Research On Multi-sensor Information Sensing Method Based On Compressed Sensing

Along with the increasing scale and perception types of Wireless Sensor Networks(WSNs),energy consumption and sampling efficiency has been deemed as the key problems to be solved up to now.It is necessary to collect lots of precious data during perception,like multimedia information such as environmental information,pictures and videos of surrounding area.High efficiency cannot be realized according to the present technology.Furthermore,it is still difficult to collect high-end signals.Therefore,it is of significant research meaning and practical application value in efficiently perceiving multiple information during using on the WSNs node with limited resource and energy.As a mew sensing mode,Compressed Sensing(CS)could help compress the obtained signals while perceiving measurement.Besides,it also manifests huge potential in many application fields such as radar,image and code correction.Analog to Information Converter(AIC),a new-type signal acquisition scheme based on CS theory,brings about great challenge to analog-digital converter sampling methods based on traditional Shannon /Nyquist to a great extent in recent years.Through conclusion of existing sampling models,the advantages & disadvantages of sampling structure that is applied into multiple sensors and the problems existing during existing perception information methods are analyzed in the paper.Firstly,we studied a single environment-aware node and built a new AIC system and corresponding hardware modules.Subsequently,according to WSNs of multiple sensing data sources,sensing theory of AIC is mainly studied.Through in-depth learning of AIC theory characteristic analysis on time and frequency domain of multisource signal,new sampling scheme is formulated by combining the AIC theory with sampling structure of time/frequency division multiplexing.According to the small number of sensor nodes and signal with high discreteness,time division multiplexing is adopted when compressed sampling.Besides,the observation matrix with time division/frequency perception characteristic based on stochastic demodulation is built.According to the experimental result,it shows that: the problems are effectively solved through the scheme.Besides,the hardware cost of WSNs is also effectively reduced.As for the signal with obvious frequency division in node,the frequency division multiplexing is adopted to perceive data and reduce data size transmitted with the multi-sensor.