Optimum Training Power Derivation for Affine Precoding and Decoding based CDMA-FBMC/OQAM System

In this work, the least square (LS) channel estimation of the affine precoding and decoding-based code division multiple access filter bank multicarrier offset quadrature amplitude modulation (CDMA-FBMC/OQAM) system is investigated. In this system, the training sequence matrix is superimposed with the precoded data matrix before FBMC modulation. The power coefficient of the precoded data $\left( {\sigma _s^2} \right)$ and training sequence $\left( {\sigma _c^2} \right)$ are normalized for single user such that $\sigma _s^2 + \sigma _c^2 = 1$. The effective average signal-to-noise power ratio (SNR
eff
) is derived. By differentiating SNR
eff
with respect to $\sigma _c^2$, the theoretical values of optimum training power coefficient $\left( {\sigma _{{c_{opt}}}^2} \right)$ is derived for which the SNR
eff
is maximum. Further, $\sigma _{{c_{opt}}}^2$ is obtained through bit error rate (BER) simulation where $\sigma _{{c_{opt}}}^2$ is the point for which BER is minimum and its closeness with theoretical value is observed.