Dynamic Resource Allocation for Compressive-Sensing-Based Wireless Visual Sensor Networks with Energy Harvesting

Abstract

Wireless visual sensor network (WVSN) is emerging with many potential applications. The lifetime of a WVSN is seriously dependent on the energy shored in the battery of its sensor nodes as well as the adopted compression and resource allocation scheme. In this paper, we use the energy harvesting to provide almost perpetual operation of the networks and compressed-sensing-based encoding to decrease the power consumption of acquiring visual information at the front-end sensors. We propose a dynamic algorithm to jointly allocate power for both compressive-sensing-based visual information acquisition and data transmission, as well as the available bandwidth under energy harvesting and stability constraints. A virtual energy queue is introduced to control the resource allocation and the measurement rate in each time slot. The algorithm can guarantee the stability of the visual data queues in all sensors and achieve near-optimal performance.