- This topic is empty.
21st July 2009 at 06:44 #58089GemGuest
My EDGE network is terribly slow despite a throughput figure of 130 Kbps.Could there be any specific reasons for this.Have checked the RX Quality and plotted it against Data Throughput and have found no correlation at all.Have heard from field that the coding schemes changes very freuently but have not been able to confirm and am not very sure.Our initial MCS coding is 6 spanning to 9.
Now shoul I furnish any more data.Pls reply21st July 2009 at 08:33 #58090PixGuest
The main reasons for low throughput are:
1- poor RxQual (= poor C/I)
2- low RxLev
3- congestion in the cell (congestion CS + PS)
4- congestion in the Abis
5- bad parameters (max pdch high load, max pdch, max number of tbf multiplexed in one pdch, etc)
6 – Congestion in the Ater PS or in the Gb or in the Gn or in the Gi interfaces.
Since you are talking about a network wide problem, you should focus on the case 3, 4 and 5 and 6. I doubt you have interference and poor coverage in your whole network.
If you’re talking about just one cell in particular, investigate all possible cases.
how many trx in the cell?
how many erlang at busy hour?
how many pdch are allocated in average in the cell ?
how much abis is used and free?
what is the congestion on the ater ps?
what is the congestion on the other interfaces ?
For one cell, perform a drive test a low load hour, and then same place at busy hour. Do you see any difference ?
(which vendor are you using ?)
pix21st July 2009 at 09:58 #58091GemGuest
Yep you are right.1,2,3 and 4 does not apply for me since i have tested out the coorelation already.
Gb link is sufficiently provided and hence there should not be any utilization issue and so we strike off 6th point
Hence only 5th point is left out which i am not able to make out.Could you explain those parameters in detail and their threshold values.I am using Nokia,BTS,BSC and MSC.I am just looking at stats of a group of BTS in a medium level city and not the whole network.21st July 2009 at 10:03 #58092GemGuest
and Pix,time is no dependency here.It is slow and is slower in DBH21st July 2009 at 15:37 #58093PixGuest
Well, I don’t work with Nokia eqiupment, so I can’t help much than this. Try to find the right person to help you out 🙂
Usually, I always look at interface utilization first. Most of EGPRS issues come from one interface which is congested.
Pix22nd July 2009 at 06:05 #58094GemGuest
OK Pix,I am at the dead end.Let me rephrase my question.The problem is with USB sticks alone.Normal GPRS operations through phones are OK.
K2540 is the model I use.Can u throw some light22nd July 2009 at 08:33 #58095JackyGuest
hi all, concerning your question GEM i remember that we faced the same problem in our network and as a recommended solution we did some configuration in TCP/IP window size on PC. otherwise u could check the point 5 and affect more PDCH ressources to the concerned area to observe thr improvment.22nd July 2009 at 10:06 #58096PixGuest
The thing is that with a phone, you never need a lot of throughput. The throughput can only be tested with USB sticks. So I think you would have the same problem with a phone, if you tested a download with it.
There is no dead end in GPRS. There are so many possible problems that the only limit is your lifespan 🙂
I agree with Jacky, the window size in your PC could slightly limit your throughput. Try doing some TCP/IP optimization (you have some freeware to do that, google “TCP IP optimization)
Gem, other questions:
during your data drivetest…
you observe a peak throughput at 130kb/s : what is the current MCS ?
what is the current number of TS allocated ?
what is the retransmission rate ?
in MCS-6, the max throughput with 4 TS is 30 * 4 = 120kb/s. So probably around 100 or 110 kb/s in real life.
If you are in MCS9 but limited at 120kb/s, then it could be a TCP/IP setting problem.4th August 2009 at 13:56 #58097RNP2Guest
Can u please explain what do u mean by “configuration in TCP/IP window size on PC”.
thanks10th August 2009 at 10:16 #58098JPGuest
As per your information regarding window size and “TCP IP optimization”…..Can anybody explain what kind of TCP IP optimization exactly you mean by i mean is it at our system end or Computer end or SGSN or etc….10th August 2009 at 12:45 #58099JackyGuest
Computer End Parameter Tuning 🙂