[LENA] multiple base stations

73 views
Skip to first unread message

Xing Xu

unread,
Mar 6, 2016, 12:42:34 AM3/6/16
to ns-3-users
Dear all,

Have you guys tried multiple base stations? My setting is pretty simple, that I have N UEs and N remote host, the only traffic is each remote host is using BulkSender to send data to its UE via TCP.

When there is only 1 base station, the instant sending rate (I got it from .pcap) is pretty stable for each UE; however, when I move to multiple base stations, this instant sending rate is really fluctuating. I attached two figures, each line is the instant rate for one UE, you can see stable rate for 1 base station case but fluctuated results for 3 base stations. 

I think this is due to inter-channel interference, but I don't know if this makes sense and want to understand this more. Is this due to DL MCS value fluctuating caused by interference? Or the interference that affect UL acknowledgement which eventually affect the DL rate (interaction of TCP)?

To troubleshoot this, I want to make sure the DL sending rate is not limited. To do so, I really want to hide L3 packet loss, in order to create the bufferbloat phenomena which is well-known in LTE. But I don't know what should I do (not a LTE expert), should I use AM RLC instead of UM? Will that help? (I assume the default is UM). 

Thanks.
1_bs.png
3_bs.png

Xing Xu

unread,
Mar 6, 2016, 2:59:04 PM3/6/16
to ns-3-users
Also, when there are multiple base stations, do you guys observe that, some UEs have significantly worse performance than others (but they collocate together so MCS should be similar)? By significantly worse, I mean, almost no throughput compare to others.

Tommaso Pecorella

unread,
Mar 6, 2016, 3:27:32 PM3/6/16
to ns-3-users
Hi,

unless the various BS are set to use different, orthogonal resources, then the graph below is pretty much expected. The interference will lead to random packet losses and a variation in throughput, even reaching zero sometimes.

T.
Reply all
Reply to author
Forward
0 new messages