The Research Of Weak Life Signal Detection Based On FPGA

With a vast territory, China has natural disasters attacking frequently. Earthquakesand landslides have been threatening people’s lives and property. Combined with moderntechnology and a variety of methods, post-disaster search and rescue can locate thevictims quickly and accurately then give rescue. Non-contact radar life detectiontechnology detects living body with radar without contact through unknown regions,depending on principles of Doppler Frequency Shift, etc. But life signals are weak andrandom with a low frequency range and strong noise, weak signal extraction is particularlyimportant in the life detection technology.Some problems of signal processing in life-detection systems has been studied, aFPGA-based life signal processing module has been designed, the wavelet transform insignal processing application has been discussed, focusing on multi-resolution analysis ofwavelet transform. Virtual instruments has been used as experimental platform for analysisof life signal characters to extract different frequency band signals of life. Experimentaldata has shown that life signal exists0.2-2Hz harmonic component of breathing andheartbeat, and2-15Hz harmonic component of the body movements. Through differentlayers of wavelet decomposition, different frequency band signals of life has beeneffectively extracted. Our group adopts the second generation wavelet lifting scheme todesign wavelet decomposition and reconstruction modules on the basis of FPGA, and usepipelining to optimize the structure. An independent data register module has beendesigned to isolate the wavelet transform data I/O and decomposition steps, and improvethe independence of each module.Additional registers isolate forecast and update steps,achieve pipeline architecture optimization to improve their data processing capabilities.FPGA-based wavelet transform module completes the real-time signal processing inlife-detection system. It can be seen from the probe data, body movements signal is more easily detected due to amplitude and high frequency. It can quickly and accurately identifyhaving life or not. Breathing and heartbeat signals is weak, and easily drowned inbackground noise. After further wavelet decomposition to extract less than2Hz signals,the peaks in the frequency domain exist, which are the heartbeat and breathing signalssuperimposed.Experimental results show that the designed wavelet transform signal processingmodules can effectively extract weak life signal from real-time radar echo signal, meetingthe demand of signal processing in life-detection system. Finally, to further improvesystem performance, the adaptive decomposition level and threshold settings can befurther studied.