Vol 24 no.2 2024
1,5Department of Electrical and Electronics Engineering, Faculty of Engineering, Olabisi Onabanjo University, Ago-Iwoye, Ogun State, Nigeria; 2,4Department of Electrical and Electronics Engineering, College of Engineering, Federal University of Agriculture, Abeokuta, Nigeria; 3Department of Electrical and Electronics Engineering, College of Engineering, Afe-Babalola University, Ado-Ekiti, Nigeria;
Abstract
The increasing demand for enormous data rate is propelling interest in 5G millimeter wave communication. Several methods have been proposed for millimeter wave channel estimation but there is little or dearth of information on the impact of training overhead on the performance of these methods. This work investigated the effect of training overhead on the performance of orthogonal matching pursuit (OMP), compressed sampling matching pursuit (CoSAMP), deep learning (DL) and least square (LS) techniques by employing normalized mean square error (NMSE), spectral efficiency (SE) and bit rate as performance indices. It was observed from NMSE profile that smallest errors were recorded for DL and OMP when the training overhead was 60 while for COSAMP and LS, lowest errors were recorded when the training overheads were 55 and 65, respectively for 4-bit ADC. It was seen also that, SE and bit rate exhibited dissimilar characteristics over increasing values of training overheads