Accelerometer-based fall detection using optimized ZigBee data streaming

Wireless body-area sensor networks (WBSNs) are key components of e-health solutions. Wearable wireless sensors can monitor and collect many different physiological parameters accurately, economically and efficiently. In this work we focus on WBSN for fall detection applications, where the real-time nature of I/O data streams is of critical importance. Additionally, this generation of alarms promises to maximize system life. Throughput and energy efficiency of the communication protocol must also be carefully optimized. In this article we investigate ZigBee’s ability to meet WBSN requirements, with higher communication efficiency and lower power consumption than a Bluetooth serial port profile (SPP) based solution. As a case study we implemented an accelerometer-based fall detection algorithm, able to detect eight different fall typologies by means of a single sensor worn on the subjects’ waist. This algorithm has a low computational complexity and can be processed on an embedded platform. Fall simulations were performed by three voluntary subjects. Preliminary results are promising and show excellent values for both sensitivity and specificity. This case study showed how a ZigBee-based network can be used for high throughput WBSN scenarios.