Book Description

"In many sensor network applications, the sensor readings are inter-node correlated. In such cases, efficient gathering of sensor readings requires distributed compression. Distributed source coding provides practical solutions for compression of these correlated readings when the appropriate rates for the marginal encoding is known at the sensor nodes. In this thesis, we present a data-gathering technique for sensor networks that exploits correlation between sensor data at different locations in the network. Contrary to distributed source coding, our method does not rely on knowledge of the source correlation model in each node although this knowledge is required at the decoder node. Similar to network coding, our proposed method (which we call Quantized Network Coding) propagates mixtures of packets through the network. The main conceptual difference between our technique and other existing methods is that Quantized Network Coding operates on the field of real numbers and not on a finite field. In this thesis, we study our quantized network coding in both lossless and lossy networks.In the study of lossless networks, we discuss the theoretical foundations for our data gathering technique. By exploiting principles borrowed from compressed sensing, we show that the proposed technique can achieve a good approximation of the sensor readings at the sink node with only a few packets received, and that this approximation gets progressively better as the number of received packets increases. Our first approach is to explain the theoretical foundations for sparse recovery from quantized network coded packets based on an analysis of the Restricted Isometry Property of the corresponding measurement matrices. Extensive simulations comparing the proposed Quantized Network Coding to classic network coding and packet forwarding scenarios demonstrate the delay/distortion advantage of quantized network coding. Furthermore, we discuss the advantages of quantized network coding in a Bayesian scenario where the prior of the sensor readings is available at the decoder node. For such Bayesian scenarios, we also discuss the adaptation of a message passing based decoding algorithm with the aid of simulations.To study the practicality of quantized network coding in lossy networks, we adapt it into the IEEE 802.15.4 standard which characterizes low rate wireless communication for sensor networks. This is done by developing a comprehensive implementation of the PHY and MAC layers of the standard and then adjusting the MAC layer settings to match with our requirements. Our computer simulations using the developed implementation show a significant decrease of the delay in many simulation scenarios. The results obtained using this implementation show more advantages for quantized network coding compared to classic routing based protocols especially for high packet drop rates." --