| Recently, wireless visual sensor networks start to gain more and more attentions due to their promising applications. In this dissertation, we develop several schemes which enable image compression and transmission in wireless sensor networks. Three schemes are developed, motivated by characteristics commonplace in wireless sensor networks; constrained resource and high unpredictability. We also analyze the scalability of hierarchical routing, commonly suggested for large scale sensor networks.; The first scheme, presented in Chapter 3, addresses the problem of minimizing the energy dissipation for transmitting compressed images over a multi-hop network subject to a specific image quality constraint. By varying parameters of a wavelet based image compression algorithm, we introduce a heuristic algorithm to select the optimal image compression parameters to minimize total energy dissipation given the network conditions and image quality constraints.; The second scheme of image compression, presented in Chapter 4, copes with the constrained resource (i.e. computation power and/or energy) of individual sensor nodes. Distributed image compression is proposed as a means to overcome this limitation by sharing the processing tasks. It has the additional benefit of extending the overall lifetime of the network by distributing the computation load among otherwise idle processors.; In addition to the constrained resources, unpredictable errors are inevitable in wireless sensor networks. In Chapter 5, a novel scheme is presented with consideration of two causes of transmission errors; wireless link impairment and node failure. We propose an "in-network" diversity combining scheme and evaluate its performance in terms of received image quality and energy consumption over a multihop wireless sensor network.; The image compression and transmission schemes we proposed assume a cluster-based routing algorithm is in place. Here, we consider the overhead of cluster-based routing in Chapter 6. Two network models are analyzed; a two-level hierarchy in an infinite Manhattan grid, and a multi-level hierarchy in a finite torus network topology. For each model, expressions of various components of the routing overhead are derived as functions of the traffic patterns modeled. |
