Autonomic Security and Self-Protection based on Feature-Recognition with Virtual Neurons

The Internet and networks are not security-oriented by design so that myriad problems are compromising today's computer systems. This paper presented an autonomic security mechanism based on the virtual neurons and feature recognition. A prototype model of the virtual neuron is designed and the distributed virtual neurons are organized in a compound peer-to-peer and hierarchical structure. Then, the autonomic security mechanism is implemented via features recognized by the distributed virtual neurons. The paper presented how the feature recognition and virtual neurons work to automatically detect various security problems that are currently hard to defend against, including eavesdropping, replay, masquerading, spoofing, and DoS. A simulation system was developed and different cases were studied