MAC/PHY Co-Design of CSMA Wireless Networks Using Software Radios

In the past decade, CSMA based protocols have spawned numerous network standards (e.g., the WiFi family), and played an indispensable role in improving the ubiquity of wireless networks. However, the accelerating evolution of CSMA brings unprecedented challenges, especially the coexistence between different network architectures and communications devices. Meanwhile, many intrinsic limitations of CSMA keep haunting its derivatives, such as ZigBee, WiFi, and mesh networks. We observe that most of these problems root in the piecemeal improvement of CSMA based network protocols, which abstract the advancement of PHY technology merely as an increase of data rate. Hence, the benefits of new PHY technologies are either not fully exploited, or they may harm the performance of existing network protocols due to poor interoperability. In this thesis, We identify such problems through extensive testbed experiments, propose a joint architecture to synthesize the basic MAC operations of CSMA with novel PHY algorithms, and leverage the reconfigurability of software radios to validate the design.

We have established five major contributions along this line of research. First, we propose CSMA/CR, which augments CSMA with a PHY layer collision resolution algorithm, in order to eliminate unnecessary collision avoidance operations. We further leverage CSMA/CR to enable delay-optimal broadcast and asynchronous cooperative relaying in wireless networks. Second, we identify a channel starvation problem caused by the rigid spectrum allocation scheme in CSMA- based WiFi networks, and propose adaptive subcarrier nulling, which enables coexistence of heterogeneous channel widths using a fine-grained OFDM spectrum access protocol. Third, we observe severe failure of CSMA in coordinating heterogeneous CSMA networks (e.g., WiFi and ZigBee), and propose cooperative busy-tone, which improves the visibility of ZigBee devices to WiFi by using a proxy-based carrier signaling protocol. Fourth, by analyzing real-world traffic traces, we pinpoint the dominant energy cost of idle listening in CSMA wireless networks, and propose E-MiLi that reduces the power consumption of idle listening by downclocking the radio, and incorporates a novel signal processing algorithm to ensure accurate packet detection at low sampling rate. Finally, we introduce NEMOx, a new network architecture and protocol that exploits the PHY layer cooperation of distributed antennas to improve the capacity of CSMA based wireless LANs.