Adaptive filterbank based blind channel estimation for multicarrier systems

We present herein a novel blind channel estimator for multicarrier (MC) systems in the presence of unmodeled narrowband interference. A generalized multi-channel minimum variance principle is invoked to design an equalizing filterbank that preserves desired signal components and suppresses the overall interference. While a channel estimate can be obtained by maximizing the filterbank output power through multi-dimensional nonlinear searches, such an approach is computationally prohibitive and suffers local convergence. To overcome this difficulty, an asymptotically (in SNR) tight lower bound on the filterbank output power is derived and used for channel estimation, which reduces to a quadratic minimization problem. The proposed scheme is shown to compare favorably with a subspace blind channel estimator in the presence of unknown narrowband interference.