12.07.2015 Views

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

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

3. Physical Layer <strong>Modeling</strong> <strong>and</strong> <strong>LTE</strong> <strong>System</strong> <strong>Level</strong> SimulationOf special interest is <strong>the</strong> feedback-wise more complex case <strong>of</strong> CLSM, where <strong>the</strong>optimum precoder is precalculated at trace generation <strong>and</strong> <strong>the</strong> optimum rank chosenat run time, for which <strong>the</strong> L2S model closely approximates link level results.3.2.1.1. Complexity EvaluationIn this section, <strong>the</strong> run-time complexity <strong>of</strong> system level simulations is compared tothat <strong>of</strong> link level simulations, showing a significant reduction in simulation run timewhen employing <strong>the</strong> L2S model. Hence, it validates <strong>the</strong> statement, much-emphasizedin this <strong>the</strong>sis, that a link abstraction model allows for significantly faster simulationtimes compared to detailed link level simulations.The same single-user, single-cell described above is simulated for <strong>the</strong> <strong>LTE</strong> channelb<strong>and</strong>widths <strong>of</strong> 1.4, 3, 5, <strong>and</strong> 10 MHz (link <strong>and</strong> system level), <strong>and</strong> additionally for<strong>the</strong> 20 MHz b<strong>and</strong>width case for system level. As from <strong>the</strong> values, listed in Table 3.4,link level simulation run times scale linearly with <strong>the</strong> number <strong>of</strong> RBs, <strong>the</strong> link levelsimulation time for <strong>the</strong> 20 MHz case has been extrapolated from <strong>the</strong> existing values.Table 3.4.: Simulation run time comparison in seconds. Marked in bold face are <strong>the</strong> systemlevel simulation times, followed by <strong>the</strong> link level simulation times.Single TX1.4 MHz 3 MHz 5 MHz 10 MHz 20 MHz6 RBs 15 RBs 25 RBs 50 RBs 100 RBs1×1 28/282 28/703 28/1 115 28/2 300 30/-1×2 30/278 28/666 28/1 089 27/2 248 32/-TxD 2×2 27/614 28/1 542 29/2 552 28/5 086 29/-OLSMCLSM2×2 32/1 308 33/3 287 34/5 446 39/10 978 33/-4×2 32/1 426 32/3 842 33/6 598 35/15 040 36/-4×4 33/2 203 37/6 029 36/10 428 40/24 531 44/-2×2 33/358 32/901 33/1 509 32/3 399 34/-4×2 31/493 35/1 257 34/2 020 33/4 394 34/-4×4 37/874 34/2 310 36/3 563 38/7 535 42/-The almost constant simulation run time <strong>of</strong> <strong>the</strong> system level simulator can be explainedby <strong>the</strong> <strong>of</strong>floading <strong>of</strong> <strong>the</strong> most computationally-intensive task, which is <strong>the</strong>channel trace generation. Although a detailed complexity analysis <strong>of</strong> <strong>the</strong> system levelsimulator has not been performed, <strong>the</strong> shown simulation results indicate that, whensimulating larger b<strong>and</strong>widths, <strong>the</strong> complexity increase <strong>of</strong> <strong>the</strong> b<strong>and</strong>width-dependentpart <strong>of</strong> <strong>the</strong> L2S model (in this scenario <strong>the</strong> link quality model <strong>and</strong> <strong>the</strong> CQI <strong>and</strong>,when applicable, RI feedback 7 ) are almost negligible in comparison to <strong>the</strong> overallrun time.7 In a single-user scenario, no actual scheduling is performed. However, scheduling algorithms do46

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!