Re: LTE Maximum DL throughput Issue
I agree that transport is very important , but in such case, you will not observe constantly 100% PRB usage ( or close to 100%).
have you observed in L3 RB Reconfig. a message used for enodeB to configure Scells? Have you observed that Scells are activated? If yes, what throughput distribution you observed? Maybe there is something in setings of CA. If you can share xml configuration of that enodeB maybe I could check.
Re: LTE Maximum DL throughput Issue
Hi,
Could try a simple test for TX limitation but you need 2 users & 2 Ue to test but second Ue (and user) can be regular one.
1. Start download with your test Ue on one sector, let it get to max throughput
2. Get another user who is camping in a different sector of the same eNB to start a FTP download
If the first Ue throughput drops, TX / FTP server / onward internet connection is bad. If it stays steady and user 2 also gets good throughput then TX good.
I would really recommend having special FTP / HTTP server (files stored in RAM not DISK) close to core without leaving the operator for these tests as too many variables to guarantee throughput once you get above 100mbit/s on the wider Internet. Had hassle with same kind of testing many times :(.
Regards
Re: LTE Maximum DL throughput Issue
Quote:
Originally Posted by
khurrambilal01
Yes, We can see 3CC active and working but Tpt never goes beyond 150Mbps.
Hi, Sir.
I`d suggest a couple things:
1) See how is the BLER across the tests. It shouldnt go above 10%.
2) Check the TBS/MCS allocation to see if they make sense. They should be at very high values for SINR=22dB.
3) Check the reported CQI. It should be in the upper range for the vast majority of the time.
4) Check the reported RI (average). It should be near 4 for the majority of time.
5) Can you test two UEs at the same time? This helps to understand if the limitation is in the air interface or from the eNodeB to the back.
6) Can you test different servers for download? All give the same results?
Can you share a small piece from the logs? Better if covers also attach, and at least a IDLE to CONNECTED transition.