[LENA] Observation of "different rate with same amount of PRBs"

[Xing Xu]

Dear all,

I'm doing a research project and currently experimenting with NS3 LTE. I found one interesting observation but I don't know why: I found that for some cases, same amount of PRBs will be allocated to 2 competing UEs, while these 2 UEs' rate are pretty different.

Setup:

To be concrete, there is 1 BS and 2 UEs, no interference or fading, UEs are collocated, only 10m away to the BS.

Two RH are sending data to UEs, but with different rate (so only DL traffic):

 - One UE is a heavy UE, whose RH's sending rate is fixed at 3Mb/s;

 - The other UE is a light UE, I vary its RH's sending rate, from 0.1Mb/s to 1Mb/s.

FYI, both RH's sending rate is pretty low, such traffic would not congest the BS.

Observation & Question:

I found that, when the light UE's sending rate is at 0.5Mb/s, it already gets the same amount of PRBs compared to the heavy UE (both of them get 6000 PRBs per second, and FYI, I see in total 12000 PRBs for every second), but its rate is only 1/6 of the heavy UE. I wonder what is the reason of this? Why the same amount of PRBs can achieve so different rate?

A relevant question, does this observation related to congestion? In my settings there is no congestion at all (traffic is low), so PRBs are used inefficiently? And if I make the BS congested, then all PRBs will be used equally efficiently and I will see same amount of PRBs can achieve the same rate? (FYI, I observed that using my method of counting PRBs, there are always 12000 PRBs per second, and I see that actually all the PRBs are allocated in my experiment, let me know if this does not make sense to you).

How I count PRB per UE:

The way I'm counting each UE's PRB is, by modifying proportional-fair scheduler (pf-ff-mac-scheduler.cc), and maintaining a counter for each UE (rnti) and log each UE's allocated # of PRBs.

(More detail: in pf-ff-mac-scheduler.cc, for each PRB, it calculates the maximum rcqi and this PRB goes to the UE with highest rcqi. I add 1 for the UE with maximum rcqi for every PRB. I clear counters for every second to get # of PRBs per second).

BTW, this post is related to my previous post: https://groups.google.com/forum/#!topic/ns-3-users/-Ob0oMqaCfE

The two posts describe the same issue.

Thanks,

Xing

[Xing Xu]

I figured out the issue. For short, the scheduler will try to scheduler and serve UE if such UE has TX queue size greater than 0, but if the PRB can transmit more than this UE's TX queue size, it will become relatively inefficient. E.g., for a light user, if it's queue size is 1000B but given its MCS value each PRB can send 2000B, then it's efficiency is pretty low, ~50%.