Blind subspace methods for code and channel estimation in Multicarrier CDMA

Abstract

In multicarrier code-division multiple access systems the linear distortion introduced by the channel causes a perturbation in the received user codes that produces multiple access interference. This paper introduces two new techniques to combat this distortion that are based on subspace decomposition. The first one consists of obtaining an approximation of the received desired user code to improve the output signal to interference noise ratio in blind linear receivers. This approximation is obtained by means of the projection of the known transmitted desired user code onto the signal subspace. The second one is a blind channel estimation method that uses the orthogonality property between the received user codes and the noise subspace. We demonstrate that the performance of channel estimation methods based on this property are severely limited by the number of simultaneous users. Nevertheless, it is shown how this limitation is overcome if user codes spanning several transmitted symbols (long codes) are used.