An Efficient Compression and Reconstruction Framework for Electromagnetic Spectrum Data

Spectrum monitoring often demands a great quantity of spectrum data, and the massive characteristics of spectrum data make it consume a lot of resources in the process of transmission and storage. At the same time, the compressed acquisition system of spectrum data often has the problem of low reconstruction accuracy of the original data, and the reconstruction accuracy and compression performance cannot be achieved simultaneously. This paper studies the factors affecting the reconstruction error in the process of electromagnetic spectrum data utilization. In this paper, an electromagnetic spectrum compression and reconstruction framework called QRK-SVD is proposed. Aiming at the problems of slow dictionary convergence and low accuracy in dictionary learning, QRK-SVD purposely uses k-means clustering to construct the initial dictionary, which effectively improves the compression accuracy and system robustness. QRK-SVD increases the minimum singular value of sensing matrix through QR decomposition to optimize the problem of low accuracy of random observation matrix in compressed system. We designed a set of spectrum data acquisition and compression system based on QRK-SVD. It can adapt to various collection scenarios, greatly reduce the amount of data transmitted and stored, and has high reconstruction accuracy. The measured data proves that the performance of QRK-SVD is better than the traditional K-SVD framework in different data compression situations.