On 2-Angular-Coverage in Wireless Visual Sensor Network Deployment for 3D Indoor Monitoring

Abstract

Listen

Translate article

Extensive efforts have been devoted to achieve k-coverage for wireless sensor networks (WSNs) deployed in the sensing area by considering omni-directional sensing ranges as seen in scalar temperature sensors, where the larger k is, fault tolerance and monitoring quality is better. However, in wireless visual sensor networks (WVSNs), the sensing range of a visual sensor is often directional and modeled as a convex triangular shape in 2D and a pyramid-shape in 3D, which renders the k-coverage solutions proposed for WSNs incapable of solving the problem in WVSNs. Moreover, a number of WVSN applications consider 3D indoor space monitoring, which makes previous solutions proposed for 2D WVSN coverage less effective. In this paper, we take an initial step to tackle this challenging problem by considering 2-coverage for WVSNs in 3D indoor space monitoring. To maximize the information that can be captured by 2 visual sensors covering a 3D location, we propose to deploy the visual sensors from different directional angles and further extend 2-coverage to to denote such unique requirements in WVSNs. We demonstrate our simulated WVSN testing platform which emulates a 3D space and highlights our objective to provide efficient 2-angular-coverage in the indoor environment by minimizing the number of visual sensors required to cover the full regions through precision angle view monitoring. We formulate the general problem in a continuous space and then discretize our model via a lattice based approach. In addition, we also create a strategy to determine the degree of coverage for the WVSN, where a given location needs to be covered by at least 2 visual sensors that fulfill the angular-coverage requirement. We design an efficient greedy heuristic and an enhanced-Depth First Search algorithm to solve the discrete version problem, where the latter, if given enough time can return an optimal solution. By extensive simulations, we demonstrate that both our solutions can achieve superior performance on minimizing the number of required visual sensors for 2-angular-coverage in 3D indoor space monitoring.