SDR for Physical Layer Authentication

• Autor(a) principal: Faria, João Francisco Lopes
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/152400
• topic: PHY; Authentication; Fingerprinting; 802.15.4; Zigbee; SDR; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Wireless networks and devices are easy and useful solutions nowadays, regardless of the context in which they are implemented. However, it is in the broadcast nature of wireless networks that some vulnerabilities arise. To protect against these vulnerabilities, encryp- tion and authentication methods are commonly used. However, such methods come at the expense of their own complexity, requiring high enough computational power to solve, and introducing latency. To try to reduce the complexity of the conventional ways of user authentication, this work has studied mechanisms to implement reliable authentication at the physical layer, analyzing the various devices signal characteristics. To achieve this analysis, the GNU Radio platform was used to process incoming signals and extract the necessary features. Given the open source nature of GNU Radio, this provides a customiz- able and low-cost solution to signal processing and feature extraction. This research uses the GNU Radio to implement a feature extraction solution and constructs a feature vector with size 1 × 95. This thesis studies the extracted features of eleven IEEE 802.15.4 devices in regards to their separability and proposes a solution for feature reduction. The feature vectors are passed through a Random Forest and a Deep Neural Network (DNN) classifier, achieving accuracies as high as 99% for short distance communication.