed for the open5GS EPC are:
mme.yaml

      s1ap:
       server:
         - address: 10.10.10.33 //  address of the PC in the Network
     gtpc:
       server:
         - address: 10.10.10.33 //  address of the PC in the Network
and the PLMN IDs of course 

sgwu.yaml

     gtpu:
       server:
        - address: 10.10.10.33 //  address of the PC in the Network

The Confiugariton for the eNodeB is as follows on PC1:
enb1.conf

mme_addr = 10.10.10.33 // makes sense, its the same as the one in the mme.yaml
gtp_bind_addr = 127.0.1.100 // from what i can gather, this has to be some local address in the Loopback address space otherwise it cant bind or it says the port is already in use if it also set to 10.10.10.33
s1c_bind_address = 10.10.10.33 // has to be same as the mme Address on this PC where the EPC is running otherwise the S1-Connection cant be established. Why? 

PC2 (Local Network Address 10.10.10.32)
This one only runs the second eNodeB and is connected to the same network as PC1

The Configuration i found that works to connect this one to the EPC running on PC1 and also enabling S1-Handover and full Internet on my COTS UE.

enb2.conf

mme_addr = 10.10.10.33 // address of PC1 running the EPC 

gtp_bind_addr = 10.10.10.32 // Address of this PC in the local network, if this is changed i wont have an internet connection on the UE. I absolutely don't understand why this has to be the local address of the PC in the Network 

s1c_bind_address = 10.10.10.32 // This also has to be this Address of the PC in the local Network otherwise the S1 Connection to the EPC cant be established. Again no clue why it would be this Address and the one of the PC running the EPC.