Low power body sensor network design based on relaying of creeping waves in the unlicensed 2.4GHz band
Body Sensor Networks are an important enabling technology for future applications in remote medical diagnostics. Practical deployments of these systems have only recently edged closer to viability, due in part to advances in low power electronics and System-On-Chip devices. Wireless communication between these sensors remains a daunting challenge, and designers typically leverage existing industrial standards designed for applications with significantly different communications requirements. This Thesis proposes a wireless communications platform designed specifically for body mounted sensors, exploiting a phenomenon in electromagnetic wave propagation known as a creeping wave. Relaying of these waves leads to a highly reliable body sensor network with very low power consumption in the unlicensed 2.4 GHz band.
A link budget is derived based on the creeping wave component of the transmitted signal, which is then used to design a spread spectrum wireless transceiver. Significant attention is given to interference mitigation, allowing the system to co-exist with other wireless devices on the internationally unlicensed band. Fading statistics from both anechoic and high multipath scenarios are used to define a channel model for the system. The link budget and channel model lead to the proposed use of relaying as a power savings technique, and this concept is a core feature of the design. This technique is shown to provide reliable total body coverage with very low transmission power, a result that has eluded body sensor networks to date.
Various relaying topologies are discussed, and robust operation for highly mobile users is achieved via sensor handoffs, a concept that resembles a similar solution in cellular networks. The design extends to define a polling protocol and packet structures.
Objective performance metrics are defined, and the proposed system is evaluated in line with these metrics. The power reduction of the suggested approach is analyzed by comparing the network lifetime and energy-per-bit to those of a reference system offering the same quality of service without relaying. The analysis results in generic closed form expressions of significant gains. The improvement in network lifetime increases with the number of sensors and settles at approximately 8x104, 7x106, 7x10 7 and 3x108 for 2,4,6 and 8 relaying nodes respectively. The energy-per-bit is shown to decrease by 2, 116, 828 and 2567 for 2, 4, 6 and 8 relay nodes respectively.