3G Wideband CDMA : packet-based optimisation for high data-rate downlink transmission

Abstract

A third generation (3G) of mobile communication systems, based on Wideband CDMA, are intended to offer high-speed packet-based services. Network operators wish to maximise the throughput in the downlink of3G systems, which requires efficient allocation ofresources. This thesis considers the problem ofmaximising throughput in an interference dominated channel. Cooperative broadcasting is a theoretical technique to mitigate this problem. Its implementation in practical systems requires efficient resource allocati.on to maximise the thr(oughput whilst meeting system and user-imposed constramts. A resource allocation approach is presented for implementing cooperative broadcasting. Users are paired and a teclmique for allocating resources between the pair is developed. Then, a method for pairing the users is considered. Simulation results are presented, which show a throughput improvement over existing resource allocation approaches. The problem ofcontrolling the distribution ofrandomly arriving data to meet the resource allocation specifications is examined. A single-threshold buffer is proposed, which requires fewer calculations than an existing double-threshold buffer. Simulation results are presented which show a throughput improvement may be realised, greater than that which would achievable using other rate control schemes. Cooperative broadcasting may lead to transmissions to some users being allocated low power. When full channel infonnation is available at the transmitter, a water filling solution may be used to maximise capacity. However, when combined with buffer management, erasure may result. This erasure may be overcome using an erasure protection code. Such a code is examined. When combined with Turbo coding, ajoint detector may be used for providing error and erasure protection. Analysis ofthis detector shows a lower limit on the error rate, dependent on the probability of erasure. Simulation results show that using this approach the error rate is significantly improved. This code can then be used to increase capacity, whilst achieving low error rates.