Spatial usage and power control in multihop wireless networks

In wireless networks, because of co-channel interference, concurrent transmitters must be chosen such that they provide a minimal amount of interference to each other. In a simple path-loss based propagation model, this implies that the concurrent transmitters and receivers must be spaced sufficiently far apart. When network traffic is high, space becomes a limited resource for which every node has to compete. Thus, improving spatial reuse among wireless nodes or reducing spatial usage of wireless transmissions is crucial to improving overall network throughput. Among the techniques for improving spatial reuse, transmit power control is fundamental. In this dissertation, we first analyze the impact of transmit power on potential network throughput. To do this, we propose a spatial usage metric and then investigate the impact of transmit power on the spatial usage of single and multihop communications. Motivated by our analysis, we propose a Media Access Control (MAC) and physical layer power control scheme, Optimized Transmit Power (OTP), to balance the spatial usage of each individual transmission and co-channel interference. This scheme assumes the worst possible interference at the receiver and reduces transmit power to be just great enough to guarantee reliable signal reception. Further study shows that OTP is overly conservative, because the worst case interference does not occur much of the time. Therefore, we develop an Enhanced OTP (EOTP) to tradeoff a possible occasional collision for lower power and better spatial usage. Our simulation results show that EOTP outperforms OTP and both schemes improve overall network throughput to a moderate or significant degree. Because MAC layer power control schemes favor short sender-receiver distances, we study a mini-hop routing strategy that discovers routes consisting of short distance hops and develop a Mini-Hop Routing (MHR) protocol. When combining MHR with EOTP, network performance, including throughput, end-to-end packet delivery latency, and routing overhead, is improved substantially. Finally, we study a load-sensitive routing strategy that bypasses hot spots and utilizes idle space. Our investigation demonstrates that the existing blind flooding technique is able to circumvent hot spots to a significant degree. Load-sensitive routing outperforms the blind flooding technique substantially only when flow lifetime is short or node mobility is high.