Blind adaptive multiuser detection for cellular systems using stochastic approximation with averaging

Abstract

Abstract-In this paper, we consider blind adaptive multiuser detection in correlated waveform multiple-access-based cellular radio networks. A common stochastic approximation (SA)-based framework is proposed from which three blind adaptive algorithms for linear minimum mean squared error detection are obtained. Two of them coincide with previously proposed algorithms and the third is shown to be best suited for implementation at a base station. The work here also improves these SA-based adaptation algorithms in the context of cellular radio networks, in terms of convergence properties by using the more recent results on the SA technique with averaging. Convergence issues of the different adaptations are investigated and numerical examples are presented to demonstrate the performance improvement due to averaging.

Distributed Algorithms for Joint Optimization of Multiuser Receivers and Power Control

“…In the latter case, with the powers fixed at w(n N ), (where n N > 0 is large enough to ensure that the user powers are sufficiently close to the optimum powers), the MSIR filters can be obtained blindly using well known adaptive algorithms in [14], [17] and more recently in [15], [16]. For blindly estimating the optimum MSIR receivers f * ij concurrently with the power updation, one can use the following stochastic approximation based recursion indexed by (n, m), where m is the updation number within the n th M -length block of received vectors (recall that the powers are updated only after every M -length interval and, therefore, remain constant for each f ij (n, m + 1)…”

Standard and Quasi-Standard Stochastic Power Control Algorithms

“…The proof of Theorem 2 is given in Appendix C. In the quasi-standard stochastic PC algorithm, define (24) Similar to (19), we define the scaled noise and the scaled bias as (25) When , from (9) and (11), we obtain (26) Both Theorem 2 and (26) are key results for the proof of the following theorem, which is one of the major steps toward the proof of the main result. , is not considered in [16], it can be taken into consideration by straightforward modifications to [16]; therefore, the proof is omitted in this paper.…”