Research On Narrow-Band Interference Detection And Suppression For Synchronous CDMA Systems

Due to NBI (Narrow-Band Interference) with high power makes spread spectrum communication invalid, the demand of optimal interference rejection with different NBIs for synchronous CDMA (Code-Division Multiple-Access) systems was investigated in this dissertation, whose major reseach issues were how to recognise the type of NBIs and how to select the optimal techniques for different NBIs.To solve the problem that traditional recognition techniques could only detect partial common interferences for CDMA systems, a TSMD (Transform Selective Mode Distinguish) recognition technique was proposed. The TSMD technique could recognise impulse interference, linear frequency modulation interfereence and NBI based on CG (Compression Gain) performance in different fields. More, recognise the type of sinusoidal tones, digital NBI or AR (Autoregressive) stochastic process based on mean and covariance of NBI'mode distinguish performance.In the case of signal's statistical properties known, to solve the problem of high computational complexity with existent DMI (Direct Matrix Inverseve) code-aided technique, a subspace code-aided technique was proposed for sinusoidal tones and digital NBI. This technique could decrease computational complexity effectively with improving the distributing of matrix. And for AR stochastic process, in order to solve the problem of difficile determination on low order with above subspace technique, an improved linear prediction & code-aided technique based on subspace was proposed to achieve nice order estimating and low complexity with whiting the distributing of subspace'eigenvalue.Besides, in the case of information such as train sequence known, to solve the problem of slow constringency with existent DMI LMS (Least Mean Square) code-aided technique by high power NBI and high complexity of existent DMI RLS (Recursive Least Square) technique, an improved adaptive RLS prediction & LMS code-aided arithmetic was proposed. The improved arithmetic could constringe rapidly, achieve steady performance closing to RLS with low complexity through reducing the power of NBI after prediction module.Furthermore, in the case of signal's statistical properties and train sequence unknown, to solve the problem that the detection rate of blind code-aided technique based on DMI was not good, a blind code-aided technique based on subspace was proposed. This technique achieves high performance through improving the SINR (Signal to Interference and Noise Rate) performance.Then, the realization of linear blind adaptive filter was analyzed to decrease the complexity of batch blind code-aided techniques above. To solve the problem of bad steady performance with blind adaptive LMS code-aided technique by high maladjustment, conflict constringency performance and steady performance with blind adaptive RLS code-aided technique, an improved blind adaptive RLS prediction & LMS code-aided arithmetic was proposed. The improved arithmetic could achive high steady performance with rapidly constringency through reducing the power of NBI after prediction module. Besides, a blind PASTd (Projection Approximation Subspace Tracking with deflation) code-aided arithmetic which was better than blind adaptive DMI RLS code-aide technique was proposed for sinusoidal tones and digital NBI. And for AR stochastic processes, to solve the problem of difficile determination on low order with above PASTd technique, an improved blind adaptive RLS Prediction & PASTd code-aided arithmetic was proposed. For whiting subspace' eigenvalue distributing, the improved technique with low complexity has better performance than bind adaptive RLS technique.Finally, this dissertation points out some open problems and future research areas.