Forward error correction of FSK alphabets for noncoherent transmissions over AWGN channel

We show that several coding schemes, in particular turbo encoding, dramatically improve the performance of a noncoherently detected frequency shift-keying modulation with correlated signals. The proposed a posteriori probability (APP) decoder based on bit-by-bit decisions proves to be highly flexible and more efficient than a classical block decisions decoding scheme. Moreover, the choice of the FSK modulation tone spacing is important. Simulation results show that the turbo-coded FSK exhibits the lowest error probability when compared to TCM, convolutional and lattice encoding schemes.