Abstract
Accurate and fast power control is perhaps the most important aspect in W-CDMA, in particular on the uplink. This thesis describes the work earned out to evaluate and enhance the link-level performance of closed-loop power control (CLPC) for the UMTS-FDD uplink. Factors that influence the performance of the CLPC scheme amongst others include the round-trip delay, UE (User Equipment) speed, dynamic range of the handset, the level of diversity available at the receiver and the accuracy of SIR (Signal-to-Interference-plus-Noise Ratio) estimation and channel estimation. The above-mentioned performance improvement is obtained by focussing on the SIR measurement aspect of the power control scheme, and to this end, two new signal quality estimation techniques are proposed and evaluated. Based on the two algorithms, a further comparison is made to investigate the effects of SIR measurement before and after RAKE combining of the signal. Firstly, an improved Eb/N0 estimator is proposed and its performance evaluated both in a stationary (Gaussian) as well as a Rayleigh channel. The improvement in the estimator's performance is obtained by the addition of a smoothing filter to an existing algorithm. The improved algorithm, based on maximum likelihood estimation, is shown to improve the performance of CLPC with an average gain of 0.25 dB (i.e. the average gain for a range of UE speeds) in the power control error (PCE) in a narrowband channel, over the performance without the filter; the algorithm would also offer gains in the wideband channel when the SIR measurement is done before RAKE-combining of the signal. The second algorithm is a traffic-based SIR measurement scheme, so-called because it uses a dedicated traffic channel for the SIR measurement, and hence specific to UMTS-FDD. Again, using the standard deviation of the PCE as the performance metric, it is shown that this scheme achieves gains in the PCE of up to 1.4dB (1.15dB on average) for the 4.75-12.2kbps speech service and up to 2.75dB (2.5dB on average) for 144kbps data, both in UMTS Vehicular environments. The gains achieved are with respect to the performance obtainable with the SIR measurement technique proposed by 3GPP (using dedicated pilot symbols), and they translate to improvements in the CDMA capacity at system-level. Furthermore, the algorithm, which currently assumes perfect blind rate detection, provides a higher improvement in the performance of CLPC for higher data rates (as the figures exemplify), which is an additional advantage since higher data-rate users cause more interference to the network by transmitting higher powers. Results obtained using the PCE performance metric also indicate that the higher path diversity available in W-CDMA (compared to narrowband systems) does not necessarily convert to improved system performance, especially when combining weak taps in the channel, i.e. the performance either stays the same or somewhat degrades. Finally, using the uncoded bit error rate (BER) as the performance metric, two further SIR measurement techniques, both based on variations of the two algorithms discussed earlier, are proposed and compared. The Pre-RAKE scheme, in which the SIR measurement is carried out before RAKE combining of the signal vs. the Post-RAKE scheme where the measurement is carried out after, are compared. The overestimation problem that occurs with the Pre-RAKE scheme is highlighted; hence, it is concluded that Post-RAKE schemes offer superior performance over their Pre-RAKE counterparts. The work has highlighted several issues that need careful consideration as far as SIR measurement is concerned, whether it be for power control or any other radio resource management technique; the overestimation problem which occurs for low SIR values as well as the issue of the number of symbols available for the measurement, can contribute significantly to the SIR measurement error, degrading the performance of CLPC in turn. In addition, it is also useful to have an idea of how much the true SIR can vary within the duration of measurement. Last but not least, the little or no improvement in performance obtained when combining weak taps is attributed to the performance and robustness of the channel estimation algorithms employed; they tend to add more noise than signal power to the RAKE-combined signal when used to combine paths with low Eb/N0 values. The SIR measurement algorithms proposed and discussed in this thesis can be implemented directly in UMTS receivers. On the other hand, the knowledge and information contained herein and the logical approaches considered can lead to the development of new ideas as well as new algorithms.