Abstarct: In this thesis we present a new adaptive algorithm, baxsed on particle swarm optimization (PSO) for echo cancellation. The proposed algorithm is done on the first echo path model of ITU-T G.168. This algorithm used to optimized the parameter of model which are the FIR coefficients, to create the replica echo and minimize it. The result shown 70db reduction on the echo level. The main advantage of PSO is that, it escape from local minima and also reduces the computational complexity of conventional methods. We modified PSO algorithm to increase convergence speed of the basic PSO. In this algorithm after some iteration, the algorithm measures fitness function of the particle and get a chance to search new areas in solution space. In this new algorithm when a sudden change occurred on the echo path model, the algorithm can adapt it.