Window-based Rate Adaptation in 802.11n Wireless Networks

Abstract

In this paper we use real experiments to study MIMO 802.11n Rate Adaptation (RA) on a programmable AP platform. The results reveal nontrivial inter-, intra- mode opportunistic gains with respect to transient channel variations in both MIMO operation modes of Diversity and Spatial Multiplexing. To exploit these short-term gains, we design and implement a new Window-based Rate Adaptation (WRA) algorithm. WRA differently from existing MIMO 802.11n proposals, runs an independent RA in each MIMO mode in parallel, to address the unique characteristics of each MIMO mode. To isolate the best candidate rates, it maintains and adjusts a Sliding Rate-Selection Window for each individual mode, and opportunistically selects the best-goodput rate among the candidates on a per-transmission basis. WRA also applies novel techniques to limit probing overhead and to improve responsiveness to mobility. Our experiments show that WRA gives 52.8% goodput gains over Atheros MIMO RA algorithm, and 72.5% gains over practical legacy 802.11a/b/g designs in field trials.