Research On AIC Information Demodulation Technique Based On Sparse Decomposition

With the rapid development of electronic information technology, the bandwidth of the signalbecomes wider. If we still use conventional Nyquist sampling theorem to achieve signal informationacquisition，it will bring a big challenge for hardware implementation. However, CompressSampling theory--a new signal acquisition theory--has risen in recent years at home and abroad.The theory makes the use of the sparsity of original signal on a transform basis and an observationmatrix which is unrelated with the transform basis, gets a small amount of information samples byprojecting the original signal onto a low dimensional space with the frequency which is far belowthe Nyquist frequency, and the original signal can be reconstructed by optimal sparse reconstructionalgorithm. Comparing to the traditional signal acquisition and processing, when the signal issampled, the data also compressed. In other words, it combined with the data acquisition andcompression. This paper discusses the theory of Analog-to-Digital Converter technology which isbased on compressive sampling, introduces the main idea of the AIC architecture, analyses andprocesses the information data which is obtained by AIC, and further studies the problem of thesparse frequency hopping signal demodulation based on AIC. The main contents and innovation ofthe paper is as follows:First, a brief description of the basics of compressive sampling theory is presented, and studiesthese parts: the sparse decomposition, measurement matrix design and reconstruction algorithms.We analyzed three analog-to-information structure--pseudo random modulating AIC structure,parallel AIC structure, sectional integration AIC structure and numerical simulations were carriedout, comparing the different effect on the signal sampling structure reconstruction.Second, the signal reconstruction algorithm based on compressive sampling of AICinformation is studied, the model and procedures of convex relaxation algorithm and greedyalgorithm is analyzed and compared with each other by simulation experiments. Then, combiningwith the advantages of signal rapid reconstruction of regularization orthogonal matching pursuitalgorithm and the sparsity self-adaptation of sparsity adaptive matching pursuit algorithm, animproved sparsity adaptive variable step regularized matching pursuit algorithm is proposed.Simulation result demonstrates the feasibility of this method.Three, the interference suppression of AIC information is analyzed. For the fixed-frequencyinterference mixed in hopping AIC information, three methods to suppress are compared:Least-mean-square method, Annihilating filter method and Controllable information domain trapmethod. Simulation result shows the hopping information can be well selected from the mixed AIC information by the three fixed-frequency suppression methods.Four, we study the frequency-hopping signal demodulation technique based on AICinformation. A hopping demodulation model which is based on1-sparse model is constructed, andthe symbol hopping time is estimated by maximum likelihood method, which ensure thesynchronization of signal sampling and symbol changing, and reduce the impact on the compressionsampling signal at symbol hopping time. Meanwhile, a hopping signal demodulation method basedon the sparse decomposition matrix is proposed, which makes the use of constructed sparsedecomposition matrix to estimate the hopping time and frequency hopping parameters first, andthen demodulated signal with the estimated parameters. These two methods both use the data ofinformation field to estimate the demodulation parameters, greatly reducing the amount ofcomputation. Finally, simulation experiments verify the correctness and rationality of the theory.