Iterative Decoding of Serially Concatenated CPM in Fading Channels with Noisy Channel State Information

We use the sum-product algorithm (SPA) to propose a reduced-complexity detector for serially concatenated continuous phase modulated (SCCPM) signals with noisy channel state information (CSI) at the receiver. Two channel effects - phase noise (PN) and flat fading - are tackled using the statistical model and associated factor graph (FG) of the problem. Although optimal performance is not assured due to the loopy nature of the FG, as well as simplifying assumptions made in the algorithm development, simulations show that very significant performance gains are possible, and the gap between systems with noiseless and noisy CSI can be reduced to about I dB using the proposed methods. The increase in complexity compared to the conventional decoder, which treats CSI estimates as if they were equal to their true values, is small.