An Adaptive Kalman Filter based Digital Phase Detector for All Digital Phase-Locked Loop

This work proposes an adaptive Kalman filter (AKF) based digital phase detector that compresses the phase detector's (PD) output noise in all-digital phase locked loop (ADPLL). The ADPLL is modeled mathematically and the noise sources are extracted. An AKF is designed based on the ADPLL model and is inserted after the PD to optimize its output noise. This AKF takes the advantage of the statistical properties of the noise and adapt its parameter to make itself an optimal filter. Simulation is done in MATLAB and the result reveals the correctness and stability of the design. The PD gets a 16.9 dB noise power decrease in the simulation.