Energy Efficient Mac Layer Design for Wireless Sensor Networks

Energy efficient communication is a critical design objective for wireless sensor networks which are usually highly energy constrained. In addition, the throughput and latency performance is also important for several sensor network applications. To simultaneously achieve the seemingly contradictory goals, this dissertation identifies the three major sources of energy wastage in communications, i.e., idle listening, overhearing, and packet retransmissions, and proposes three mechanisms to optimize the energy consumption while maintaining high throughput and low latency. To deal with the idle listening problem, we design a new low duty-cycle MAC layer protocol called Convergent MAC (CMAC). CMAC can work at low duty cycles and requires no synchronization when there is no traffic. When carrying traffic, CMAC first uses anycast to wake up forwarding nodes, and then converges gradually from route-suboptimal anycast to route-optimal unicast. Experiments and simulations show that CMAC significantly outperforms other duty cycling protocols in terms of latency, throughput and energy efficiency. The MAC layer anycast technique is also an efficient technique to cope with low link quality and interference in wireless sensor networks. By utilizing the nature of broadcast wireless communication medium and allowing multiple nodes in the forwarding set to compete to be the packet forwarder, links with poor reception quality but good progress in a given routing metric space can be opportunistically used to ii forward packets. However, the impact of unreliable communication in the reverse channel on anycasting has not been studied before. The second part of this dissertation analyzes the impact of unreliability of reverse links on the performance of existing anycast protocols, proposes a new metric characterizing the number of transmissions in the network for anycast based MAC protocols, and presents an efficient solution for computing the forwarding sets. The third part of this dissertation is dedicated to optimizing the schedule of packet retransmissions. CSMA relies on carrier sensing to decide if retransmissions should be performed immediately. However, in cases where the poor channel quality persists or packet losses are due to interference undetectable by carrier sensing, the channel assessment alone is not a good indicator of successful transmissions. To schedule retransmissions at appropriate moments, we propose a new technique called transmission pushback to reduce such losses by delaying retransmissions. This technique overcomes periods of poor channel quality while ensuring a throughput matching the incoming packet rate. In order to determine the optimal pushback period, we devise an adaptive channel prediction technique based on estimating the parameters of a simple hidden Markov model (HMM) which represents the channel. We dynamically update the parameters of the HMM based solely on the ACK sequence for the previous packet transmissions. By considering both the packet incoming rate and the packet loss pattern, the appropriate pushback period is calculated and applied for future retransmissions.