Abstract: Fire Wrap, a network based cyber defence system proposal which detect network anomalies by analysing the pattern of an incoming attack and distinguishes the attacker from the existing networked machines using the Boltzmann machine learning algorithm, then re-routes the incoming signal using double tunnelling approach to a sandbox environment where, the exit node vulnerability of onion routing is exploited to extract the raw data. The attacker will execute the attack in this sandbox environment and we can analyse the behaviour of the attack virtually without affecting the original network system, there by obtaining vital information which will help in forensic studies. The analysis part of Fire Wrap is carried out through Hopfield neural network, which is a simple recurrent network that can work as an efficient associative memory and can store and analysis data in a manner similar to the brain. The system is far advanced than any currently used firewalls and is developed to protect the surface web users from the cybercriminals which use the tor anonymous network to launch cyber-attacks from their hotspots in the dark web, which will steal all the confidential data and causes fatal damage to users including the defence network of a country. We are formulating this system with a provision of planning a counter attack which will make the system future proof.
Keywords: Tunnelling, dark web, neural network, Boltzmann, sandboxing