DeepBlock: a Collaborative Intrusion Detection Framework Based on Blockchain and Deep Learning

With the recent proliferation of the Internet, network security has gained considerable research attention. Collaborative Intrusion Detection Systems (CIDS) have been recognized as essential mechanisms for protecting networks against malicious attacks. A variety of approaches have been developed for CIDS, but they still face two critical issues data and trust management; because not all network devices are trusted and secured. This paper proposes a novel framework, DeepBlock, for CIDS using machine learning and blockchain technology. Machine learning is incorporated for the detection of both novel attacks (through fuzzy clustering) and known attacks (through Long Short-Term Memory LSTM). Blockchain smart contract is a vital part of the blockchain and is a program that can be jointly executed by a network of untrusted nodes. Motivated by this, the rules extracted from the machine learning sub-model will be encoded in the smart contract and distributed among the blockchain nodes. The security and immutability of blockchain will ensure the trust and confidentiality of data.