Complex FIR block adaptive algorithm employing optimal time-varying convergence factors

The Complex Least Mean Square algorithm (Complex LMS) has been widely used in various adaptive filtering applications, e.g. in the wireless communications and biomedical fields, due to its computational simplicity. However, the main drawback of the Complex LMS algorithm is its slow convergence. In addition, the performance is dependent on the choice of the convergence factor or learning rate. In this paper, a novel complex block adaptive algorithm is presented that overcomes the performance limitation of the Complex LMS. The proposed algorithm (Complex OBA-LMS) derives independent time-varying convergence factors for the real and imaginary components of the FIR complex adaptive filter coefficients. Furthermore, the convergence factors are updated at each block iteration. The convergence speed and accuracy of the Complex OBA-LMS algorithm are investigated and compared with the Complex LMS algorithm. Simulation results show that the proposed technique exhibits superior performance at the expense of a modest increase in computational complexity for different training inputs.