Semi-blind channel estimation for Zero Padded-OFDM systems with asynchronous interferers

摘要

In this paper, a subspace-based semi-blind channel identification scheme is developed for a zero padded-orthogonal frequency division multiplexing (ZP-OFDM) system with asynchronous interferers. In our model of multiuser OFDM signals, an asynchronous interferer has a guard interval within the transmitted data block when the observation window is synchronized with respect to the desired user. As a result, the interfering user's channel vector lies in a different null space relative to the desired user's subspace, allowing the desired user's channel vector to be separated and determined uniquely up to a multiplicative scalar. However, in practice, a noisy sample correlation matrix produces a number of small eigenvalues, which can cause poor performance when conventional subspace techniques are employed. This problem is overcome by using several eigenvectors corresponding to the smallest eigenvalues; the channel estimates are obtained by linear combinations of these vectors. In order to find the linear combining coefficients, known pilot symbols are required. However, the pilot-only estimation method is not feasible when the channel order and the number of active users are large. Therefore, we use the preceding method to develop a subspace-based semi-blind algorithm that guarantees estimation of every user's channel vector. Simulation results demonstrate that the semi-blind channel estimator offers more than a 2dB gain over pilot-only channel estimation in the bit error rate (BER) performance.