Iterative receivers and multichannel equalisation for time division multiple access systems

Abstract

The thesis introduces receiver algorithms improving the performance of TDMA mobile radio systems. Particularly, we consider receivers utilising side information, which can be obtained from the error control coding or by having a priori knowledge of interference sources. Iterative methods can be applied in the former case and interference suppression techniques in the latter.

Convolutional coding adds redundant information into the signal and thereby protects messages transmitted over a radio channel. In the coded systems the receiver is usually comprised of separate channel estimation, detection and channel decoding tasks due to complexity restrictions. This suboptimal solution suffers from performance degradation compared to the optimal solution achieved by optimising the joint probability of information bits, transmitted symbols and channel impulse response. Conventional receiver utilises estimated channel state information in the detection and detected symbols in the channel decoding to finally obtain information bits. However, the channel decoder provides also extrinsic information on the bit probabilities, which is independent of the received information at the equaliser input. Therefore it is beneficial to re-perform channel estimation and detection using this new extrinsic information together with the original input signal.

We apply iterative receiver techniques mainly to Enhanced General Packet Radio System (EGPRS) using GMSK modulation for iterative channel estimation and 8-PSK modulation for iterative detection scheme. Typical gain for iterative detection is around 2 dB and for iterative channel estimation around 1 dB. Furthermore, we suggest two iteration rounds as a reasonable complexity/performance trade-off. To obtain further complexity reduction we introduce the soft trellis decoding technique that reduces the decoder complexity significantly in the iterative schemes.

Cochannel interference (CCI) originates from the nearby cells that are reusing the same transmission frequency. In this thesis we consider CCI suppression by joint detection (JD) technique, which detects simultaneously desired and interfering signals. Because of the complexity limitations we only consider JD for two binary modulated signals. Therefore it is important to find the dominant interfering signal (DI) to achieve the best performance. In the presence of one strong DI, the JD provides major improvement in the receiver performance.

The JD requires joint channel estimation (JCE) for the two signals. However, the JCE makes the implementation of the JD more difficult, since it requires synchronised network and unique training sequences with low cross-correlation for the two signals.