Performance analysis of the linearly constrained constant modulus algorithm-based multiuser detector

Abstract

This paper quantifies the adaptive performance of a blind adaptive multiuser detector (MUD) based on the linearly constrained constant modulus algorithm (LCCMA) in both a stationary and nonstationary channel. A framework is developed to apply the feedback analysis method to analyzing adaptive MUD schemes. A closed-form expression for the excess mean square error (EMSE) of LCCMA blind adaptive MUD in a CDMA communications system is derived for both of the steady-state and tracking cases. The effects of additive noise and multiple access interference are considered. A transient analysis is performed that predicts the learning curve of the adaptive filter. Computer simulation is used to verify the accuracy of the analysis.Index Terms-Blind adaptive multiuser detection, code division multiple access, constant modulus algorithm, excess mean square error, feedback analysis, steady-state analysis, tracking analysis.

A class of constant modulus algorithms for uniform linear arrays with a conjugate symmetric constraint

“…Changing the weight vector in the negative direction of the gradient in (20), scaled by a constant stepsize µ, we have the following update equation for the weight vector w w[n…”

“…One can follow the general steps presented in [20] and a recent paper with a more complete analysis to the CM based algorithms [13], and it is omitted here.…”

“…The first step in (35) is exactly the same as the original LCCMA derived without the proposed conjugate symmetric constraint [20]. Therefore the effect of the weight constraint is again equivalent to adding a second step to the original algorithm.…”

Steady-state performance analysis of HOS-based blind adaptive multiuser detection

“…[7] to obtain the accuracy steady-state EMSE of the CM multiuser detection. The assumption for the EPEbased analysis is similar to that of ODE.…”

“…In ref. [7], Whitehead introduced an energy preserving equation (EPE) to derive the closed-form expression of the steady-state excess mean-squareerror (EMSE) of the constant modulus multiuser detection. Simulations validate its feasibility and preciseness.…”