Loss aware sample packetization strategy for improvement of body sensor data analysis

With limited bandwidth and high channel loss in wireless body area networks (WBANs), it remains a challenging issue to transmit time series body sensor data for satisfactory data analytics. In this paper, we investigate the interesting problem of sample packetization, i.e., assembling multiple samples in each packet before transmission over the wireless link. We first illustrate the ineffectiveness of the default sequential packetization under channel loss. Then, we propose Loss Aware Sample Packetization (LASP), a heuristic loss aware sample packetization strategy to improve the data quality for improvement of the sensor data analysis. The core idea of LASP is to minimize contiguous sample loss during the data transmission, thereby significantly increasing the accuracy of data recovery at the aggregator side. Extensive simulations are conducted to evaluate the effectiveness of the proposed approach on data recovery quality as well as data analysis accuracy. Results show that LASP yields a nicer sample loss pattern, therefore significantly improve the data recovery quality.