An Energy Scalable Computational Array for Sensor Signal Processing

Harvesting energy from environmental sources can extend wireless sensor network node lifetime beyond the limits of battery technology. However, the output power from an energy harvester is highly variable. We propose a domain-specific computational array which maximizes sensor performance by matching system power consumption to the available scavenged energy through power scalable approximate signal processing. The array consists of distributed arithmetic (DA) based functional units coupled with a reconfigurable interconnect structure. Each unit implements several core linear and nonlinear signal processing functions in an area efficient manner which also minimizes leakage power. Several sensor DSP applications (FIR, IIR, and FFT) have been mapped onto the array. Post-layout simulations confirm that the proposed domain-specific computational array is energy efficient and energy scalable