You are on page 1of 3

1) 2) 3) 4)

A8 and A9 interface lies between BS and PCF on cdma patch path. A8 is about data A9 is about signaling A8 and A9 interface is shown below

5) basic call flow between BS(AN in evdo) and PCF is shown below. It wont be much different in cdma-2000 AT BS ---session est-------> PCF PDSN

-------- A9 setup-A8-----> ------A11 Reg request ------------> <--- A11 Reg Resp ---------------<-------A9 connect-A8-------------------------------est ppp connection ---------------------------------------------================Data Transmission===========================

6) stack details are given below for a9 and a11

7) stack details for A8 and A10 is given below.

8) We end up having different state machine for A9 based on call flows on this interface. 9) GRE keys between BS and PCF may be different from GRE keys between PCF and PDSN, so when we establish session in CAE with A9 messages, we would be seeing GRE keys in A9 for GRE tunnel between BS and PCF. So data processing would remain same. But there may be some changes in control path processing inside kdp based on A9 state machine and how cae is going to deal with A9 messages 10) A8 connection is released when session moves from active to dormant unlike A10 where a10 connection is maintained even when session during transition from active to Dormant. This is like preservation cases in UMTS, bit more complicated than A10/A11. 11) meid will be available in A9 messages 12) cell identifier is available in A9 messages 13) imsi is available in A9 messages 14) some of the vendors can use very proprietary interface between BS and PCF (we need to figure out what kddi uses this. Like last time, if they provide specifications, it would be great) 15) Data path processing at simulators would be very minimal, but we need A9Sim in simulators 16) some effort is required from QA in qualifying A9Sim , but much less than 4.0 ci qualification and 2.9.6 Sim qualification. 17) Based on conversation with kumar last time when he visited Bangalore, band width monitoring is required for every one second unlike existing implementation at netmon. So serious effort is required at netmon if we understood requirement correctly 18) From QA side, 2 months time is required to deliver release (on telco) with decent quality with 3 folks, these 2 months does not include time taken to stabilize simulators

19) QA-1 takes care of netmon testing, QA-2 takes care of signaling,control plane testing and QA-3 takes care of rest of testing. 20) QA needs to come up with A9 message based call flow scripts similar to A11. But just effort , no risk 21) 3GPP2 A.S0016-D v3.0 deals with A9 interface.

You might also like