Simulation of a subspace tracking algorithm in the mobile environment

This paper studies the performance of a noniterative subspace tracking algorithm in mobile CDMA, including mismatch loss, convergence speed and system capacity. The array response vector of the uplink channel can be estimated by computing the principal eigenvector of data covariance matrix, which can be decomposed into an one-dimension signal subspace and multiple-dimension noise subspace. By averaging the noise subspace, we convert the principal eigenvector update into a 2×2 eigen problem, which largely reduces the computation complexity. Simulation results show that real-time tracking for the mobile users can be implemented, and the SNR loss due to estimation error is less than 0.5 dB