Vector-Valued KLMS based Multiple Target Range and Velocity Estimation using IEEE 802.11p Waveform for Autonomous Vehicle

An autonomous vehicle equipped with radar is used to monitor the movement of other vehicles (targets) in the vicinity. Radars on autonomous vehicles are primarily used for accurate estimation of targets' (neighboring vehicles) range and velocity to avoid collision and fatal accidents. The existing linear estimator based on Fourier transform, in the multi-target scenario, is prone to yield inaccurate estimates of targets' parameters. In this paper, to jointly estimate the range and velocity of multiple vehicles, an adaptive non-linear estimator based on vector-valued kernel least mean square algorithm is proposed. The proposed estimator optimizes the convex cost function in reproducing kernel Hilbert space and consequently yields better estimates. Further, the complexity of the proposed estimator is reduced by incorporating the sparsification technique. Analytical expressions for Cramer-Rao lower bound are derived to evaluate the performance of the proposed estimator. Lastly, simulation results validate the performance of the proposed estimator via convergence to a lower normalized mean square error with reasonable computational complexity.