Research On Narrowband Interference Suppression Technologies In Spread Spectrum Communication Systems

Direct Sequence Spread Spectrum (DSSS) communication systems have been widely employed in civilian and military fields, for their unique features such as multiple accessing capability, low probability of intercept transmission and multipath rejection etc. Even though DSSS communication systems are inherently robust to the effects of narrow band interference, they have been shown that the use of additional processing aimed at interference suppression can result in substantial performance improvement. In the past thirty years,many researches are focused on stationary narrowand interference suppression techniques in DSSS communication systems. In this dissertation ,not only the stationary narrowband interference suppression technologies, but also the cyclostationary narrowband interference suppression technologies were studied. The improvement on algorithms and narrowband interference suppression technologies was performed to make them feasible for practical realization.A simplified signal model of receiver was given in this dissertation. Then through analyzing the statistical character of interference, this dissertation built up the stationary model and cyclostationary model of interference.Based on stationary model of interference, first, the suppression performances of different time domain interference suppression techniques were summed up and compared. Second, a kind of blind adaptive algorithm of minimum bit error ratio (MBER) receiver was discussed., and its stability performance was discussed. Finally, a different approach of time domain interference suppression was studied by the eigenanalysis of the receiving data. An eigencanceler based on predictor/interpolation filter was simplified; it reduces computational complexity greatly. A new MBER eigencanceler was deduced and its suppression performance was analyzed. A subspace tracking method-Robust Recursive Least Square (RRLS) algorithm implements these two eigencancelers. Aiming at the shortcoming of the RRLS algorithm, an improved algorithm was proposed. It added the steps of eigenvector orthogonalization,rank estimating adaptively and diagonal loading technique. The improved algorithm become more stable and its estimation ranks more precisely,when interference is strong.Based on cyclostationary model of interference, the dissertation gived the structure of the linear frequency shift (FRESH) predictor/interpolation filter and deduced their output SINR. When there was a cyclic frequency error (CFE) of the desired signal, The FRESH filter geted severe performance degradation. A recursive algorithm of cyclic frequency was proposed. It makes the estimative value to converge to the true one gradually. So it can largely depress the sensitivity of frequency shift filter to CFE. On cyclostationary property of interference, a suppressor optimized in the MBER criterion was studied. It was changed into the optimum suppressor in the MMSE criterion and the constrained MMOE criterion. These two optimum suppressors could be completed by two RLS algorithm. This dissertation called them Cyclo (cyclostationary-based)-RLS and Cyclo-BRLS algorithm respectively. These two adaptive algorithms had many problems. The IQR decomposition was used to overcome them. The systolic arrays of IQR decomposition for parallel implementation of these adaptive algorithms were designed.The Principle of reduce rank based on multistage wiener filter (MWF) had been analyzed. Then a MWF based on Lanczos method was simplified. Its computational cost is less than before. Based on cyclostationary model, a MWF is used for interference suppression ,which has low computational complexity and good performance.It had been proved that the Krylov subspace equal the interference subspace on some conditions. The method which was used to get orthogonal vector of Krylov subspace could be used to get interference subspace. So a new eigecanceler based on Krylov subspace was proposed. Using PN code sufficiently, a new algorithm which combined the forward recursive steps of MWF with RLS algorithm was proposed. It has lower computational complexity and better suppression performance than RLS algorithm.