System Level Modeling and Optimization of the LTE Downlink
System Level Modeling and Optimization of the LTE Downlink
System Level Modeling and Optimization of the LTE Downlink
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
D. Evaluation <strong>of</strong> Multi-User GainAvg. cell throughput [Mbit/s]240220200180160140120100800Throughput 4x45 10 15 20 25 30number <strong>of</strong> UEs/cellFairness <strong>of</strong> UE throughput10.90.80.70.60.50.40.30.20.100Fairness 4x45 10 15 20 25 30number <strong>of</strong> UEs/cellRound Robin Proportional fair Best CQI fitFigure D.3: 4×4 CLSM transmit mode multi-user gain results. Left: throughput results.Right: fairness results. Vertical lines mark <strong>the</strong> 95% confidence intervals.achievable unless unrealistically high SNRs are present, even in <strong>the</strong> case <strong>of</strong> uncorrelatedchannels, as previously shown on Figure 3.17.An analysis such as <strong>the</strong> one conducted here also quantifies <strong>the</strong> reduction in multi-usergain seen between <strong>the</strong> round robin <strong>and</strong> proportional fair schedulers when switchingfrom a 2×2 to a 4×4 antenna configuration due to <strong>the</strong> higher suboptimality (dueto <strong>the</strong> higher number <strong>of</strong> degrees <strong>of</strong> freedom), <strong>of</strong> <strong>the</strong> PMI <strong>and</strong> RI feedback in 4×4MIMO.The results <strong>of</strong> practically-employed schedulers, such as round robin <strong>and</strong> proportionalfair can in this manner be compared to <strong>the</strong> upper threshold <strong>of</strong> <strong>the</strong> best CQI scheduler,which is not used due to it starving <strong>the</strong> majority <strong>of</strong> UEs in a cell.Alternately, <strong>and</strong> similarly to <strong>the</strong> evaluation in Chapter 5, throughput <strong>and</strong> fairness(see Section 5.4) results can be evaluated toge<strong>the</strong>r. As observed from Figures D.1 toD.3, proportional fair scheduling consistently converges to <strong>the</strong> same fairness value<strong>of</strong> roughly 0.7, with a decrease in multi-user gain for higher antenna counts, while<strong>the</strong> varying fairness results <strong>of</strong> round robin hint at it operating at different points <strong>of</strong><strong>the</strong> fairness-throughput trade-<strong>of</strong>f shown in Section 5.6.99