Clustering Quantization Short-Time Energy Feature Extraction Method for MAC Protocol Identification in Non-cooperative UWANs

Abstract

The identification of the MAC protocol in non-cooperative underwater acoustic networks (UWANS) is of great significance in the field of underwater acoustic countermeasures, where feature extraction is one of the most important tasks. By taking into consideration UWANs characteristics such as long propagation delays, multipath effects, and non-Gaussian noise, this research provides a receiving signal model for UWANs. To effectively identify three common types of MAC protocol, including TDMA, ALOHA, and CSMA, we propose a feature extraction method called clustering quantization short-time energy (CQSTE). This method can clearly reflect the change of energy with time, resulting in a feature set more suitable for MAC protocol identification of non-cooperative UWANs. The received signal data set of UWANs is established in this research, from which the CQSTE is extracted and the feature set is produced. To validate our work, random forest (RF) and support vector machine (SVM) are utilized to identify the MAC protocol. The experimental findings demonstrate that the CQSTE and the RF classifier features are more suited for complicated underwater acoustic environments and can obtain good results in MAC protocol identification of non-cooperative UWANs.