臺灣博碩士論文加值系統

展頻通訊有許多優點，包括抑制干擾、提供多使用者使用同一通道、消除多路徑干擾、低攔截機率等，在這眾多優點中排拒干擾的能力是相當重要的部份，此能力主要取決於系統的處理增益(processing gain)。因為實際系統的增益有限，故若干擾的強度大到超過其能力，則系統無法正常工作，此時就需利用一些訊號處理的技術將干擾減弱以利後端處理。本研究的目的便是針對直接序列展頻通訊系統的各種窄頻干擾，包括單一弦波干擾、單一自動回歸模式干擾及多重窄頻干擾等，找出其有效的加強抑制窄頻干擾能力的方法。 本論文主要是以線性與非線性最小均方(least mean square, LMS)自適性濾波器來抑制各種不同頻帶、不同種類的窄頻干擾。結果顯示非線性最小均方自適性濾波器在單一弦波干擾、單一自動回歸模式干擾下的訊雜比改善值較優於線性最小均方自適性濾波器，但非線性濾波器的收斂速率較線性慢且有較差的收斂特性。另外此二種自適性濾波器在對抗多重窄頻干擾時都有一定的限制，故本論文也針對頻帶接近的多重干擾源提出以小波包(wavelet packet)分析的方式來克服。 本論文最後提出一個時頻分析的架構，以自適性時頻分解的方式來抑制直接序列展頻通訊中的窄頻干擾。此架構是利用干擾源的不同特性來決定在時域中或在頻域中抑制干擾，並且可克服使用自適性濾波器在無干擾源時會追蹤微弱相關的展頻訊號之缺點，因此極適合處理穩態或非穩態的干擾。

Spread spectrum communication systems have many advantages, including interference rejection, multiple accessing, multipath suppression, and low probability of intercept. Among all the potential applications, the ability for a spread spectrum system to withstand interferences, both intentional and unintentional, is fairly important. Any practical spread spectrum receiver can only suppress a given amount of interference. If the interference level exceeds some threshold, the system will not function properly. In this case, the interference immunity can be improved significantly by using signal processing techniques which complement the spread spectrum modulation. The goal of this study is to find an effective method to reject various narrowband interferences in direct-sequence spread spectrum systems, including a single-tone interference, an AR model interference, and multiple narrowband interferences. This thesis is aimed at both linear and nonlinear least mean squares adaptive algorithms to reject the narrowband interferences. The simulation results show that the nonlinear LMS filter performs better than linear one in terms of the SNR improvement for both AR model and single-tone interferers. However, the former has slower convergence rate and worse convergence behavior than the latter. Furthermore, both adaptive filters have some limitation on multiple-interference rejection. Therefore, this thesis also presents a wavelet packet analysis to reject multiple interferences located in close frequency bands. Finally, this thesis proposes a time-frequency framework to enhance the interference rejection capability in a direct-sequence spread spectrum system. The principle of the framework is to detect both time and frequence localization characteristics of the interferences before an appropriate time or transform domain processing technique is performed. Incidently, it can overcome a drawback inherented in an adaptive filter to track a weakly correlated spread spectrum signal when there is no interference. This scheme is thus well suited for handing stationary as well as nonstationary interferences.