Study On Narrowband Interference Suppression Technology In DSSS Based On Time-Domain Nonlinear Processing

Spread spectrum communication techniques have many advantageous properties, such as inherent capability to reject interfering signals, privacy, increasing transmission capacity, and sharing bandwidth with other narrowband communication systems, which have been widely used in military communication system, navigation and localization system, personal wireless communication etc. Because equipments and systems are often operated in complex electromagnetic surroundings, and faced with more and more serious interference, the external interference, especially the strong narrowband interferences (NBIs), cannot be rejected efficiently only by the spread spectrum (SS) gain. Some other measurements of rejecting NBIs should be applied to guarantee the performance of the SS communication systems.The technology of the time-domain prediction filtering can approach the optimal solution when the observation noise is Gaussian, which has been widely researched. Since the direct-sequence spread spectrum (DSSS) signal can be modeled as an independent identically distributed (i.i.d.) binary sequence, which is highly non-gaussian, the optimal scheme of NBI suppression at this condition is nonlinear prediction filtering. The researches of the existing time-domain nonlinear prediction filtering techniques against NBIs mainly focus on adaptive algorithms. Seldom researches focus on filters' structure design. The orders and the complexity of existing prediction filters are too high to acceptable for practical applications under the complex electromagnetic surroundings. The thesis is performed mainly on the design of the prediction filters' structures. The main results of this paper are as follows:1. The structures and the adaptive algorithms of existing prediction filtering have been systematically studied and the inherent problem of them has been analyzed.2. The characteristics of laguerre filters have been studied in theory and some institutive conclusions have been deduced. A novel adaptive nonlinear laguerre transversal prediction filter and three kinds of improved nonlinear adaptive algorithms based on it are proposed in this paper. The memory depth of the filter is not coupled to the order of the filter and the realizing complexity of the filter can be largely reduced. The three kinds of improved algorithms are as follows:1) The adaptive nonlinear laguerre normalized LMS algorithm. This proposed algorithm outperforms the other exiting nonlinear adaptive prediction schemes for the NBI suppression in DSSS, and the orders of the proposed filter can largely be reduced.2) Variable step size adaptive nonlinear laguerre normalized LMS algorithm. Theconverging stability can be enhanced and the converging speed can be accelerated by using this algorithm, and the tremble to update the tap weight proceeded from the irrelated noise can be effectively reduced.3) Evolutionary programming algorithm is applied for the proposed filter structure and an adaptive nonlinear laguerre evolutionary programming (L-EP) algorithm is proposed, signal-to-noise (SNR) improvement performance of which performs substantially better than that of common signal processing methods.3. The traditional adaptive nonlinear lattice prediction filters all have the problem that the filter order couples with the memory depth, which restricts the performance of suppressing NBIs. An improved adaptive nonlinear laguerre lattice prediction filter is proposed and an adaptive nonlinear laguerre lattice algorithm is given. The proposed lattice prediction filter has many advantages, such as low complexity of realization, parallel process with high speed, insensitive to round off error of finite word length etc.