An Efficient Spoofing Attack Detection Using Deep Learning Based Physical Layer Security Technique

Abstract

Spoofing attack detection plays a crucial role in the defence field, involving critical and highly secured data processing. The accurate attack detection mechanism prevents unauthorised access to sensitive information, thereby protecting National security. Physical Layer Security (PLS) is a promising emerging technique that uses the wireless channel’s randomness to secure the communication network. The spoofing attack is one of the severe threats to the wireless network, where the attacker imitates the legitimate user to launch an attack against the network. This paper investigates the channel characteristics-based physical layer technique to detect spoofing attacks. For static radio environments, the two-sample independent hypothesis testing is used to identify the spoofing attack, showing an improvement in detection accuracy of 97 %. The attack detection problem is considered a Reinforcement Learning (RL) based classification problem for a challenging dynamic radio environment. It is simulated using the actor-critic-based Deep Reinforcement Learning (DRL) technique with the help of the Reformed Deep Deterministic Policy Gradient (Re-DDPG) algorithm. The simulated results show that the proposed method performs better than the existing strategies and achieves a Receiver Operating Characteristics (ROC) value of 0.96. The detection accuracy of the proposed method can reach up to 98 %, with precision and recall of about 98 % and 99 %, respectively.