| The Algebraic Constant Modulus Algorithm (ACMA) works on signals with constant modulus by exploiting the property of signal independence between different sources. The Semi-Blind Algebraic Constant Modulus Algorithm (SB-ACMA) is a variation of ACMA which uses the availability of training data. It works in a similar way but constraints the subspace of the solution using the training symbols. The resulting beamformer is unambiguous in phase and computationally less expensive to calculate. In this thesis we study the asymptotic properties of the SB-ACMA and prove that it converges to the (non-blind) Wiener Beamformer Solution. We also analyze the performance of the algorithm with respect to the Cramer-Rao Bound, calculated for one CM and all other Gaussian sources, and show that the empirical performance is close to the lower bound. |
