Analytical Investigation of Non-Coherent Mutual FMCW Radar Interference

Resource-Efficient Deep Neural Networks for Automotive Radar Interference Mitigation

“…One such burst is illustrated in Figure 4, where an ideal h[n] = δ[n] is assumed with δ[n] being a Kronecker delta. Classical interference mitigation methods are mostly signal processing algorithms that are applied either on the time domain signal s IF [n, m] or on the frequency domain signal S R [n, m] after the first DFT [1]. The CNN-based method used in this paper, also termed Range-Doppler Denoising (RDD), is applied on the RD map after the second DFT.…”

“…The radar signal contains reflections from static and moving objects as well as receiver noise. The simulated interference, that is added to the time domain measurement signal according to (1), is generated by sampling uniformly from the ego radar and interferer radar transmit parameters. See [35] for a detailed description of the simulation parameters and [11], [36] for an extensive analysis of the used measurement signals.…”

“…For a non-regulated spectrum, the most common form of mutual interference is non-coherent, where the transmitters send with non-identical parameters. This results in burstlike interferences in time domain and a decreased detection sensitivity in the range-Doppler (RD) map [1], [2]. Thus, the detection and mitigation of interference is crucial in a safety context and must be addressed.…”

Decentralized Automotive Radar Spectrum Allocation to Avoid Mutual Interference Using Reinforcement Learning

“…Frequency modulated continuous wave (FMCW) is widely employed in automotive radar due to its low hardware complexity [2]. The consequences of mutual interference among FMCW radars have been investigated in [1,[3][4][5][6]. The probability of ghost targets is raised if an FMCW radar is interfered by another with the same chirp rate [1,3,4].…”

“…To prevent ghost targets, [5] proposed to use random chirp rates so that interference resulted in an elevation of the noise level. Such interference is referred to as non-coherent interference in [6]. The MOSARIM project also concluded that the most probable P. Liu, Y. Liu, T. Huang and X. Wang are with the Department of Electronic Engineering, Tsinghua University, Beijing, China (e-mail: liupf16@mails.tsinghua.edu.cn; yiminliu@tsinghua.edu.cn; huangtianyao@tsinghua.edu.cn; wangxq ee@tsinghua.edu.cn).…”