LTE: How PGW will select the SGW for Incoming Packet?

Question

How P GW will select the S GW for Incoming Packet?
Consider two conditions
1. UE is in idle mode.
2. UE is in ECM CONNECTED and ECM IDLE.

Comments

I hope I got your question correctly, PGW does the TFT mapping which helps in SGW selection and tunnel mapping.

Answer 1

For incoming packet on PGW for a UE means, a valid PDP Context must exist on it. When UE attaches to APN, it's PDP context is stored on MME and P-GW. This context info contains several info e.g. UE IP Address, APN , IMSI, MSISDN, Charging ID and peer & local end TEID. This TEID is kept for both local and peer end till the UE is detached from network. So once tunnel is setup between SGW & PGW it's not deleted even when UE goes IDLE.

P-GW is selected by MME during attach process and SGW setups Tunnel to destination PGW. So for PGW, SGW is always known for a UE (as long it remains attached with network)

Answer 2

On behalf of Bart -
I read your answer and I don't get it.

How the TFTs are a condition based on which the sgw/pgw pair is selected?

The SGW is associated with TAs and based on this condition this and not the other is used when UE is in specific geographical area. At least this is how I understood it, please share if my view is wrong.