Enhancing the availability of wireless visual sensor networks: Selecting redundant nodes in networks with occlusion

摘要

Availability is a major design issue that should play an important role when deploying and operating wireless sensor networks, specially for critical monitoring applications. Actually, sensing redundancy can be exploited to enhance the attainable availability level of sensor networks, since redundant nodes can replace faulty nodes. When employing camera-enabled sensors, the perception of sensing redundancy is considerably changed when compared to scalar sensors, with direct impact on network availability. In such way, some characteristics as deployment density, viewing angle and sensing range should be properly evaluated in wireless visual sensor networks, in order to better estimate the network availability. Nevertheless, when deploying visual sensors on occluded environments, viewed areas and resulted overlapping may be significantly altered, redefining sensing redundancy. We then propose an algorithm to automatically select redundant nodes in wireless visual sensor networks deployed on areas with occlusion, according to network configurations and application availability requirements. Additionally, an algorithm to adjust cameras' orientations in occluded environments is also proposed. Mathematical assessment of the proposed algorithms are performed, allowing the discussion of how parameters of deployed networks can influence applications monitoring availability.