In this work the benefits and the limits of turbo-detection for GSM-systems are examined. Turbo-detection applies the turbo-principle to iterative equalization and decoding. The channel encoder and the transmission channel are regarded as a serial concatenation of convolutional codes, and hence, can be iteratively decoded. Since turbo-detection requires modifications of the transmission only at the receiver, it can be adopted to existing mobile radio systems without any amendment of the transmission standard. Before applying this method to existing GSM-services, e.g. full-rate speech, the original turbo-detection scheme must be changed to enable acceptable transmission delays. New schemes are developed that differ in complexity, delay, memory requirement, and performance. The performance difference is evaluated using time-invariant intersymbol-interference channels. It is shown that the performance of the original turbo-detection scheme can be maintained for these channels. The influence of channel parameter estimation and suboptimum detection algorithms is investigated. However, in order to evaluate the benefits for GSM-systems, the transmission channel has to comprise an exact model of the modulator and the mobile radio channel. Applying the turbo-detection schemes to GMSK modulated GSM-services, only small iteration gains are obtained after decoding due to the orthogonal intersymbol-interference of the modulator. Even for multipath environments with large delay spreads the iterations gains do not exceed 0.7 dB since the paths are fading independently. Higher-order modulation schemes will be used for new high-data rate services within GSM. Introducing the symbol soft-values, the turbo-detection scheme can be easily adopted to these modulation techniques. It is shown that for the new packet switched services large iteration gains of up to 2 dB are achieved. By considering the signal-to-noise distribution in the network, the throughput of the system, and hence, the spectral efficiency can be improved by up to 30%. Simulation results show that turbo-detection also guarantees large gains in severe mobile radio environments. Additionally, turbo-detection is applied to a system of serial concatenated convolutional codes. By iteratively detecting this double serial concatenated system the performance is further increased compared to simple turbo-decoding. For fast-fading channels an adaptive Max-Log-MAP equalizer is developed that can be incorporated in the turbo-loop. This adaptive equalizer profits from the decoder decisions and accurately tracks the channel variations. However, no significant iteration gain is achieved after decoding for GMSK-modulated services.
«
In this work the benefits and the limits of turbo-detection for GSM-systems are examined. Turbo-detection applies the turbo-principle to iterative equalization and decoding. The channel encoder and the transmission channel are regarded as a serial concatenation of convolutional codes, and hence, can be iteratively decoded. Since turbo-detection requires modifications of the transmission only at the receiver, it can be adopted to existing mobile radio systems without any amendment of the transmis...
»