Camera management in wireless visual sensor networks based on a quality aware resource allocation scheme

Wireless visual sensor networks (WVSNs) are widely applicable to industrial surveillance, target tracking, and health monitoring applications. Energy conservation in WVSNs is one of the most important challenges because sensors are battery-operated. In addition, the visual data which is provided by such networks is much more voluminous than that of conventional wireless sensor networks, so the compression of visual data in WVSNs is inevitable. However, this process is highly energy-consumer. Apart from that, the quality of images delivered at the fusion center is highly affected by the number of quantization bits. Therefore, in this paper, a compression strategy that sets the number of quantization bits to meet the desired level of quality of service (QoS) has been employed. It is noted that the entropy of the image received at the fusion center and the distortion caused by compression are regarded as QoS criteria. We also aim at presenting a resource allocation algorithm along with a node selection scheme. To do that, we study the energy optimization problem in WVSNs. In this regard, an energy minimization problem that is subject to some QoS constraints is formed. We formulate the problem as a convex problem, and then we propose a solution based on Karush-Kuhn-Tucker (KKT) conditions. As a result, in the first stage, the optimal resources, represented by the number of quantization bits and the power consumed in the compression unit, are computed leading to a quality-aware resource allocation (QARA) algorithm. Next, based on the optimal value of resources for each visual sensor (VS) node, a node selection (NS) algorithm is proposed to select an appropriate node to capture an image from a target at a given time. The simulation results illustrate that our proposed algorithm is superior to benchmark algorithms in terms of energy consumption and convergence time.