Error minimization and energy conservation by predicting data in wireless body sensor networks using artificial neural network and analysis of error

Wireless Body area Sensor Network (WBSN) is a recent concept that can dramatically benefit healthcare applications through advances in wireless technology. Physiological and biokinetic parameters that require continuous monitoring are sensed by small and lightweight body sensors that transmit the values of these parameters over wireless links for monitoring at the other end. The sensors employed in WBSNs are limited in resources, with battery power being at the premium. Conservation of energy used by the network has a direct bearing on the longevity of the network. Therefore, there is no need to send data periodically and need to transmit selectively when needed. This paper presents a dual framework for predicting when to transfer physiological parameters in such a network that could save energy consumption while maintaining error to minimum level. The framework utilizes an artificial neural network (ANN) for prediction that not only saves energy, but also does it with lesser error than popular prediction algorithms. A comparison of performance of five data prediction algorithms in predicting physiological data is presented. The amount of network energy saved as a result of prediction is also considered in detail.