The Design And Research Of Adaptive Filter In Eliminating Mixed-noise Interference

Speech signal in the process of receiving and transmitting will always besubjected to interference from outside world and the natural environment in differentdegrees, which make actual information affected, so how to filter out the noise andinterference has been the goal of the field of digital signal processing. The traditionalway of processing need to calculate the statistical knowledge of signal and noise, butthese basic conditions are difficult to obtain in an unknown environment. Adaptivefilter provides an attractive filtering method, it can be based on certain predeterminediterative algorithm, following environmental change automatically adjust itself filtercoefficients and eventually realize filter function. In addition, real communicationenvironment are very complex, we are often exposed to a variety of noise, in whichthe periodic interference and broadband Gaussian white noise are more common, so itis very necessary to study an adaptive filter which can simultaneously removemixed-noise.This paper proceeds from the significance of research of adaptive filter ineliminating mixed-noise interference, it describes the structure of adaptive filter, andelaborates the working principle through detailed theoretical derivation and analysis,and summarizes several commonly used adaptive filtering algorithms. LMS algorithmdeveloped on the basis of the method of steepest descent, whose performance likeconvergence, convergence rate, offset and computational complexity has a closerelationship with the step size factor and filter order, due to its simple iterative process,a small amount of calculation, easy to implement, and widely used features, thispaper design uses LMS algorithm.Gaussian white noise canceller and periodic interference splitter is the basic partof the design of, the core principles of them are all the LMS adaptive filter, Gaussiannoise canceller is to offset the reference signal that estimated by the filter from the Gaussian white noise pollution, periodic interference splitter is to add delay unitbefore the original input signal, which makes the periodic signal and delayedbroadband signal irrelevant, so that obtain a valid signal.The design of adaptive filter in eliminating mixed-noise interference based onLMS algorithm and has two stages, the first stage is two parallel working periodicinterference separator in order to separate periodic interference from received signal,the second stage is only a Gaussian white noise canceller whose function is tocontinue to filter out the white noise from information and output the best signal. Thetwo-stage adaptive filter has been simulated using SIMULINK block library. Thesimulation results are shown that: the proposed filter can effectively filter outmixed-noise including sinusoidal narrow-band interference and Gaussian white noise,and meet the design requirement.System Generator launched by Xilinx Inc. is a high-level development tool indigital signal processing, it can complete FPGA design process such as system modelestablish, simulation, implement, verification and configuration download on a sameplatform, the speed is accelerated by this feature. This article uses this innovative wayto successfully build a hardware model of adaptive mixed-noise canceller based onSystem Generator, by comparing and analyzing simulation results form a number ofaspects, we verify that this hardware design can remove periodic interference andwhite Gaussian noise simultaneously, and the filtering effect is consistent withoriginal algorithm design, this design can be successfully implemented on FPGA.