Abstract

User identification provides valuable information about the active users in a Code Division Multiple Access (CDMA) system that could be used in numerous applications. Most importantly, it can produce a reliable estimate of multiple access interference, whose cancellation is the main objective of multiuser detection schemes. Several schemes have been introduced for user identification in a synchronous CDMA system. The main distinguishing factor between different user identification schemes is the amount of information about each user required to be available in advance at the receiver. The existing schemes require prior knowledge of users' spreading codes or some other forms of assistance. However, the proposed method does not need the prior knowledge of the spreading codes nor relies on any assistance. Thus, using the scheme proposed in the thesis, one will be able to perform complete blind identification for a group of users. This identification involves determination of the spreading codes of active users and estimation of their relative timings. In this thesis, a scheme for blind identification of users in a chip-synchronous Direct Sequence CDMA (DS-CDMA) system is presented. The proposed approach is based on the subspace decomposition and searches for a solution that minimizes a defined cost function. The search is performed over both signal space and time domain. By completing the identification of the users, other important parameters such as user timings and power can be evaluated. Blind identification of users for an asynchronous CDMA system lends itself to several applications such as CDMA up-links, ad-hoc networking, signal interception and non-intrusive tests.