An Adaptive Compressed Sensing Signal Restoration Method Based on Greedy Algorithm

Abstract

The invention discloses an adaptive compressed sensing signal recovery method based on a greedy algorithm. On the basis of the OMP algorithm, a detector is designed to judge whether there is still a signal component in the remaining vector, and through the hypothesis testing model in the signal detection theory, Calculate the detector, and then give the threshold of the detector through the preset false alarm probability. When the detector is smaller than the given threshold, it can be judged that there is no signal component in the remaining vector. At this time, the algorithm can stop iterating, and Get the restored value of the signal, otherwise continue to iterate. The present invention uses the greedy algorithm as channel estimation, does not depend on the multipath number of the channel; Compared with MDL, the method of the present invention does not need multiple observations, thereby saving communication resources; The method of the present invention has less computing power of the greedy algorithm The advantage is that the greedy algorithm solves the problem of high real-time requirements for sparse signal recovery, and its application in the problem of unknown signal sparsity.

Description

technical field [0001] The invention belongs to the field of signal processing, in particular to an adaptive compressed sensing method. Background technique [0002] Compressed sensing is a signal processing method for dealing with sparse vectors, and it is one of the most important theories proposed in this century. Compressed sensing technology can recover sparse signals from fewer sample points than the Nyquist sampling rate, ie y=Ax+e, where, is the observation vector, through the observation matrix for a k-sparse signal For observation, the dimension of the observation vector y is much smaller than the dimension of the signal x, m<n, and the compressed sensing technology can recover the signal x from the observation vector y. According to the compressed sensing technology, the signal can be directly compressed when it is acquired, which saves the complexity of sampling and saves the storage space. The so-called sparse signal means that most of the positions of...

AI Technical Summary

Problems solved by technology

[0016] In practical applications, when the problem dealt with by compressive sensing technology requires high signal processing speed, it is generally inappropriate to use the convex relaxation method with a large amount of calculation.

In addition, in the actual signal processing problem, usually the sparsity of the signal is unknown, therefore, the greedy algorithm cannot be directly used to restore the sparse signal

Based on the dependence on sparsity, it is difficult for the greedy algorithm to be directly applied to problems with unknown sparsity in compressed sensing technology.

Images