Implementation of energy efficient WBAN using IEEE 802.15.6 scheduled access MAC with fast DWT based backhaul data compression for e-Healthcare

Abstract

This work describes the implementation of a complete Wireless Body Area Network (WBAN) that is capable of monitoring multiple physiological signals of a patient by means of IEEE 802.15.6 scheduled access MAC protocol. In the WBAN setup, data from multiple sensors are sent to a Body Network Controller (BNC) using low power transceivers. To this end, the BNC is designed to multiplex the data from multiple sensors, and send them to a remote server over the Internet using a backhaul cellular network, thereby enabling ubiquitous remote health monitoring. Furthermore, to facilitate an energy efficient backhaul transmission that incurs low data transfer costs to the users, we introduce the concept of data compression at the BNC. In this regard, we propose a fast Discrete Wavelet Transform (DWT) based data compression algorithm at the BNC, termed herein as B-DWT, that is implementable in real-time using the limited on-board resources. The remote server is configured to accept data from multiple patients, de-multiplex different data of a single patient and store them in a database for pervasive access. Issues related to the hardware implementation of sensor nodes and BNC, and the design of the scheduled access mechanism and B-DWT are addressed. Detailed performance analysis of the WBAN is performed in OPNET simulator to determine the optimum allocation intervals for the sensor nodes that maximizes network capacity while maintaining a frame delay constraint. Further, in order to prolong the battery life of sensor nodes, we obtain the optimal payload sizes that maximizes their energy efficiency. Additionally, through implementation of B-DWT at the BNC we determine the optimal wavelet filer and compression levels, that allow maximum data compression within acceptable limits of information loss. The resulting B-DWT algorithm is shown to outperform traditional DWT with significant gains in execution speed and low memory footprint at the BNC.