Optimal transmission policies for a star wireless sensor network

Composed of wireless sensors directly or indirectly connected to the information sink, a wireless sensor network is typically employed to survey a coverage area. Sensors directly connected to the information sink form a star structure. Regardless of network topology, the central star structure shoulders the burden of relaying all information that reaches the sink. Elevated traffic in this region increases mutual interference and probability of congestion, which in turn accelerate battery depletion and ultimately limits the total amount of data that reaches the sink. It is the goal of this dissertation to provide novel insight into factors that affect communication in a star wireless sensor network. The data received at the sink until the last sensor expires, referred to as the data volume, is introduced as a suitable performance criterion for networks of limited energy sensors. Data volume takes into consideration both the network capacity and the period in which it is available. With an information theoretic approach, optimum transmit powers and scheduling policies that maximizes the data volume for a star wireless sensor network are analytical derived. While accounting for sensor processing powers, solutions for both the stationary and time-varying wireless channels are presented. Parallel solutions derived without considering processing power in the sensor model are used to emphasize its importance. The effects of processing power, initial battery energy and network size on the data volume and the activity duration of the network are discussed. Once the optimum solution is known, various methods are introduced to reduce the computational complexity of calculating it. Less computation allows calculations for larger networks and networks with more energy. Calculation shortcuts do not compromise the optimality of the solution and simulations show their significant effects. Suboptimal transmission policies that produce data volume close to the optimum scheme but require significantly less computation are also discussed. Allowing sensors that are not transmitting to enter low-activity cycles is regarded as an energy saving measure. An analytical derivation for the optimum transmission and sleep policy is presented. Simulations show that allowing sleep cycles extends sensor lifetimes and increases data volume.