| The constant modulus algorithm is an effective adaptive filtering technique, which compensates for interference caused by multipath propagation and cochannel signals. A system of serial stages, each stage consists of constant modulus array followed by an adaptive signal canceler, resulting in an effective adaptive processing technique for separation of cochannel signals received by an array of antennas.;The advantages of the constant modulus algorithm over the equivalent adaptive algorithms are that it does not require a reference signal and it does not rely on the inverse of the correlation matrix of the input data, which may be singular. These advantages make the constant modulus algorithm suitable for applications such as cellular mobile communications and electronic reconnaissance, where it is difficult to obtain the reference signal, and the received signal is usually correlated with interference signals. The application limitations of the constant modulus algorithm are mainly due to its slow rate of convergence.;Two new algorithms are introduced in this dissertation, which keep the advantages of the original constant modulus algorithm, while accelerating the convergence of the adaptation process. Each of the new algorithms provides an average reduction by ninety percent of the number of iterations required for the original constant modulus algorithm to converge.;The behavior of the considered algorithms and the system for separation of cochannel signals are analyzed analytically and supported by computer simulations for different classes of constant envelope signals. |
