Abstract: Aiming at the problems that device-free sensing technology had high parameter dependence in practical applications and existing low-parameter-dependent solutions based on optimization algorithms tended to fall into local minima, we proposed a low-parameter-dependent device-free sensing method based on the differential evolution algorithm. The proposed method optimized the parameter configuration of the sensing system through the differential evolution algorithm, combined iterative search and performance evaluation of multi-parameter combinations to gradually approach the global optimal solution, and introduced multiple classifiers to verify the effectiveness of the method.  The experimental results show that this method can effectively overcome the problem of parameter dependence, improve the accuracy of data collection and target detection, with a classification accuracy of 97.06% on the human behavior recognition dataset and 94.02% on the brain nerve dataset. It  has good practicability and robustness, and  effectively solves the defects of parameter dependence and local optimal solutions of traditional sensing methods, providing a new  parameter optimization idea for the engineering application of device-free sensing technology.

Key words: device-free sensing, differential evolution algorithm, low parameter-dependent, optimization algorithm