Energy efficiency and reliability in wireless sensor networks

Abstract

Motivated by their tremendous potentials in civil and military applications, wireless sensor networks have raised tremendous research interests in recent years. The fundamental function of wireless sensor networks is the gathering of information from a covered area. Information thus gathered by the sensor nodes are processed by the sink nodes according to the applications deployed in the network. Therefore, delivering information from sensor nodes to sink nodes effectively is one of the fundamental challenges in wireless sensor networks. To achieve flexibility in deployment, sensor nodes are generally battery-driven. Furthermore, battery replacement is very difficult, if at all possible, due to the random deployment nature of wireless sensor networks. Thus, sensor nodes have limited operational time and the operational time of the sensor networks is also limited. Hence, energy efficiency is another fundamental challenge in wireless sensor networks. To improve the performance of wireless sensor networks, data delivery should be both energy efficient and effective, which is the focus of this thesis. First, we observe the conflict of the two objectives -viz., the application performance and the network lifetime—when optimizing the performance of wireless sensor networks. We find that the tradeoff between them can be studied by investigating the interaction between the network lifetime maximization problem and the rate allocation problem. Then, we formulate the tradeoff problem as a constrained convex optimization problem by introducing a tradeoff factor. We first tackle this problem at the transport layer, and then from a cross layer perspective. Using Lagrange dual decomposition, algorithms are obtained to achieve the best tradeoff. Then, we note that the notion of reliability in certain wireless sensor networks is probabilistic. Based on this, the problem of providing minimum energy probabilistic reliable data delivery is studied. First we tackle the problem at the MAC layer, using p-persistent CSMA protocols. We derive adaptive algorithms to tune the persistence probabilities of these p-persistent CSMA protocols. Then, we obtain the optimal allocation of per-hop reliability requirements, which further reduces the energy consumption with adaptive p-persistent CSMA protocols at the MAC layer.