The Research Of Compressive Sensing Reconstruction Algorithm In IOT

The Internet of Things (IOT), as the extension of Internet, collects all the information, via sensing devices and becomes a huge network with Internet. As one of the most important applications scenarios, sensor network has been most potential market. The world of big data causes nodes’ capability of computing, storage, endurance not meet the demand, that becomes the barrier of development of IOT.In recent years, compressive sensing (CS), which is a new type of sampling theory and an innovation, has attracted all fields of science and engineering’ attention. Compressive sensing surpasses the traditional limits of sampling theory, using sparsity of signals. The most of information in distributed network will be filter. Because of this special type of network of nodes in IOT and data acquisition, if the redundant information will be missed at the beginning, that will make IOT rapid progress.This paper has deep research about the application of compressive sensing in information transmission of IOT and focuses discussion on the application of distributed compressive sensing in video service of IOT. Combined with Bayesian theory, the main fundamental work and innovation ideas are as followed.After a brief historical overview, we detailed the key of traditional compressive sensing reconstruction algorithm and make simulations about performance of algorithms.Also, we analyze the application of distributed compressive sensing in video/image for the spread of IOT in real-time monitoring. Based on the reconstruction algorithm we made, make comparisons of performance of these algorithms.This paper considers Bayesian theory in compressive sensing, Bayesian compressive sensing algorithm utilizes the property that different signals satisfy the same priori probability distribution, to jointly estimate the probability distribution parameters of original signals. In order to solve the limitation of traditional distributed Bayesian compressive sensing, we propose the robust the Laplace prior probability based distributed Bayesian compressive sensing algorithm, which is more suitable for complex noise model.