Wireless sensor networks: Energy efficiency, delay guarantee and fault tolerance

Wireless sensor systems contain new types of computing machines and run different kinds of applications in a variety of physical environments, which offers many challenges. The main design constraints of many sensor network applications are energy efficiency, delay guarantee and fault tolerance. Since these constraints impact multiple protocol layers, a joint design across all layers becomes indispensable in designing a protocol for sensor networks. In this thesis, we take the cross-layer design approach to improve the performance of protocol layers by taking into account parameters in other layers instead of merging them into one layer to preserve modularity. In addition to the interactions from lower to higher layers, the interactions from higher to lower layers are included to exploit application-specific characteristics of sensor networks such as many-to-one communication, collaborative processing and asymmetry of constraints.;We first illustrate the use of sensor networks in traffic applications. We show that sensor nodes provide a vehicle detection accuracy higher than that of the current standard vehicle detection scheme, inductive loops. Moreover, the high spatial density associated with the magnetic sensors allows us to classify the vehicles and even re-identify them. We then propose a TDMA (Time Division Multiple Access) MAC protocol, PEDAMACS, that achieves both energy efficiency and delay guarantee. Although it is known that there is a tradeoff between delay and energy, PEDAMACS demonstrates that one can achieve both of them by using special characteristics of sensor networks: many-to-one and asymmetric communication. We then develop TDMA scheduling algorithms that take into account the many-to-one communication characteristic of sensor networks to decrease the delay.;We formulate the lifetime maximization as a linear programming (LP) problem with the goal of determining optimal routing paths in the network. We extend the energy efficient routing to introduce anytime delay guarantee. We show that the traditional way of introducing delay constraint via capacity constraints only provide average delay performance. Anytime delay guarantee is a restriction on the routing trees after the decomposition of the LP solution. A distributed algorithm is proposed to illustrate the trade-off between delay and energy efficiency. We also extend the energy efficient routing to include multiple paths between each source and destination in the network to provide the resiliency of the network to link failures. We furthermore investigate the effect of physical layer and application layer characteristics on multi-hop routing. (Abstract shortened by UMI.).