Multi-person Respiration Detection Method Based on Complex Independent Vector Analysis

Frequency modulated continuous wave(FMCW) radar is widely used for non-contact vital signs detection due to its high accuracy and performance. Blind source separation (BSS) algorithms have become essential for multi-person respiration detection due to their low hardware dependence and robustness. However, traditional blind source separation algorithms only handle signals from a single range bin. Due to spectrum leakage, respiratory signals may appear in multiple range bins. Mutual interference between the respiratory signals of multiple people can lead to the failure of range bin selection. This paper proposes a complex independent vector analysis (IVA) algorithm based on FMCW radar for multi-person respiration detection. The method avoids the range bin selection problem by considering multiple range bins as multiple datasets for joint solutions. Meanwhile, the method overcomes the alignment ambiguity problem in traditional independent component analysis(ICA). The experimental results show that this multi-person respiration detection method based on FMCW radar is feasible.