Generalized Least Squares Based Channel Estimation for High Data Rate FBMC-OQAM

This paper addresses the problem of preamble-based Channel Estimation (CE) in filter bank multi-carrier systems with offset quadrature amplitude modulation (FBMC-OQAM), a task that is problematic due to the presence of intrinsic Inter-Carrier Interference (ICI) and Inter-Symbol Interference (lSI). In this work, we present a theoretical analysis of the Mean Squared Error (MSE) of several CE schemes based on the virtual symbol (VS) approach such as the Interference Cancellation Method (ICM) and the Interference Approximation Method (IAM). We also propose and analyze a generalized least squares (GLS) based CE scheme which is capable of achieving improved performance by taking into account the correlation properties of the noise and interference among sub-carriers. Simulation results are presented which verify the accuracy of the theoretical analysis, and also demonstrate that the proposed GLS-based estimation method yields an improvement of almost 3dB in the CE-MSE with respect to the VS-based approach. Finally it is shown that, for higherorder modulation schemes, this CE-MSE reduction translates into significant improvement in the system's overall bit error rate (BER) performance with only minor additional complexity.