Behavioural reconfigurable and adaptive data reduction in body sensor networks

Body area sensor networks have been attracting more and more applications which focus on human behaviour and monitoring, ranging from simple positioning to medical applications. These BSNs inherit unique specifications since are composed of light-weight embedded systems. In this paper, we focus on energy and lifetime requirements of these systems which is one of the most challenging design constraints. We study this problem from the angle of data compression and sampling which are both known to be very efficient in energy reduction specially when large amount of data is to be transmitted wirelessly. We introduce the notion of functional compression which utilises classes of data patterns to efficiently represent information through regenerative functions. Furthermore, we propose a reconfigurable compression methods which dynamically uses different compression methods to optimise compression ratio and energy savings. Later we study how sampling rate can be adaptively altered based on the behaviour of data pattern for further reduction in sample counts. We use the data from a wearable sensing system to illustrate the effectiveness of these methods which utilise the context and behaviour of the environment in system optimisation process.