Rank constrained temporal-spatial matrix filters for CDMA systems

• Published in: IEEE transactions on wireless communications Vol. 3; no. 6; pp. 1974 - 1979
• Main Authors: Filiz, O., Yener, A.
• Format: Journal Article
• Language: English
• Published: Piscataway, NJ IEEE 01.11.2004; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Applied sciences; Array signal processing; Complexity; Constraint optimization; Constraints; Detection, estimation, filtering, equalization, prediction; Exact sciences and technology; Filtering; Information, signal and communications theory; Interference constraints; Interference suppression; Iterative algorithms; Least mean squares algorithm; Mathematical models; Minimization; Minimization methods; Minimum mean-squared error (MMSE); Multiaccess communication; Multiuser detection; Nonlinear filters; Optimization; Receivers; Signal and communications theory; Signal, noise; Studies; Telecommunications and information theory; temporal-spatial filtering; Multiuser detection; Interference suppression; Filtering; Simulation; Wireless telecommunication; Code division multiple access; Receiver; Beam forming; Optimization; Mean square error; Least mean squares methods
• ISSN: 1536-1276; 1558-2248
• DOI: 10.1109/TWC.2004.837393

Efficient interference suppression techniques are needed to maximally utilize the potential gains of code-division multiple-access systems. In this letter, a receiver structure which combines multiuser detection (temporal filtering) and receiver beamforming (spatial filtering) in a multipath environment is considered. Following previous work, we model the receiver as a linear matrix filter and use the minimum mean-squared error (MMSE) as the performance criterion. Motivated by the high complexity of the optimum receiver, we propose rank constrained temporal-spatial filters which are simpler and near-optimum. The MSE is minimized subject to a structural constraint, using an iterative algorithm based on alternating minimization. The constraint on the receiver matrix filter narrows down the solution space, which helps to solve the optimization problem more efficiently. The constraint can be set appropriately by the system designer to achieve the desired tradeoff between performance and complexity. Numerical results indicate that a performance close to that of the optimum filter can be achieved with a simple iterative structure, even in highly loaded systems. Adaptive implementation of the rank constrained filters is derived. A new adaptive scheme is proposed which is a combination of the alternating minimization and the least mean squares methods. The convergence properties are investigated along with the effect of the number paths.