Adaptive Filter-based Noise Suppression Technique

All systems will be influenced by noise. How to effectively eliminate the noise is one of hot subjects for years. Noise suppression is classified into two classes: Passive Noise Control and Active Noise Control. Passive Noise Control suppresses noise by sound-absorbing material and sound-absorbing construction. This method acts on high frequency noise suppression, but not on low frequency noise, and it is difficult to handle because of weight and bulk. Active Noise Control acts on low frequency noise, and the system is small in size and light in weight. With the development of control system theory and digital signal processing Active Noise Control puts concentration on adaptation. The paper studied the Active Noise Control and its application method based on adaptive filter approach.The paper begins with the principle of adaptive filter. Based on the principle, Least Mean Square and Least Squares are researched deeply. They are two important algorithms of adaptive filter. The ratiodnative process and convergence performance of the algorithms is given. Parameter effects on performance of the algorithm are studied. Based on the MATLAB platform, simulation is carried out for the applications of adaptive algorithms in noise canceling, analysis of convergence rate and steady state error results are given under various conditions. Accordingly, merits of the different algorithms are discussed.On the basis of theoretical investigation and simulation, high performance implementations of adaptive filter are achieved in the paper. There are two ways: DSPs and FPGA implantation. In the DSP implementation, a type of DSP processors is used. It is TMS320C33 processor produced by TI Corp. Peripheral interface drcuit is designed. C language is used for program. The method has advantage on rate and performance of convergence.In the FPGA implementation, the method of top to down is considered. Adaptive filter is divided into several small modules. On the one hand the modules are prone to design; on the other hand they can be used by other projects. On the QUARSTUS Π platform, digital filters are described and designed by use of VHDL. The hardware design and logic synthesize of the adaptive filter is completed. Simulation is done. TheFPGA implimentation is preferable to DSPs in speed because of working in parallel mode. The work that needs a great deal of operation instruction by DSP processor to complete will only take only a time of clock period for the FPGA to complete. By use of the In-System Programmability technology, the user can amend the hardware of the system in real time. So the design of this system is vivid, easy to maintain and update.