Parallel Semi-blind Joint Timing-Offset and Channel Estimation for AF-TWRNs

The estimation of the channel coefficients and the timing offset plays a critical role in the performance of asynchronous amplify-and-forward two-way relay networks. A highly accurate semi-blind algorithm for the joint estimation of the channel coefficients and the timing offset has recently been developed assuming generic pulse shaping filters. This algorithm is based on the expectation maximization (EM) framework, where the system is modeled using a hidden Markov model (HMM), and the posterior probabilities of the data are calculated using the Baum-Welch forward-backward algorithm. The high computational complexity of the Baum-Welch algorithm is the main challenge for real-time implementation of this algorithm. This challenge is addressed in this work by developing an efficient parallel implementation of the EM, including the Baum-Welch. Deep performance and time analysis are performed to study the opportunities for parallelization and choose the suitable technique. Using multithreading, we are able to achieve a maximum net speedup of 4.8X.