Lossy Data Aggregation with Network Coding in Stand-Alone Wireless Sensor Networks

Abstract

This work focuses on a special type of wireless sensor networks (WSNs) that we refer to as a stand alone network. These networks operate in harsh and extreme environments where data collection is done only occassionally. Typical examples include habitat monitoring systems, monitoring systems in chemical plans, etc. Given resource constrained operation of a sensor network where the nodes are battery powered and buffer sizes are limited, efficient methods for in-network data storage and its subsequent fast and reliable transmission to a gateway are desirable. To save scarse resources and to prolong the life time of the whole network, the lossy data aggregation method can be applied. It is especially viable in the networks where several sensors are measuring the same physical phenomenum and only average values of sensor readings are of interest. In this paper, we present a method for effiecient lossy data aggregation using a network coding technique. We demonstrate that with low communication and processing overhead, a gateway can retrieve data fast from the network and the retrieved values are close to the sample means of the collected measurements. This method requires a special choice of coefficients for linear data encoding: the condition number of the coefficient matrix should be small and close to one.