NeuroNet: Towards an Intelligent Internet Infrastructure

Malicious attacks against Internet infrastructure are one of the most damaging threats to modern society. Due to the inter-dependence between networks, attackers can paralyze or isolate the victim network without attacking it directly. Such kind of attacks could be a main weapon of cyber war in the foreseeable future, potentially detrimental to many national interests. Today´s network security solutions designed under the end-to-end paradigm cannot address the malicious activities inside the core network effectively. This paper proposed to strengthen the network infrastructure by developing NeuroNet, a network neural system that plays a critical role similar to the role of the nervous system in the human body. A distributed information collection and processing mechanism is proposed, which coordinates the activities of core network devices, monitor for anomalies, construct alerts, and initiate countermeasures. Through intensive experiment of a distributed detection scheme against the low-rate TCP-targeted DDoS attacks, the effectiveness of the NeuroNet architecture is verified.