3. April 2017 at 18:38 #7873Nakrop JinapornParticipant
Dear The Vienna Team,
As mentioned in the Vienna book, REUSE-1 has been used for macro and femtocell tiers. Can we examine the FFR in this two-tier heterogeneous cellular networks?
I try running LTE_sim_main_launcher_demo_femtocells.m with the FFR configuration for 1*1 (Mode 1) and 2*2 (CLSM) as follows:
LTE_config.scheduler = ‘FFR’;
LTE_config.scheduler_params.FR_scheduler.scheduler = ‘prop fair Sun’;
LTE_config.scheduler_params.FR_scheduler.av_window = 25;
LTE_config.scheduler_params.PR_scheduler.scheduler = ‘prop fair Sun’;
LTE_config.scheduler_params.PR_scheduler.av_window = 25;
LTE_config.FFR_params.beta_FR = 0.75;
LTE_config.FFR_params.SINR_threshold_value = 12.75;
All results (peak, average, edge, cell throughput and fairness index) with FFR were lower than the case without FFR. To the best of my knowledge, the results should be improved by using FFR in particular edge throughput.
I also checked the scenario without femtocell deployments. Peak, average and cell throughput could be improved significantly by using FFR, while edge throughput and fairness index with FFR were smaller than the case without FFR.
Maybe this is caused by inappropriate beta_FR and SINR_threshold. I am not sure whether this FFR for macro-femto networks is supported by the Vienna simulator.
Please let me know if you have any suggestions. Your advice is very important for my research direction.
NJ16. May 2017 at 18:03 #8092Martin MüllerMember
First for the comparison of a scenario without femtos: when comparing the results with and without FFR and the values that you specified (2×2, prop fair scheduler), I see higher values for FFR than for plain reuse-1.
For using femtos with or without FFR: There my simulation results also show a better performance when FFR is not used. I am not sure however if this is not expected, due to the larger interference, e.g., for edge users assigned to femtos.
Could you maybe first verify, that for a FFR setup without femtos, the performance is improved compared to not using FFR?
Martin28. May 2017 at 1:58 #8135NaserParticipant
I think the performance degradation with FFR setup is a result of “low RB occupancy” !!
I’ve already asked this question in this form since 2 months with no answer so far !
with FFR, there is a low RB occupancy, which lead to loss in TP !
Naser31. July 2017 at 11:59 #8367Martin MüllerMember
the results-calculation for “low RB occupancy” is not valid when using FFR, since the RB_occupancy information from the scheduler does not fit (RBs for the different re-use parts are all occupied but are averaged over all FFR parts and therefore not correct – this will have to be adjusted).
As I wrote earlier: when simulating FFR for your specified values (only macros), I see an improvement in the results. For combining femto-cells and FFR, this does not have to be the case, since there are overlapping areas of frequency re-use from femtos and macros. Since the gain of FFR comes from the better spatial separation, this might destroy any performance gain, when using het-nets.
- You must be logged in to reply to this topic.