Abstract

Wideband Code Division Multiple Access (WCDMA) system uses orthogonal channelization codes to distinguish physical channels in a base station, while base stations are identified by different downlink scrambling codes. User equipments (UEs) must achieve synchronization to the downlink scrambling code before decoding any messages from base stations. The process of searching for a base station and synchronization to the downlink scrambling code is often referred to as cell search. The performance of cell search has a significant impact on a UE's switch-on delay, and thus it is very important to UE design. The goal of designing a cell search algorithm is to achieve a balance between speed, accuracy and complexity. A basic three-stage cell search procedure has been defined by 3GPP. It employs synchronization channels and the common pilot channel to facilitate a fast cell search. This cell search scheme only works well if there is no frequency offset between a base station's transmitter and a UE's receiver and if sampling timing is perfect on a UE. In practice, however, imperfection of oscillator in a UE may cause a big frequency error as well as clock error. It usually results in phase rotations and sampling timing drifts, which may degrade cell search performance significantly. Some advanced cell search algorithms have been proposed for mitigating impacts of frequency error or clock error. However, there is no much discussion on comprehensive solutions that can deal with the two negative impacts at the same time. In this thesis, we propose an algorithm that considers both frequency error and clock error. A fast and accurate cell search with a relatively low level of complexity is achieved. The algorithms are based on a combination of four existing enhanced cell search algorithms that are designed for a toleration of either frequency error or clock error. We first introduce the 3GPP-defined cell search algorithm as a basis. Then the four existing enhanced algorithms, PSD (partial symbol de-spreading), DDCC (differential detection with coherent combining), STS-1 (serial test in stage-1) and RSPT (random sampling per trial) are presented. Next, we propose four possible combinations of the existing algorithms: PSD+STS-1, PSD+RSPT, DDCC+STS-1 and DDCC+RSPT. Through extensive computer simulations, we find the DDCC+RSPT algorithm to be the best one. It is superior to other combinations and also outperforms any existing algorithm in terms of acquisition time, detection probability and complexity. Therefore, it is highly recommended for practical uses.