Turbo equalization with low complexity decoder

Turbo equalization (TE) is an iterative receiver algorithm repeating equalization and channel decoding to improve receiver performance. As equalization and decoding are performed several times, the complexity of the TE receiver increases respectively. Furthermore, the decoder needs to provide probability information as output, which increases the computational burden even more. In this paper we introduce a new method to reduce the decoding complexity of the TE receiver. This soft trellis decoding (STD) technique utilises information obtained from the previous decoding step to select those trellis metrics that need to be recalculated. Hence, the most unlikely trellis branches are eliminated. We evaluate STD performance in the Enhanced General Packet Radio System (EGPRS) platform by simulations. We show that STD reaches performance close to full trellis decoding, but requires only a fraction of the computational power