Improved TCP performance in wireless IP networks through enhanced opportunistic scheduling algorithms

Current and next-generation wireless networks rely on multiuser diversity and scheduling techniques, such as the commonly used proportional fair (PF) algorithm, to achieve greater system throughput and higher efficiencies for wireless data applications over a time-varying wireless channel. We show that the variability of the inter-scheduling intervals, as introduced by the PF scheduling algorithm, can have adverse effects on TCP and its congestion control mechanism and lead to spurious timeouts and unnecessarily low throughput. We propose an enhanced scheduling algorithm that is explicitly tuned towards throughput performance at the TCP layer. However, this algorithm does not use any explicit information from the TCP layer and solely relies on information readily available at the link layer at which the scheduler resides. The performance of this improved algorithm is assessed through extensive simulations to show an average TCP throughput improvement of 12% compared to PF. In addition, the TCP-level fairness across all users is increased, as is the individual user throughput.