Transmission Schemes for Resource-constrained Wireless Sensor Networks

Abstract

Wireless sensor networks (WSN) provide a versatile monitoring infrastructure to track physical processes for autonomous and manual decision-making. In WSN and wireless networked control systems, sensors observe physical processes and transmit measurements to remote estimators or a fusion center that tracks process parameters. The information update rate from sensors is limited by the number of communication channels and the sensor nodes’ energy storage capabilities. Commonly, sensors share a limited number of communication channels, and if the number of transmitting sensors exceeds the number of channels, interference occurs.Furthermore, sensor nodes in WSN are usually not supported by power grids, and instead, they rely on small batteries, from which radio communication consumes a significant amount of energy. Therefore, to prolong the lifetime of sensor nodes, there is a need for approaches that reduce the number of data transmissions from the sensors without compromising the data accuracy. Also, given the limited resources for communication channels, the data transmission should be coordinated to maximize system utility. This thesis develops transmission schemes that determine when a sensor should transmit an observation to achieve a high level of state-estimation in channel- and energy-constrained WSN for remote estimation. For a channel-constrained WSN, we exploit spatio-temporal dependencies among sensors to improve the overall estimation accuracy for remote estimators tracking different processes. We derive an optimal scheduling policy that minimizes the time average mean squared error (MSE) by modeling the scheduling problem as a Markov decision process. We also consider event-triggered transmission schemes, where a sensor transmits a measurement if it exceeds a predefined threshold. As an extension of the dual prediction scheme framework, a cost-aware dual prediction scheme is presented to further reduce data transmission in a WSN where sensors observe non-stationary processes. Finally, we consider a system of multiple sensors implementing threshold-based transmission over a limited number of shared channels, resulting in collisions. From statistical parameters, we derive optimal thresholds minimizing the MSE.