Leveraging Human Mobility for Communication in Body Area Networks

When a person is walking the RF signal strength of an on-body communication link may exhibit significant fluctuation with peak-to-peak amplitudes beyond 20 dB. Instantaneous signal strength may be noisy, but the smoothed signal typically exhibits a period that matches the person's stride period. We present an opportunistic packet scheduler that extracts a set of Received Signal Strength Indicator (RSSI) samples from application traffic and utilizes an accelerometer to monitor the person's gait cycle. Packets are scheduled based on previous RSSI peaks and the current offset within the gait cycle. We formulate the task of finding a nonoverlapping packet schedule among the different body area network (BAN) devices as a linear programming problem and present an efficient way of solving it with the simplex method. Our experimental evaluation shows that outdoors BAN links with PRR (ratio of correctly received to transmitted packets) values between 50% and 90% can typically be turned into reliable links with PRR values well above 90%. Indoors the improvements are smaller, but still significant at low transmission power. The main price is an increase in packet delivery latency. The energy consumed by the devices is marginal, but the coordinator spends more energy due to signal processing.