Post-Processing Phase Noise Mitigation Performance Comparison in a Coherent Distributed Passive Radar System Andrew N Morabito ∗ , John D Sahr ∗ , Zac M.P. Berkowitz ∗ , Laura E. Vertatschitsch ∗ ∗ University of Washington, Department of Electrical Engineering, Campus Box 352500, Seattle, WA 98195-2500, USA e-mail:morabito@u.washington.edu Abstract: Phase noise in a radar system propagates through processing algorithms and affects the quality of the output radar data products by making targets harder to detect from noise and potentially confounding multiple targets. Usually referred to as autofocus algorithms, techniques for mitigating the effects of phase noise in these images have been proposed and studied in other applications including terrestrial telescope imaging and synthetic-aperture radar. We examine the effectiveness of autofocus algorithms as they attempt to mitigate phase noise in a coherent distributed passive radar system. We also propose an alternate technique based on deconvolution of power spectral densities, which has produced better results as well as requiring far less computation since no optimization problem exists within this method. 1. Introduction Phase noise can be introduced into radar systems in a variety of ways, such as by instability in the system’s local oscillator(s) and by atmospheric effects. It is well known that [8] phase noise on raw data in a radar system affects the output image by causing a target range’s true Doppler ambiguity to be convolved with the power spectral density (PSD) of the phase noise process: A φ [ν]= A 0 [ν] ∗ F τ  e jφ[τ]  , (1) where e jφ[τ] is the phase noise autocorrelation, so F τ  e jφ[τ]  is the phase noise power spectrum. This convolution causes large Doppler bins to leak into adjacent bins, thereby misrepresenting the measurable Doppler spread of a target and potentially confounding multiple targets or even obscuring them altogether from detection. We will examine processing techniques that attempt to mitigate the effect of phase noise using data from the Manastash Ridge Radar (MRR). MRR is a coherent distributed passive radar system based at the University of Washington, Seattle WA. Several reports have described MRR and its observations [5, 6]. By observing FM broadcasts, MRR detects meter scale plasma density irregularities in the ionosphere. For a given range, r, MRR computes the ambiguity signal as 1 : A[ν; r]= |F n {x ∗ [n]y[n + r]}| 2 , (2) which is the magnitude-squared of the Fourier transform of the product of the conjugated transmitted signal with a shifted scattered signal. 1 In actuality, MRR also includes a decimation step.