Coding in 802.11 WLANs

Forward error correction (FEC) coding is widely used in communication systems to correct transmission errors. In IEEE 802.11a/g transmitters, convolutional codes are used for FEC at the physical (PHY) layer. As is typical in wireless systems, only a limited choice of pre-specified coding rates is supported. These are implemented in hardware and thus difficult to change, and the coding rates are selected with point to point operation in mind. This thesis is concerned with using FEC coding in 802.11 WLANs in more interesting ways that are better aligned with application requirements. For example, coding to support multicast traffic rather than simple point to point traffic; coding that is cognisant of the multiuser nature of the wireless channel; and coding which takes account of delay requirements as well as losses. We consider layering additional coding on top of the existing 802.11 PHY layer coding, and investigate the tradeoff between higher layer coding and PHY layer modulation and FEC coding as well as MAC layer scheduling. Firstly we consider the joint multicast performance of higher-layer fountain coding concatenated with 802.11a/g OFDM PHY modulation/coding. A study on the optimal choice of PHY rates with and without fountain coding is carried out for standard 802.11 WLANs. We find that, in contrast to studies in cellular networks, in 802.11a/g WLANs the PHY rate that optimizes uncoded multicast performance is also close to optimal for fountain-coded multicast traffic. This indicates that in 802.11a/g WLANs cross-layer rate control for higher-layer fountain coding concatenated with physical layer modulation and FEC would bring few benefits. Secondly, using experimental measurements taken in an outdoor environment, we model the channel provided by outdoor 802.11 links as a hybrid binary symmetric/packet erasure channel. This hybrid channel offers capacity increases of more than 100% compared to a conventional packet erasure channel (PEC) over a wide range of RSSIs. Based upon the established channel model, we further consider the potential performance gains of adopting a binary symmetric channel (BSC) paradigm for multi-destination aggregations in 802.11 WLANs. We consider two BSC-based higher-layer coding approaches, i.e. superposition coding and a simpler time-sharing coding, for multi-destination aggregated packets. The performance results for both unicast and multicast traffic, taking account of MAC layer overheads, demonstrate that increases in network throughput of more than 100% are possible over a wide range of channel conditions, and that the simpler time-sharing approach yields most of these gains and have minor loss of performance. viii Finally, we consider the proportional fair allocation of high-layer coding rates and airtimes in 802.11 WLANs, taking link losses and delay constraints into account. We find that a layered approach of separating MAC scheduling and higher-layer coding rate selection is optimal. The proportional fair coding rate and airtime allocation (i) assigns equal total airtime (i.e. airtime including both successful and failed transmissions) to every station in a WLAN, (ii) the station airtimes sum to unity (ensuring operation at the rate region boundary), and (iii) the optimal coding rate is selected to maximise goodput (treating packets decoded after the delay deadline as losses).