Joint Routing and 2D Transform Optimization for Irregular Sensor Network Grids Using Wavelet Lifting
IPSN '08 Proceedings of the 7th international conference on Information processing in sensor networks
Transform-based distributed data gathering
IEEE Transactions on Signal Processing
Practical data compression in wireless sensor networks: A survey
Journal of Network and Computer Applications
Compression in wireless sensor networks: A survey and comparative evaluation
ACM Transactions on Sensor Networks (TOSN)
| Hi-index | 0.00 |
We address the problem of compression for wireless sensor networks from a signal processing point of view. Each of the sensors has limited power, and acquires data that should be sent to a central node. The final goal is to have a reconstructed version of the sampled field at the central node, with the sensors spending as little energy as possible. We propose two distributed wavelet compression algorithms for multihop, distributed sensor networks based on the lifting scheme. The first algorithm introduces extra transmissions in the network so as to give the nodes access to neighboring data, making it possible to compute the transform coefficients. The second algorithm exploits the natural data flow in the network to aggregate data by computing partial wavelet coefficients that are refined as the data flows towards the central node. We study the impact of quantization of partial data on the final distortion obtained, and propose a rule to determine how many bits should be used to quantize the partial information so as to achieve a target level of degradation, in the form of added distortion as compared to calculating coefficients without partially quantized data. We also introduce a framework where the network is represented as a graph, with sensors associated to nodes, transmission costs to edge weights, and different coding schemes associated to modes of operation. Dynamic programming techniques are used to optimize the network, assigning coding schemes to each of the nodes so that the overall energy cost is minimum. The proposed coding methods and optimization are extended to bidimensional networks, with irregular sensor deployment. In this scenario, our algorithm operates by first selecting a routing strategy through the network. Then, for each route, an optimal combination of data representation algorithms is selected. We then propose a simple heuristic to determine the data representation technique to use once path merges are taken into consideration.