Bias analysis and removal of a microphone array based road traffic speed estimator

Recently, a maximum likelihood estimate has been proposed for road vehicle speed based on two omnidirectional microphones. It is observed that this estimate may be severely biased if the target is moving fast and/or the acoustic signal present high frequency components. We undertake a bias analysis to explain this effect and present a modification which is capable of drastically reducing the bias at the same computational complexity as the original estimate.