Cooperative Cross-Correlation Algorithm to Optimize Linearity of Fused RF Sensors

Abstract

Fusing data from multiple sensors can improve performance beyond that of any individual sensor and at worst is limited to the best individual performance among the sensors. In this work, nominally identical RF sensors are spaced closely enough such that they receive equivalent signals. However, the sensors have independent variable front-end attenuations and thermal noise. If all pair-wise cross-correlations of signals among the sensors are averaged, a proper choice of attenuation settings can optimize the linearity of the result as measured by signal to distortion and noise (SINAD) ratio. With the receiver gain and 1dB compression point (IP1) as variables, a closed-form expression for the optimal attenuation settings is derived for two sensors and is extended for any number of coherent sensors with phase-aligned reception. The expression is verified with experimental measurements up to four sensors. However, in reality, the sensors measure differing signal powers, which violates the original assumptions of the derivation. Nonetheless, the derived result is robust to slight variations since the measured SINAD is within 1dB of the optimum as long as the difference in measured signal power between two sensors is less than 5dB. For the cases in which the measured signal powers differ by more than 5dB, this work presents an algorithm to adjust the attenuation values and the number of signal captures to compensate the loss in linearity. All results are corroborated with experimental measurements taken with four low-cost RadioHound sensors.