Compressed Sensing in Multi-Hop Large-Scale Wireless Sensor Networks Based on Routing Topology Tomography

Abstract

Data acquisition from multi-hop large-scale outdoor wireless sensor network (WSN) deployments for environmental monitoring is full of challenges. This is because of the severe resource constraints on tiny battery-operated motes (e.g., bandwidth, memory, power, and computing capacity), the data acquisition volume from large-scale WSNs, and the highly dynamic wireless link conditions in outdoor harsh communication environments. We present a novel compressed sensing approach, which can recover the sensing data at the sink with high fidelity when a very few data packets need to be collected, leading to a significant reduction of the network transmissions and thus an extension of the WSN lifetime. Interplaying with the dynamic WSN routing topology, the proposed approach is both efficient and simple to implement on the resource-constrained motes without motes' storing of any part of the random projection matrix, as opposed to other existing compressed sensing-based schemes. We further propose a systematic method via machine learning to find a suitable representation basis, for any given WSN deployment and data field, which is both sparse and incoherent with the random projection matrix in compressed sensing for data collection. We validate our approach and evaluate its performance using a real-world outdoor multihop WSN testbed deployment in situ. The results demonstrate that our approach significantly outperforms existing compressed sensing approaches by reducing data recovery errors by an order of magnitude for the entire WSN observation field while drastically reducing wireless communication costs at the same time.