Deployment, Coverage and Network Optimization in Wireless Video Sensor Networks for 3D Indoor Monitoring

摘要

As a result of extensive research over the past decade or so, Wireless Sensor Networks (WSNs) have evolved into a well established technology for industry, environmental and medical applications. However, traditional WSNs employ such sensors as thermal or photo light resistors that are often modeled with simple omni-directional sensing ranges, which focus only on scalar data within the sensing environment. In contrast, the sensing range of a wireless video sensor is directional and capable of providing more detailed video information about the sensing field. Additionally, with the introduction of modern features in non-fixed focus cameras such as the Pan, Tilt and Zoom (PTZ), the sensing range of a video sensor can be further regarded as a fan-shape in 2D and pyramid-shape in 3D. Such uniqueness attributed to wireless video sensors and the challenges associated with deployment restrictions of indoor monitoring make the traditional sensor coverage, deployment and networked solutions in 2D sensing model environments for WSNs ineffective and inapplicable in solving the Wireless Video Sensor Network (WVSN) issues for 3D indoor space, thus calling for novel solutions.;In this dissertation, we propose optimization techniques and develop solutions that will address the coverage, deployment and network issues associated within Wireless Video Sensor Networks for a 3D indoor environment. We first model the general problem in a continuous 3D space to minimize the total number of required video sensors to monitor a given 3D indoor region. We then convert it into a discrete version problem by incorporating 3D grids, which can achieve arbitrary approximation precision by adjusting the grid granularity. Due in part to the uniqueness of the visual sensor directional sensing range, we propose to exploit the directional feature to determine the optimal angular-coverage of each deployed visual sensor. Thus, we propose to deploy the visual sensors from divergent directional angles and further extend k-coverage to "k-angular-coverage'', while ensuring connectivity within the network. We then propose a series of mechanisms to handle obstacles in the 3D environment. We develop efficient greedy heuristic solutions that integrate all these aforementioned considerations one by one and can yield high quality results. Based on this, we also propose enhanced Depth First Search (DFS) algorithms that can not only further improve the solution quality, but also return optimal results if given enough time. Our extensive simulations demonstrate the superiority of both our greedy heuristic and enhanced DFS solutions. Finally, this dissertation discusses some future research directions such as in-network traffic routing and scheduling issues.