Summary
International Conference on Emerging Technologies for Communications
2020
Session Number:D3
Session:
Number:D3-2
K-Factor Estimation based on Spectrogram Images by Convolutional Neural Network
Shun Kojima, Kosuke Shima, Kazuki Maruta, Chang-Jun Ahn,
pp.-
Publication Date:2020/12/2
Online ISSN:2188-5079
DOI:10.34385/proc.63.D3-2
PDF download
Summary:
In the next generation mobile communications systems, accurate and fast acquisition of the communication environment is essential to achieve high-speed and low-latency communication. A K-factor in Rician fading is a major determinant of the link quality. Therefore, K-factor estimation is a very important task in order to achieve high performance of adaptive control in wireless communications that is dependent on the link quality. Conventional methods for estimating the K-factor have been widely studied, including the method using moments of the received signal and the method based on the received signal envelope and channel statistics. These methods require sampling of the signal on a large scale, which increases the amount of computation and processing time. In this paper, we focus on spectrograms containing features of the K-factor and propose a novel its estimation method using convolutional neural network (CNN) from a single packet. Simulation results reveal its effectiveness in terms of estimation accuracy and resultant adaptive modulation performance.