A frequency-modulated continuous-wave (FMCW) radar interference mitigation technique using the interpolation of beat frequencies in the short-time Fourier transform (STFT) domain, phase matching, and reconfigurable linear prediction coefficients estimation for Coherent Processing Interval processing is proposed. The technique is noniterative and does not rely on algorithm convergence. It allows the usage of the fast Fourier transform (FFT) as the radar's beat-frequency estimation tool, for reasons such as real-time implementation, noise linearity after the FFT, and compatibility with legacy receiver architectures. Verification is done in range and in range-Doppler using radar experimental data in two ways: first by removing interferences from interference-contaminated data and second by using interference-free data as the reference data, and processing it-as if it had interferences-using the proposed technique, inverse cosine windowing and zeroing for comparison. We found that processing with the proposed technique closely matches the reference-data and outperforms the inverse cosine windowing and zeroing techniques in 2-D cross correlation, amplitude, and phase average errors and phase root-mean-square error. It is expected that the proposed technique will be operationally deployed on the TU Delft simultaneous-polarimetric PARSAX radar. Index Terms-Frequency-modulated continuous wave (FMCW), linear prediction (LP), multiple-input and multipleoutput radars, polarimetric radars, radar interference mitigation techniques. I. INTRODUCTION F REQUENCY-modulated continuous-wave (FMCW) radars might suffer from interferences from other radars operating within their vicinity, as in multiple-input and multiple-output radar networks and in automotive scenarios, or from themselves as in the case of fully polarimetric radars with dual-orthogonal signals [1], where there is a leakage between two mutually orthogonal channels (cross-channel interference). In deramp FMCW radars (stretch-processing), targets' range is deduced from beat-frequency estimation. Processing interference-contaminated beat frequencies with fast Fourier transforms (FFTs) yields poorer radar detection, due to undesired artifacts such as a noise-floor level increase in range

A reconfigurable range-Doppler processing method for FMCW radar is presented. By concatenating beat-frequency signals from more than one sweep, a continuous beat-frequency signal for the whole coherent processing interval (CPI) can be created. As a result, continuous targets' observation time is extended beyond that of a single chirp duration, leading to range resolution improvement. The created continuous beat-frequency signal can be split in the digital domain to any two-dimensional slow-time and fast-time matrices with the same number of elements as in the original signals, which offers a realization of a software-defined pulse/sweep repetition rate in range-Doppler processing. The signal concatenation is done in the short-time Fourier transform (STFT) domain, where the beat-frequency slices are extrapolated to compensate for the observation time lost in the transient region between sweeps, and then a phase correction is applied to each frequency-slice as appropriate, followed by an inverse STFT (ISTFT). The proposed technique is verified with simulation and experiments with an FMCW radar for stable and moving target scenarios. We found that the method allows for range resolution improvement without the transmission of additional bandwidth and also allows for the ability to observe different resolution granularities in parallel from one CPI. It additionally allows the decoupling of the transmitted PRF from the Doppler processing PRF, permitting the facility to observe different unambiguous Doppler velocity intervals from one CPI, without compromising on the total CPI processing gain.

Classical saw-tooth Frequency Modulated Continuous Wave (FMCW) radars experience a coupling between the maximum unambiguous Doppler-velocity interval, maximum operational range, range-resolution and processing gain. Operationally, a trade-off is often necessarily made between these parameters. In this paper, we propose a waveform and a processing method that decouples the aforementioned parameter dependencies at the price of using multiple receiver channels within the radar. The proposed method exploits the fact that beatfrequency signals have the same baseband frequency, even if the transmitted and received chirps occupy different radio frequency bands, and have different center-frequencies. We concatenate those baseband signals in the time-frequency domain to restore the range-resolution and processing gain. An overview of FMCW parameters trade-off for related waveforms and a feasibility and limitations analysis of implementing the proposed processing method are presented. The method is verified by simulations and experiments with an FMCW radar for stable, moving and extended-moving targets. We additionally have highlighted its non-idealities in the simulations and experiments. We found that the proposed method indeed alleviates the trade-off between FMCW operational parameters and allows the extension of the Doppler ambiguity interval without compromising on those parameters.

No abstract

No abstract