System Level Modeling and Optimization of the LTE Downlink

3. Physical Layer Modeling and LTE System Level Simulationrun time [s]10 410 310 2link levelsystem levelSISOSIMO 1x2TxD 2x2OLSM 2x2OLSM 4x2OLSM 4x4CLSM 2x2CLSM 4x2CLSM 4x41.4MHz 5MHz 10MHz 20MHzFigure 3.18: Simulation run time. Legend and colors as in Figures 3.16 and 3.17.While for OLSM the gains are higher (up to a 1 150x speed-up, as shown in Table3.5), significant gains can be achieved for CLSM, of high interest due to theemployed more elaborate MIMO feedback and higher model accuracy. In this case,a speed-up gain of up to 359x can be obtained when considering a 20 MHz bandwidth.Such a computational complexity reduction enables performing multi-usersimulations with high channel bandwidths, necessary to evaluate complex schedulingscenarios or multi-user gain with more practical simulation time durations, such asin the example in Appendix D.Table 3.5.: Calculated system level simulator speed-up compared to link level simulation runtime, 20 MHz bandwidth scenario.Single TX TxD OLSM CLSM1×1 1×2 2×2 2×2 4×2 4×4 2×2 4×2 4×4speed-up 153x 140x 351x 665x 836x 1 150x 200x 259x 359x3.2.2. Multi-cellThe L2S model presented in this thesis, on which the LTE system level simulatoris based, enables simple simulation of dense networks. As such, it is an idealtool for performance modeling of heterogeneous networks, where an LTE macrocelllayer is coupled with a tier of smaller, low-power cells (e.g., pico-, or femto-cells).Although issues related to LTE femtocell deployment are currently already beinginvestigated [100–102], results commonly lack accurate link abstraction modelingdue to the extra overhead necessary for it.increase in complexity with the number of PHY resources (RBs) they allocate. In multi-usersimulations scenarios where scheduling algorithms are applied, simulation run time may not,depending on the complexity of the scheduling algorithm, stay constant over bandwidth.47