Font Size: a A A

Research On Sub - Nyquist Sampling And Reconstruction Algorithm For Sparse Multi - Band Signals

Posted on:2016-09-09Degree:MasterType:Thesis
Country:ChinaCandidate:Z HuangFull Text:PDF
GTID:2208330470955395Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
The highly information-oriented era sees the explosive growth of the amount of various data. In signal and information processing, for the huge amount of data, there are two development directions, one goes toward the big data processing theories and algorithms to extract useful information from a huge amount of data; another uses compressive sampling methods to reduce the total amount of data while preserving the useful data for a large class of sparse signals, such as many signals with sparse characteristics in radio astronomy. Sub-Nyquist sampling is a compressive. sampling method based on the traditional Nyquist sampling theory. In this paper, the extensive analysis of Sub-Nyquist sampling and recovery methods is performed for multi-band signal sampling and recovery.Sub-Nyquist sampling does not breach the traditional Shannon-Nyquist theorem and it is an innovation based on the Shannon-Nyquist theorem. Sampling data can be employed to accurately reconstruct the original signal using reconstruction algorithms. Sub-Nyquist sampling includes the sampling algorithms and reconstruction algorithms. The purpose of compressive sampling is that as less data as possible are weighted to represent the original signal. The reconstruction algorithms are to identify these small data by using optimal query.In this paper, three common Sub-Nyquist sampling frameworks, i.e., random demodulator (RD), multi-coset sampling (MC sampling) and modulated wideband converter (MWC) are first introduced; This paper focuses on the modulated wideband converter; especially, the three common reconstruction algorithms-OMP, ROMP, and STOMP are analyzed. Then an improved reconstruction algorithm called normalized OMP is proposed based on these algorithms. Normalized-OMP introduces a factor vector which is the main diagonal elements of the matrix obtained by the compressive matrix multiplied by its transpose matrix. The improved parts of the algorithm include three steps:(1) Multiply the transpose of the compressive matrix by residuals;(2) Obtain a vector by summing each line of the resulting matrix;(3) Divide the column vector by the factor vector, then find the position corresponding to the largest element; The rest part of the steps are similar to the OMP algorithm. In this paper, MWC is used to sample the sparse multiband signal. OMP, ROMP, STOMP and Normalized-OMP are employed to reconstruct the signal, respectively. From the analysis of the experimental data, we can see that Normalized-OMP has advantages over the previous algorithms in running time and the signal reconstruction error. Experiments show that the improved reconstruction algorithm is effective, and it sets a foundation for radio astronomy signal collection and recovery by using MWC.
Keywords/Search Tags:Compressive sensing, Sparse multiband signal, Compressive sampling, Sub-Nyquist sampling, Reconstruction algorithms
PDF Full Text Request
Related items