High Performance Rate Adaptation on IEEE 802.11 Networks

Abstract

IEEE 802.11 standard has evolved from the basic transmission rates in early days to multiple rates today with advanced encoding and antenna techniques. The performance of IEEE 802.11 wireless networks is supposed to benefit from the support of multiple transmission rates. To take advantage of multiple available rates, a strategy is required to choose the most appropriate rate in transmission: rate adaptation in wireless networks is the selection of the optimal transmission rate for data frames under current channel conditions. For this purpose, a rate adaptation scheme must assess the channel condition and then accordingly adjust the data rate if necessary. In this dissertation, we extensively survey the rate adaptation schemes in literature. We progressively proposed three strategies to address some open problems in rate adaptation. First, we exploit the periodical mandatory beacon frames in IEEE 802.11 networks to estimate the initial rate for a stream of data frames and yield the scheme Beacon Assisted Rate Adaptation (BARA). In IEEE 802.11 networks, particularly with basic CSMA/CA access, the assessment of channel condition may become complex because frame losses likely result from channel condition degradation or transmission collision, and each type of frame loss requires a different response in rate adjustment. To diagnose the cause of a frame loss, we propose the basic rate retransmission technique and a rate adaptation scheme Loss Differentiated Rate Adaptation (LDRA), based on Signal-Noise-Ratio (SNR). However, SNR and Received Signal Strength Indication (RSSI) are deemed not to be good channel condition indicators. Then we investigate rate adaptation schemes based on frame losses with loss differentiation ability. We discuss some anomalies observed in implementing a Linux based rate adaptation testbed. Based on the observations, we propose Effective Rate Adaptation (ERA) to effectively adapt rates in mixed frame lossy environments. ERA judiciously exploits the IEEE 802.11 fragmentation mechanism to accurately diagnose the cause of a loss and responds accordingly. We analytically prove that the fragmentation mechanism incurs less overhead and is more effective than using RTS/CTS at the basic (lowest) rate for rate adaptation in a collision dominated environment. ERA also takes effective actions to overcome abnormalities observed in other rate adaptation schemes. For performance evaluation, we first simulate all our three schemes on ns-2. Moreover, we implement ERA and four selected most recent representative rate adaptation schemes on a Linux based testbed and evaluate them with extensive experiments in both controlled and public field tests. Experiment results demonstrate that our proposed schemes outperform their peers in most scenarios.