Research On Interference Suppression Algorithms Via Sparse Representation For GNSS

Global navigation satellite system(GNSS) can provide real-time, all-weather and global services. However, the satellite navigation accuracy and integrity are extremely vulnerable to radio frequency interferences because the power of received satellite signal is far lower than the power of noise. Thus, anti-jamming has become a major concern in the GNSS community. In the existing anti-jamming technology, as a core technology of satellite navigation terminal, interference suppression based on array processing has become the focus of all countries in the world. In addition, compressed sensing theory regarding sparse signal acquisition and reconstruction has become a new hot topic in signal processing. In this thesis, sparse array model is developed by utilizing sparse characteristics of interference spatial spectrum. In order to overcome the shortcoming of traditional interference suppression algorithms, sparse recovery algorithms are applied to anti-jamming.Firstly, some fundamental concepts of array signal processing and sparse representation are introduced. Then the fitness of anti-jamming based on sparse representation is discussed in order to establish the theoretical basis for follow-up work. Secondly, the conventional interference suppression algorithms based on covariance matrix are ineffective in high dynamic environment. In view of the above-mentioned fact, the greedy algorithm is applied in high dynamic interference suppression by utilizing sparse characteristics of interference spatial spectrum. Estimation of the covariance matrix is avoided by using few snapshots, even only one snapshot. Besides, traditional greedy algorithms require a priori information of the number of interferences. For this reason, this paper proposes improved greedy algorithms. Finally, the power minimization approach can not provide flat gains in other directions. In consideration of this situation, a new interference suppression algorithm is proposed by combining with eigenvalue thresholding method and L1-norm constraint. New algorithm forms automatically deep nulls in the direction of arrival of interferences and provides approximately flat gains in other directions. The simulations demonstrate the effectiveness of all methods in this paper.