System Level Modeling and Optimization of the LTE Downlink

This thesis presents the design and application of a Link-to-System (L2S) model capable of predicting the downlink throughput performance of cellular mobile networks based on the 3GPP Long Term Evolution (LTE) standard. The aim of a L2S model is to accurately abstract the physical layer at a fraction of the complexity of detailed link level simulations. Thus, it dramatically reduces the necessary simulation run time and by extension enables the simulation of much more complex scenarios. The thesis is divided in four main parts. First, the basics of the LTE standard are presented, with the link abstraction model being presented afterwards. Extensions for the L2S model for the cases of Hybrid Automatic Repeat reQuest (HARQ) and imperfect channel state information are presented in the third section. In the last chapter, the performance of the application of Fractional Frequency Reuse (FFR) to LTE is evaluated by means of the developed model. The presented LTE link abstraction model employs a zero-forcing receiver and is based on the calculation of the post-equalization Signal to Interference and Noise Ratio (SINR which for the Closed Loop Spatial Multiplexing (CLSM) MIMO transmit mode employs a high-Signal to Noise Ratio (SNR) precoder approximation. The designed model is capable of accurately predicting the throughput performance of the following LTE-defined transmit modes and antenna configurations, as validated against link level simulations: Single transmit antenna with MRC combining; 2×2 Transmit Diversity (TxD); 2×2, 4×2, and 4×4 Open Loop Spatial Multiplexing (OLSM); and 2×2, 4×2, and 4×4 CLSM. The results presented in this thesis have been obtained by the Matlab implementation of the L2S model, which is released including its source code as the Vienna LTE System Level Simulator. Reproducibility scripts for each of the previous works on which this thesis is based are also available for download, which enables the presented results to be independently replicated. As of March 2013, the simulator has already been downloaded more than 22 000 times and is being used both by universities and industry.