Estimation of the success probability of a malicious attacker on blockchain-based edge network

Modeling highly accurate cyber-attacks on blockchain-based edge networks may be almost impossible in real-world circumstances due to unanticipated adversary (malicious) behavior. In this study, we propose a novel, distributed blockchain-based security architecture that today's industrial edge-enabled IoT systems may use to strengthen their defences against cyberattacks. We develop a probabilistic model to estimate the success probability of a malicious attacker on blockchain-based edge network by considering (i) hardware -level attack, (ii) network-level attack (IoT, Edge), (iii) software-level attack, wallet, smart contract, and (iv) blockchain network-level attack. We analyze the proposed attack model for sixteen different cyberattacks: False Data Injection, Sybil, DDoS, Identity Spoofing, Side-Channel, Botnet, Backdoor Trojan, Targeted Code Injection, Social Engineering, Phishing, Sinkhole, Man-in-the-middle, SQL Injection, Consensus, Eclipse, and Block Mining. Our model considers scenarios where private keys are stolen, lost, or forgotten by certain nodes. To determine the probability that a malicious attacker will be successful, we develop a simulation environment (a blockchain-based edge network with 200 total nodes, which generates 47,540 samples). We then estimate the success probability of a malicious attacker based on the blockchain resiliency (fault-tolerance) provided by the consensus algorithm, attack types, attack location, and the network size for various scenarios. Our results demonstrate that blockchain-based edge networks are more vulnerable to malicious attacks based on: (i) cyber-attack types (90.96% for Botnet and Backdoor Trojan attacks higher than for other attacks, such as DDoS, SQL Injection and Sybil); (ii) attack location (90.18% for software-level attacks higher than hardware, network, and blockchain network-level attacks); and (iii) consensus algorithm (68.85% for Byzantine fault tolerance, BFT, higher than the Proof-of-Work, PoW). The IoT network vulnerability factor, or exposure factor, of cyberattacks, depends on the strategic importance of the application to the attacker. Our results should be validated in real-world experiments with a large number of nodes.