Kernel LMS-Based Estimation Techniques for Radar Systems

Relationship between the delay and Doppler shift with the radar return is nonlinear in nature. Therefore, a nonlinear estimator based on sparse kernel least mean square algorithm is proposed. Further, an adaptive kernel width optimization technique is proposed to lower the computational complexity and for simple implementation. An expression for the Cramer-Rao lower bound is derived and validated for the proposed estimator over linear frequency modulated, and orthogonal frequency division multiplexed radar systems.