On the Integration of Approximate Computing in GNSS Signal Processing for Improved Energy-Efficiency

Abstract

Continuous miniaturization of GNSS receivers has led to reduction of energy consumption over the years. Yet, these remain one of the highest energy consuming sensors in embedded electronics, rendering their integration difficult in platforms with high energy-constrains. In parallel, the developments of novel computation paradigms, such as Approximate Computing, have opened new possibilities to further reduce the receiver energy consumption by trading-off computation accuracy. In this paper, we analyze the application of AxC operators inside the GNSS processing chain. More precisely, we apply various AxC multipliers inside the correlation operations used for signal tracking and look at the impact on the post-correlation SNR values. We show that the degradation in SNR values can be low, depending on each AxC multipliers implementation, thus offering significant energy saving opportunities in correlators.