Code-Aided Channel Tracking and Decoding over Sparse Fast-Fading Multipath Channels with an Application to Train Backbone Networks

In a fast-fading environment, e.g., high-speed railway communications, channel estimation and tracking require the availability of a number of pilot symbols that is at least as large as the number of independent channel parameters. Aiming at reducing the number of necessary pilot symbols, this work proposes a novel technique for joint channel tracking and decoding, which is based on the following three ideas. 1) Sparsity: While the total number of channel parameters to be estimated is large, the actual number of independent multipath components is generally small; 2) Long-Term versus short-Term channel parameters: Each multipath component is typically characterized by long-Term parameters that slowly change with respect to the duration of a transmission time slot, such as delays or average power values, and by fast-varying fading amplitudes; and 3) Code-Aided methods: Decision-feedback techniques can optimally leverage past, and partially reliable, decisions on the data symbols to obtain 'virtual' pilots via the expectation-maximization (EM) algorithm. Numerical results show that the proposed code-Aided EM algorithm is effective in performing joint channel tracking and decoding even for velocities as high as 350 km/h, as in high-speed railway communications, and with as few as four pilots per orthogonal frequency-division multiplexing data symbol, as in the IEEE 802.11a/n/p standards, outperforming existing schemes at the cost of larger computational complexity.