Abstract:

An improved channel estimation algorithm for orthogonal frequency division multiplexing (OFDM) systems with virtual carriers is proposed. As conventional estimator cannot estimate the channel transfer function (CTF) at virtual carriers, which makes channel energy leakage after inverse discrete Fourier transform, and the time domain filtering method is not directly applicable. To circumvent this problem, the least square (LS) method is applied to estimate the CTF at virtual carriers by using the limitation of channel impulse response (CIR). Further, by exploiting the noise correlativity between signal subspace and noise subspace, the maximum a posterior probability (MAP) criterion is applied to estimate the noise in signal subspace and then the estimation error brought by it is also suppressed. In addition to the training mode, the proposed method can also be extended and used in the tracking mode with dicison aided feedback. Simulation results show that the improved scheme is free of symbol error rate (SER) floor, and its SER attains 2 dB signal-to-noise ratio gains compared with that of regular LS estimator.