LL Logo




Quadratically Constrained RLS Filtering for Adaptive Beamforming and DS-CDMA Multiuser Detection

Kristine L. Bell, Zhi Tian, and Harry L. Van Trees
George Mason University
341 Science and Technology Building
Fairfax, VA 22030-4444
tel: (703) 993-1707
email: kbell@gmu.edu

Abstract The Minimum Output Energy (MOE) detector used in Direct-Sequence Code Division Multiple Access (DS-CDMA) communications systems is analogous to the Linearly Constrained Minimum Power (LCMP) beamformer used in array processing. In both applications, a quadratic constraint on the weight vector norm can improve robustness to mismatch in the temporal or spatial signature vector.  We describe a technique for implementing a quadratic inequality constraint with Recursive Least Squares (RLS) updating in both the direct LCMP and the Generalized Sidelobe Canceler (GSC) (also called the Partitioned Linear Interference Canceler (PLIC)) structures. A variable diagonal loading term is incorporated at each step, where the amount of loading required to satisfy the quadratic constraint is given in closed form. Comparisons are made with several other robust RLS and Least Mean Squares (LMS) algorithms. Simulations show that the variable loading RLS technique offers better convergence and robust control over mismatch than other LMS and RLS implementations for both beamforming and multiuser detection.

Presentation (pdf format)



LL Logo Disclaimer

Direct comments and questions to: webmaster@ll.mit.edu

MIT Lincoln Laboratory. All rights reserved.