Decision Directed and Constant Modulus Algorithms Derived and Evaluated for Multicarrier Systems

摘要

The purpose of this research is to derive and examine blind adaptive algorithms for equalizing multicarrier (MC) communication systems. The author analyzes how they perform under varying environmental conditions and parametric variations, focusing on equalizers set in cascade with the channel. Two well- accepted and widely known cost functions, Decision Directed (DD) and Constant Modulus (CM), are applied to the MC signal structure, and gradient descent algorithms based on both DD and CM functions are derived, analyzed, and compared. Comparison of the new algorithms, Multi-Carrier Decision Directed (MCDD) and Multi-Carrier Constant Modulus (MCCM), focuses on detailing how each algorithm performs when the factors of noise power and symbol synchronization are varied. Additionally, a Frequency Domain Equalizer (FEQ) is developed and employed to de-rotate and resize the output symbol constellation. Both MCDD and MCCM are compared against other blind and trained adaptive MC equalization algorithms in the areas of bit error rate (BER) versus signal-to-noise ratio (SNR) and BER versus synchronization delay. Results are presented showing that MCDD and MCCM perform worse than a MC-Trained (MCT) approach, but better than both Multicarrier Equalization by Restoration of Redundancy (MERRY) and Carrier Nulling Algorithm (CNA).