Two-tier Multi-zone Consensus: Enable Intelligence Sharing for AIoT with Enhanced Security

With the advent of intelligence-of-everything in 6G, the explosion of Internet of Things (IoT) devices will bring massive volumes of data. Though injecting artificial intelligence (AI) technologies into IoT, namely AI of things (AIoT), can release the data value, it still faces many challenges, such as poor efficiency, lack of trust, etc. As a promising technology with decentralization, scalability, and security, leveraging blockchain in AIoT has attracted widespread attention in recent years. However, isolatability and scalability are two critical issues that hinder blockchain from building efficient, secure, and trusted intelligence-sharing platforms for AIoT. Blockchain interoperability addresses isolatability between heterogeneous blockchains, while sharding-based blockchain achieves scalability by solving isolatability of homogeneous blockchains running in different shards. To ensure atomicity between different blockchains, both blockchain interoperability and sharding-based blockchain need to enhance the resistance to double-spending attacks of the blockchain to which the cross-chain transaction originator belongs. To this end, this work presents a two-tier multi-zone consensus, where Consensus Zone in tier-1 features blockchain interoperability and sharding-based blockchain by allowing different blockchains run in different zones, while Coordination Layer in tier-2 is to enhance each zone’s security by allowing them to interoperate as a Direct Acyclic Graph (DAG) in a distributed and parallel manner. Then, stochastic models are used to capture the consensus process and analyze the probability of successful double-spending attacks. From the numerical results, this consensus is proved to be with enhanced security.