Abstract
A novel bind estimation method for encoder parameters operating over the noisy received signal is proposed in this paper. This scheme can blindly identify the turbo encoder adopted at the transmitter so as to correctly decode the received signal sequence. An iterative expectation-maximization algorithm is designed to estimate the coding parameters, which are the weighting coefficients in a recursive convolutional encoder. These coefficients are associated with the feedback and forward connections in the encoder. To tackle this blind encoder-parameter estimation, we separate the feedback portion from the forward structure and then convert the recursive systematic convolutional encoder into a non-systematic convolutional encoder preceded by a feedback encoder. Our new encoder structure will be investigated. The effect of the separate feedback encoder on the state sequence resulting from the forward convolutional encoder will be studied. Monte Carlo simulation results will be demonstrated to evaluate the effectiveness of our proposed new scheme. © 2012 IEEE.