Soft -output detection and decoding algorithms with joint channel estimation for direct -sequence spread -spectrum systems
Abstract
In this thesis, the soft-output detection and decoding algorithms with joint channel estimation are developed for aperiodic random direct-sequence spread-spectrum (DS-SS) systems. Soft-decision algorithms provide more information than those with hard decisions, giving better bit-error-rate (BER) performance in the following decoding stage. The receiver considered is a single-user receiver without knowledge of any other users, and the channel is assumed to vary within tens of symbol time. For such time-varying channels, channel estimation is necessary for generating soft decisions. Especially, for a single-user, aperiodic random DS-SS receiver, the channel estimation relies additionally on interference suppression because the channel state information and crosscorrelations from interfering users are unknown random variables which change from symbol to symbol. To provide sufficient dimensions for interference suppression, multiple signal samples are taken either in time or space domain. With a state model formulation of the channel and the received signal, a joint channel estimation and MAP detection algorithm with real-time model parameter identification is proposed. First, based on the MAP detection requirement, a maximum likelihood (ML) estimator of unknown model parameters is derived. Under certain assumptions on the interference, the estimated parameters are proved to converge asymptotically to the true parameters with probability one. However, the complexity of the ML parameter estimation is prohibitively high for practical implementation. Instead of the ML estimator, the well known extended Kalman filter (EKF) is applied as part of the solution. The linear constraint in the antenna array and the non-observable model noise variance are estimated by simple and effective approximations. In the special case of time-domain oversampling, further complexity reduction is achieved, and negligible performance degradation in the following decoding stage is found through computer simulation. In addition, a unified implementation of fixed-delay, symbol-by-symbol MAP detection/decoding algorithms is proposed. With proposed modifications to lower the numerical requirement, the Optimum Soft-output Algorithm (OSA) and its suboptimal approximations can be integrated into a common algorithmic structure. Furthermore, by applying the principle of the Soft-Output Viterbi Algorithm (SOVA), the number of add-compare operations in the Sub-optimal Soft-output Algorithm (SSA) can be further reduced with little performance degradation.
Degree
Ph.D.
Advisors
Lehnert, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.