The effect of noise, fading and interference on the performance of scheduling algorithms in a cross-layer downlink model

Abstract

A system model combining space time block coding and beamforming at the PHY layer along with various scheduling algorithms at the MAC layer is used to study the effect of noise, fading and interference on the performance of a system that simultaneously serves two users. System performance expressed in terms of capacity, is assessed for line of sight (LoS) and non-LoS (NLoS) environments, for a range of SNRs, and for different number of antenna elements that are used for combating interference via beamforming. As expected, system capacity is higher for higher SNRs and for direct LoS environments for all scheduling algorithms. The improvement in capacity is also achieved due to an increased number of antenna elements at the PHY layer that leads to narrower beam widths, and aid in minimizing the interference from other simultaneously served users. The scheduling algorithm proposed by D. Arora and P. Agathoklis (2005), that explicitly considers the angular spread between simultaneously served users for combating interference, performs consistently better than the greedy and the round-robin (RR) scheduling algorithms. The results suggest that jointly addressing the PHY and MAC layer issues is important for achieving maximal system performance.