5G-FOG: Freezing of Gait Identification in Multi-Class Softmax Neural Network Exploiting 5G Spectrum

Khan, J. S., Tahir, A., Ahmad, J., Shah, S. A. , Abbasi, Q. H. , Russell, G. and Buchanan, W. (2020) 5G-FOG: Freezing of Gait Identification in Multi-Class Softmax Neural Network Exploiting 5G Spectrum. In: Computing Conference 2020, London, UK, 16-17 Jul 2020, pp. 26-36. ISBN 9783030522421 (doi: 10.1007/978-3-030-52243-8_3)

Abstract

Freezing of gait (FOG) is one of the most incapacitating and disconcerting symptom in Parkinson’s disease (PD). FOG is the result of neural control disorder and motor impairments, which severely impedes forward locomotion. This paper presents the exploitation of 5G spectrum operating at 4.8 GHz (a potential Chinese frequency band for Internet of Things) to detect the freezing episodes experienced by PD patients. The core idea is to utilize wireless devices such as network interface card (NIC), radio frequency (RF) signal generator and dipole antennas to extract the wireless channel characteristics containing the variances amplitude information that can be integrated into the 5G communication system. Five different human activities were performed including sitting on chair, slow-walk, fast-walk, voluntary stop and FOG episodes. A multi-class, multilayer full softmax neural network was trained on the obtained data for classification and performance evaluation of the proposed system. A high classification accuracy of 99.3% was achieved for the aforementioned activities, compared with the existing state-of-the-art detection systems.

• Item Type: Conference Proceedings
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Abbasi, Professor Qammer and Khan, Jan Sher and Shah, Mr Syed
• Authors: Khan, J. S., Tahir, A., Ahmad, J., Shah, S. A., Abbasi, Q. H., Russell, G., and Buchanan, W.
• College/School: College of Science and Engineering > School of Engineering > Electronics and Nanoscale Engineering
• ISSN: 2194-5357
• ISBN: 9783030522421
• Published Online: 04 July 2020
• Copyright Holders: Copyright © 2020 Springer Nature Switzerland AG
• First Published: First published in Intelligent Computing: Proceedings of the 2020 Computing Conference, Volume 3: 26-36
• Publisher Policy: Reproduced in accordance with the publisher copyright policy