Deterministic channel tracking and signal detection for multi-carrier transmissions over high-mobility radio channels

Adding into the cluster of available detection schemes designed for scenarios where user terminals are highly mobile, this paper describes a semi-self-recovery OFDM signal detection method for rapidly time-varying channels. The proposed algorithm does not assume the channel status information (CSI) at the receiver, it tracks the channel variation while detecting the data symbols. The key feature of the scheme is that it exploits both the finite alphabet (FA) property of the symbols and the finite number of the discrete significant signal replicas over multipath channels. Characteristically, it judiciously utilizes a total-variation regularization (TVR) and iteratively enforces FA property to gain the accuracy of the channel tracking. TVR was widely applied in the image processing area for image denoising. In the proposed method for multi-carrier transmissions, it is found very effective to properly employ TVR in suppressing the spurious components in the estimated channel coefficients. As a result, the algebraic method is able to well exploit the signal frame structure, it can work either in a semi-self-recovery fashion without pilot-subcarriers or work with an arbitrary pilot-subcarrier pattern. Computational simulation verifies that the proposed scheme exhibits a quick convergence, and at a benign or a high SNR case, it delivers a performance asymptotically approaching that of the inter-carrier-interferences (ICI) mitigating least-square detection method that ideally assumes perfect CSI at the receiver.