Iterative channel estimation using soft decision feedback

We present an iterative channel estimation scheme with soft decision feedback. In a block-by-block transmission system with a soft-output equalizer and/or decoder it is possible to feed back the decisions to the channel estimator to improve its performance. We propose to feed back soft decisions which will reduce error propagation. We also derive a suboptimal estimator with substantially lower complexity and present simulations for a GSM-like system. The improvement of the bit-error rate over non-iterative schemes with this low complexity estimator is about 0.8 dB (equalizer feedback) and 1.5 dB (decoder feedback).