Towards safe usage of COTS wireless devices in medical settings : a profiling - policing framework for WBAN against WiFi interference

Abstract

Medical applications nowadays are increasingly demanding Wireless Body Area Networks (WBAN). Medical WBANs usually use Industrial Scientific Medical (ISM) band due to free license and abundant supply of low cost Commercially Off-the-Shelf (COTS) devices. In the 2.4GHz ISM band, WBANs usually adopt low power wireless technologies, such as Zigbee, Bluetooth, and IEEE 802.15.6. These wireless technologies may suffer from co-channel interference from WiFi due to power asymmetry. In this thesis, we study several challenging issues on WBAN-WiFi coexistence. First, the two most widely deployed wireless technologies in the 2.4GHz ISM band are WiFi and Bluetooth. The pervasive existence of WiFi and Bluetooth threatens co-channel medical WBAN. Without loss of generality, we evaluate a typical WBAN scheme in the context of medical multi-parameter monitoring under WiFi and Bluetooth interference. The results show that WiFi is a major threat to WBAN; while Bluetooth is not. Second, we propose a general WiFi-WBAN coexistence design, called WiCop. WiCop is a cross-MAC-PHY-layer solution. It suppresses WiFi interferer by transmitting customized WiFi compliant signals. Our experiments show that WiCop can double the Packet Reception Rate (PRR) of WBAN under intense WiFi interference. Moreover, WiCop requires no modification of existing WiFi and WBAN standards. Third, we are also interested in passive profiling using sniffers. As each sniffer can only monitor one channel at a time, and cover a fixed area, Sniffer Channel Assignment (SCA) affects directly monitoring quality. Among the algorithms solving SCA, annealed Gibbs sampler is superior due to its distributed nature. We propose several improvements to annealed Gibbs sampler that offer faster convergence and higher chance to reach global optima.