Performance of adaptive equalizers for systems with code-multiplexed pilot

We derive analytical expressions for the output SINR of equalizers trained on code-multiplexed pilot using the least squares and least mean square (LMS) adaptive algorithms. The results show the exact dependence of the convergence behavior on the number of training samples, the equalizer´s tap complexity, the fraction of power devoted to the forward link pilot, and the statistics of the channel and interference. Remarkably, the latter dependence is captured entirely by the Wiener STNR achievable with the specified tap complexity.