Two-Step Pairing Algorithm for Target Range and Velocity Detection in FMCW Automotive Radar

Eugin HYUN  Woojin OH  Jong-Hun LEE  

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E98-A   No.3   pp.801-810
Publication Date: 2015/03/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E98.A.801
Type of Manuscript: PAPER
Category: Digital Signal Processing
FMCW radar,  pairing,  multi-target,  two-step,  detection algorithm,  detection probability,  

Full Text: PDF>>
Buy this Article

In automotive frequency modulated continuous wave (FMCW) radar based on multiple ramps with different slope, an effective pairing algorithm is required to simultaneously detect the target range and velocity. That is, as finding beat-frequencies intersecting at a single point of the range-Doppler map, we extract the range and velocity of a target. Unlike the ideal case, however, in a real radar system, even though multiple beat frequencies are originated from the same target, these beat frequencies have many different intersection values, resulting in mismatch pairing during the pairing step. Moreover, this problem also reduces the detection accuracy and the radar detection performance. In this study, we found that mismatch pairing is caused by the round-off errors of the range-beat frequency and Doppler frequency, as well as their various combinations in the discrete frequency domain. We also investigated the effect of mismatch pairing on detection performance, and proposed a new approach to minimize this problem. First, we propose integer and half-integer frequency position-based pairing method during extraction of the range and Doppler frequencies in each ramp to increase detection accuracy. Second, we propose a window-based pairing method to identify the same target from range-Doppler frequencies extracted in the first step. We also find the appropriate window size to overcome pairing mismatch. Finally, we propose the method to obtain a higher accuracy of range and velocity by weighting the values determined in one window. To verify the detection performance of the proposed method by comparison with the typical method, simulations were conducted. Then, in a real field test using the developed radar prototype, the detection probability of the proposed algorithm showed more than 60% improvement in comparison with the conventional method.