Channel estimation for power line OFDM systems with unknown colored noise exploiting real signal transmission

In this paper, we propose pilot-aided subspace tracking channel estimators for OFDM systems over power line multipath channels (PLC) with unknown colored noise exploiting real signal transmission. In the proposed estimators, the signal subspace and the noise subspace are identified through decomposition of pseudo covariance matrix of received time domain samples, unlike the conventional subspace tracking which is based on the decomposition of covariance matrix. The least square (LS) technique together with the identified signal subspace are then used to derive the channel estimate in time domain using pilot OFDM symbols. The proposed estimators do not assume the knowledge of statistics of colored noise. The simulations show that, the proposed pilot-aided subspace tracking channel estimators outperform conventional LS based channel estimators, however, the performance gain is degraded with the increase of number of multipath. Further, the proposed estimators require the number of delay path known as a prior.