Title: A novel skin-inspired model for intelligent object recognition in sensor networks
Authors: Aaron Rasheed Rababaah
Addresses: College of Engineering and Applied Sciences, American University of Kuwait, Salmia, Kuwait
Abstract: Extensive research work has been conducted in the area of wireless sensor networks with high concentration on target detection or/and tracking. Although, there are some works that are dedicated to target classification, but the considered sensor modalities require inexpensive hardware and complex algorithms such as video, audio, radar, infrared, etc. Our solution to this problem is to propose a novel concept inspired by the human skin. It is well-known that skin possesses simple sensing receptors, compared to sophisticated ones such as vision, through which humans cannot only detect a stimulus but can identify its type such as a needle, glass, table, ball, etc. analogues to this biological behaviour, we propose a stimulus data modelling, characterisation and classification in a simulated sensor network. The technical development of the proposed model is presented and validated via training a convolution neural network and was found to be effective and promising for future extensions.
Keywords: skin-inspired model; object recognition; object classification; intrusion detection; sensor networks; deep neural networks; convolution neural networks.
DOI: 10.1504/IJSNET.2022.123590
International Journal of Sensor Networks, 2022 Vol.39 No.2, pp.93 - 105
Received: 28 Jul 2021
Accepted: 01 Aug 2021
Published online: 29 Jun 2022 *