PeerAmbush: Multi-Layer Perceptron to Detect Peer-to-Peer Botnet

Due to emerging internet technologies that mostly depend on the decentralization concept, such as cryptocurrencies, cyber attackers also use concept develop P2P botnets. botnets are considered one of most serious and challenging threats infrastructure security. Consequently, several open issues still need be addressed, improving botnet intrusion detection systems, because is essentially a confrontational problem. This paper presents PeerAmbush, novel approach for detecting using, first time, effective deep learning techniques, which Multi-Layer Perceptron, with certain parameter settings detect this type botnet, unlike current research, entirely based machine techniques. The reason employing learning/deep besides data analysis, bots under same have symmetrical behavior, makes them recognizable compared benign network traffic. PeerAmbush takes challenge fewer selected features existing related works by proposing feature engineering method Best First Union (BFU). proposed showed considerable results, very high accuracy 99.9%, no FPR. experimental results promising approach, we look forward building it better security defenses.