Noise cancellation using LMS adaptive filter and application implemented on FPGA

Normally, sound contents are compound of various sources that is processed at human ears. Such a condition, ear tries to gather valuable information among others which may resulted in failure. In the case of failure, we conclude that noise is at harmful degree. In addition, when noise comes from different sources with changing contents like frequency and amplitudes with no periodicity, solutions to cancel noise will be a quite hard job and instantenous solutions may be needed. The varios noise kinds, the various fitler types has been found. If noise contents are always same and known having periodicity, solutions will be easy with regarded success. Sometimes contens come with no information, in such cases high level mathmatical process will be required with highly costs of hardware. Instead, adaptive filters may be chosen in these conditions. Adaptive filters try to adjust their weights according to their own outputs related with their own inputs. These kind of filters have been used in a quite range of applications. LMS or NLMS oriented adaptive filters is a good example of these filters. LMS oriented filters results are not excellent but accaptable. In addition, LMS oriented filters are one of the easiets filters which is easily implemented on hardwares. Hardware consideration is a second problem with algorithms. The most performanced algorithm may have impossible hardware solutions. For this reason, finding the right relation between the algorithm and the hardware is a part of solution. In this thesis, the process of algorithm towards hardware solution is explained in details. This thesis explains: the LMS oriented self-adaptive fitler which takes two input and produces one output. Also, real-time process of fitler is explaind with use of FPGA. Keywords: Adaptive filters, LMS, FPGA, Real Time algorithms, audio processing, noise cancellation.