QoS aware routing with admission control for video sensor networks

Wireless sensor networks (WSNs) have secured a role in many emerging applications spanning environmental, ecological, industrial, and commercial domains in which data are periodically sourced from sensors and used in monitoring and control. The ubiquity of network-based video streaming, largely in the form of monitoring equipment installed in administrative domains (cities, campuses, offices, highway systems) has been possible by increases in capability of low-cost cameras and computational devices. However the streaming of video data over wireless sensor networks has the nature of heavy data transmission load, relatively long data transmission period, severe inter- and intra-path interference, and unbalanced network energy consumption under severe energy constraints. These characteristics make it very difficult to achieve meaningful throughput performance for video data delivery over WSN. In this dissertation we focus on the development of practical throughput aware video data routing techniques optimized to data delivery cost to address the challenge of QoS support in end-to-end throughput achievement for bandwidth demanding video delivery in a large-scale, battery-operated sensor network.;We utilize video data streaming throughput as the main metric to quantify the video delivery performance. We present a benchmark data routing algorithm to assure the throughput performance (QoS) of a data path via eliminating the impact of inter-path interference through path isolation. Extending the benchmark algorithm, we develop an interference-tolerant data routing algorithm to improve the general data egress rate of the entire network with some tolerant sacrifices of the throughput performance on each stream. The routing algorithms construct throughput-guaranteed end-to-end video delivery data paths based on accurate video delivery performance estimation. We present two analytical models in conjunction with simple proactive admission and congestion control strategies to enable accurate end-to-end data delivery throughput estimation with considerations of data stream's transmission rate, location, and relative positions. Simulations demonstrate that our estimation models are highly consistent with the measurement of the real data transmission scenario and our proposed data delivery scheme adapts to the optimal data rate about three times faster than a traditional CTS/RTS scheme without generating any jitter as observed in CTS/RTS. In addition, we exploit the option of deploying mobile base stations to support improved video delivery QoS. The proposed mobile base station deployment strategy improves the video delivery performances with the best effort to minimize the data delivery cost. As a complementary solution for the deployment of our routing scheme over WSNs, we develop a new code dissemination framework for generic WSN applications. Simulation and prototype implementation illustrate that our framework is almost five times faster for application code dissemination than traditional over the air programming strategy. Another contribution of this dissertation is the constellation graph of routing algorithms. It presents a clear way of selecting the most appropriate routing algorithm for a specific WSN application.