Gesture Recognition System Resilient to Interdevice Interference Based on Direct Sequence Spread Spectrum

Gesture interaction based on ultrasonic radar is an attractive solution for wearable devices. However, interdevice interference will be a realistic problem, when multiple adjacent devices that use this kind of technology are working simultaneously. In this letter, a direct sequence spread spectrum (DSSS)-based gesture recognition system that is resilient to interdevice interference is proposed. Based on the idea of code division multiplexing, the interference of other transmitters can be suppressed through correlation processing. The range-Doppler map (RDM) is generated by fast Fourier transform on the correlation function, and then features are extracted from the RDM sequence. A random forest classification is employed to complete the gesture classification. A complete prototype DSSS gesture recognition system has been built, which consists of 40-kHz ultrasonic frontends, an field-programmable gate array (FPGA) development board, and a personal computer. It is demonstrated that the system can achieve an average accuracy up to 92% for four defined gestures while the accuracy with an interference transmitter is lowered to 88%, which is 16% higher than that of the classic scheme that does not use the DSSS technique.