Context Compression: Using Principal Component Analysis for Efficient Wireless Communications

In certain settings, like for example a Wireless Sensor Network (WSN), contextual information (context) needs to be disseminated between nodes and then interpreted. Dissemination is typically performed through a wireless network infrastructure where resources are scarce. Our focus is on the design/implementation of a context compression scheme that tries to minimize the pieces of information exchanged over the network. Our scheme heavily relies on the multi-value(vectorial) nature of context dissemination messages that flow throughout the network. We adopt the Principal Component Analysis and determine the statistical dependencies between the context vector components. We manage to reduce (compress)the transmitted contextual information down to the identified principal components. A comparative assessment with other energy efficient models is reported indicating the capability of the proposed model to minimize resource consumption.