Adaptive Rate Control based on Collision and Frame Aggregation Awareness

Abstract

The innovative advances in mobile technologies have facilitated remarkable growth in the use of mobile devices such as smart phones, e-book readers, and tablet PCs. Since a significant amount of wireless traffic is generated from these devices, the shared channel medium may suffer from serious mutual interference. In addition, various kinds of fading weaken the stability of channel quality. Though IEEE 802.11 standards specify an available set of transmission rates as one of the approaches to accommodate channel dynamics, they do not discuss in detail when and how to use them. Rate Adaptation (RA) algorithms try to solve this decision problem by determining the most appropriate data rate for the instantaneous wireless channel condition. The primary focus of Rate Adaptation studies for legacy standards such as IEEE 802.11 a/b/g has been on when to increase/decrease transmission rates or how to differentiate between collision-induced loss and channel disorder-induced error. However, with the advent of novel Multiple Input Multiple Output (MIMO), Channel Bonding, and Frame Aggregation/Block Acknowledgement technologies, it has become possible to enable Gigabit connectivity in wireless communication. At the same time, these new technologies add a new dimension of problems with rate adaptation. Even though much research has attempted to address the non-monotonic relationship between data rate and throughput via exploring MIMO technology or by identifying the usability of another channel space for bonding, little research has analyzed the impact of Frame Aggregation. This paper begins by surveying existing research on Rate Adaptation (RA) algorithms. A wide range of RAs are introduced and classified based on various design factors. Secondly, a new RA scheme called RACD (Rate Adaptation with Collision Differentiation) is presented. RACD is capable of differentiating the reasons for packet errors by using CTS-to-self control packets. After classifying the sources of errors, RACD responds accordingly with actions such as initiating RTS/CTS handshake or adjusting Contention Window size. Extensive simulation results confirm that RACD significantly improves performance, especially in simulated congestion scenarios. Finally, the new features introduced in the state of the art IEEE 802.11n standard are overviewed and discussed. Out of the three dominant innovations specified in the new amendment, the impact of one innovation, Frame Aggregation (FA), on RA algorithm design is examined from various perspectives with a case study. Based on the analysis and findings from the case study, a novel AARA (Aggregation Aware Rate Adaptation) algorithm is proposed and evaluated with extensive experiments conducted in controlled and repeatable environments. The performance enhancement over representative existing RA algorithms is validated with throughput gains up to 473% in various experimental scenarios. Potential performance benefits on RA design from Frame Aggregation awareness is discussed in detail. Although the experimental work and the proposed AARA algorithm provide insight into the significant influence of FA on RA algorithm design, a number of critical issues remain which should be considered for further research. This dissertation concludes with a discussion of potential future work related to FA and RA for IEEE 802.11 Wireless Networks.