Research And Simulation Of Background Noise Cancellation To Mobile Phones

The wide use of cellular phones has significantly increased the use ofcommunication systems in high noise environments. Intense background noise,however, often corrupts speech and degrading the performance of manycommunication systems. So noise reduction becomes increasingly important toimprove voice quality in noisy environments, especially for the future personalmobile communication, which serves people"Anytime, Anywhere, Always BestConnected". From the birth of the first handsets in 1973, the theory ofbackground noise cancellation and its applications has been always attractingmost of correlative manufacturers and graduate schools'interest. Recently, thealgorithms and products which can suppress or remove the background noisehave been announced continually overseas. For a purpose of expanding theinfluence and applications of this technology interiorly, the research onbackground noise cancellation to mobile phones was chosen as my graduatethesis with an expection to throw a sprat to catch a whale.Usually the background noise does not keep steady and it will change fromtime to time. So Adaptive Noise Cancellation is used, it could be able to workunder changing conditions, and be able to adjust itself according to the changingenvironment. To realize the adaptive noise cancellation, two distinct parts must beconsidered, a digital filter to perform the desired signal processing, and anadaptive algorithm to adjust the coefficients (or weights) of that filter. Becausethe FIR filter is always stable and can provide a linear phase response, it becomesthe choice of the design. The basic adaptive noise cancellation algorithm is LMSAlgorithm, which is the earliest and the most widely used adaptive algorithm forpractical applications for its simpleness. Also, other modified algorithms based onLMS algorithms will be presented.From above descriptions, an outline of ANC to mobile phones is obtained.The next problem is how to realize it on DSK in real time. The Mathworks Inc.provided a model-based method for developing digital signal processing differentfrom traditional development technique. Under the MATLAB developmentenvironment, the following tasks have been implemented:Use multi-domain models to design an ANC systemTest and compare the performance of different LMS AlgorithmsRealize FBLMS Algorithm on DSK through RTDX methodGenerate code automaticallyDesign a SNR model to test and verify continuouslyThis thesis can be divided into seven chapters, the theory of ANC and its realization in MATLAB and on DSK is introduced at length.Chaper 1 begins with a description of the domestic and foreign status quo ofbackground noise cancellation to mobile phones, and then brings forword thebackground, purpose, significance and main contents of the research. In Chapter2, we introduce the characteristic of speech and noise, make a summary ofbackground noise suppression technique and digital signal processing methods.Chapter 3 gives a brief introduction of adaptive filter theory, especially stresses agroup of LMS algorithms. In Chapter 4, simulink for system and algorithmmodeling are introduced. In Chapter 5, Using simulink, an ANC model is createdto complete analysis, camparison and evaluation to the performance of a familyof LMS algorithms. Chaper 6 shows the steps necessary to realize FBLMSalgorithm on TMS320VC5510 DSK in real time with RTDX method, involvingmodel design, code generation, measurement and verification of signal-to-noiseratio. A brief summary and further research advice is put forward in Chapter 7.