Agile and robust resource management in CSMA wireless networks

With the recent push towards wireless broadband, and user migration towards mobile devices, it has become imperative for wireless networks to support higher network capacity, increased battery life, and greater mobility. While a combination of both local- and wide-area wireless technologies will be needed to meet these requirements, we believe that carrier-sense multiple-access (CSMA) wireless networks, if designed well, can play an important role in the future. Traditional CSMA wireless networks, which form the basis for today's ubiquitous wireless LAN technology---IEEE 802.11 or WiFi, cannot satisfy the stated requirements mainly because they lack efficient resource management (or resource parameter adaptation) techniques. Several fundamental characteristics of these networks, and practical implementation challenges, also limit the applicability of solutions from other domains. Taken together, these considerations force us to fundamentally re-think the design of resource parameter adaptation for CSMA wireless networks. In this dissertation, we first identify that, to satisfy the requirements, resource management techniques for CSMA wireless networks should possess two fundamental, but conflicting properties, agility, and robustness. Briefly, to provide increased bandwidth to mobile users, agile solutions are required that exploit opportunities for improved performance; at the same time, solutions cannot afford to compromise on link robustness. In addition, we realize that striking this trade-off effectively calls for different solutions in indoor and outdoor environments. To prove these hypotheses, we present the design and implementation of two resource management frameworks, Symphony and Sonata, for indoor and outdoor environments respectively. Indoors, Symphony increases network capacity and battery life for mobile clients by addressing the classical problem of joint, per-link, transmit power control and rate adaptation. For improved robustness, Symphony uses novel mechanisms based on measuring the expected transmission time (ETT), and the utility of RTS/CTS frames, while relying on a learning approach to converge quickly to the right resource parameter choice. Outdoors, the Sonata framework introduces a novel and fundamental tradeoff between directionality and base station diversity for uplink transmissions. Using a new location-based approach for improved parameter convergence, Sonata is able strike the agility-robustness tradeoff effectively. Together, these frameworks prove that, achieving the right balance between agility and robustness can enable CSMA wireless networks to transition to the wireless broadband era.