A Probabilistic and Distributed Validation Framework Based on Blockchain for Artificial Intelligence of Things

Abstract

Artificial Intelligence of Things (AIoT) applications have advanced rapidly. However, most of them are inherently vulnerable to security threats and may be the source of spoofing attacks, and meanwhile, in AIoT systems, the transfer of real-time data from terminals and the cloud strains network bandwidth. To defend against attacks, save network forwarding resources, and relieve authentication pressure on the receiver end, it is essential to verify source identity of AIoT terminals on the forwarding path. In this paper, we propose PDIV, a probabilistic and distributed identity validation solution for AIoT applications. In the framework of PDIV, honest forwarders can verify the authenticity of the source identity of packets with a certain probability and filter spoofed packets to prevent them from spreading, reduce end-to-end network latency, and increase throughput as much as possible. Additionally, PDIV, a blockchain-based system that uses Merkle Patricia Trie (MPT) on the blockchain, makes it practical and efficient to realize distributed storage and verification of identity information. Moreover, we theorize about the trade-off between PDIV network performance and detection effectiveness, as well as how PDIV enables defenses against different attacks like spoofing and DDoS attacks. Furthermore, we implement PDIV on Network Simulator Version 3 (NS3) and evaluate its performance. The simulation results demonstrate that PDIV can prevent the spread of spoofed packets effectively and PDIV works better than currently available blockchain-based public key infrastructure (PKI) approaches in terms of network latency.