Power-efficient-coded architecture for distributed wireless sensing

Abstract

Most distributed wireless sensor systems avoid using error correction codes because of the added complexity and power consumption of the decoders. Therefore these uncoded sensor systems are not capable of providing a robust communication link in noisy environments. This leads to a high packet loss, limits the communication range and increases the number of sensors required to cover a region. A novel architecture with several simple sensors and a few sensors with complex decoders is proposed. This approach is power efficient, practical and at the same time provides better coverage and link quality. Several coding schemes are studied with the overall goal of maximised performance, while minimising the decoding complexity, hence saving sensor nodes' power. Furthermore, in both additive white Gaussian noise and frequency-flat Rayleigh fading channels a code is optimised to provide the lowest possible bit error rate for a specific memory constraint. It is demonstrated that overall power consumption for a coded wireless sensor system can be less than its uncoded counterpart, even considering the decoding power consumption. The proposed approach is targeted at real-time wireless environmental monitoring, where the information flow is unidirectional, justifying the usage of complex decoders in a few super-nodes.