Enhanced Wireless Technology Recognition Method Using Energy-Efficient Spiking Neural Networks
Lifan Hu, Yu Wang, Xue Gang Fu, Hao Huang, Shengnan Shi, Lantu Guo, Yun Lin, Guan Gui · 2024
Wireless Technology Recognition (WTR) has emerged as a promising solution to counteract the degradation of communication quality caused by various technological interferences. It involves distinguishing different wireless technologies by analyzing their characteristic features extracted from radio signals. While Deep Neural Networks (DNNs) have been extensively employed in WTR due to their robust ability to extract hidden data features and make classification decisions, their application is often limited by excessive power consumption. We propose a novel WTR method utilizing the Spiking Neural Networks (SNN) framework to address this challenge. Experimental results demonstrate that our proposed model achieves commendable performance at high signal-to-noise ratios on open-source datasets. Notably, it retains the inherent low power consumption characteristic of SNNs while also being competitive in recognition accuracy. The proposed SNN-based WTR method is assessed on a large-scale real-world dataset, confirming its efficacy in maintaining low power consumption while offering competitive recognition accuracy.