Design of improved space-time trellis codes with and without serial concatenation.

摘要

Space-Time-Trellis Codes (STTC) were first introduced in 1998 by Tarokh et al [1] for wireless communication channels with multiple transmitter and receiver antennas to maximize temporal and spatial diversity gain on a quasistatic fading channel. However, this space-time code is relatively weak on its own in terms of coding gain. There are many papers [2--4] that search for better space-time codes using the criteria proposed in [1]. Research results show when diversity gain is reasonably large, a multiple transmitter/receiver fading channel converges to a Gaussian channel [4]. As such, the Euclidean distance criteria (EDC) can be used for code design. In this thesis, a simple method is proposed to construct better space-time codes based on the EDC. We implement space-time trellis codes with a time-multiplexer, a convolutional encoder and a mapper, and we optimize the convolutional encoder and the mapper to obtain good STTCs. Our method can be used to construct, both non-recursive and recursive space-time codes where the latter is essential for serial concatenation. The thesis investigates the performance of the serial concatenation of the newly designed recursive space-time codes as the inner codes and with convolutional codes as the outer codes. In order to evaluate the performance of the new concatenated space-time codes, Jakes time correlated Rayleigh fading model is used on a two transmitter antenna system. The decoding method follows an iterative (turbo-like) strategy using the Maximum A-Posteriori (MAP) algorithm. The simulation results show improvement for the new space-time codes over other space-time codes both with and without serial concatenation. With serial concatenation, the bit error rate (BER) of our code is 1 dB better than in [5].