How can AI improve consensus in Blockchain?

I do not intend to explain these proof-of-X mechanisms in this article, as there are already many well-written tutorials online. Instead, I would like to take a closer look at the intriguing question mentioned in the title.

Before investigating our main question, let’s reach a consensus about what a good consensus algorithm is!

A Google Scholar search for the term “comparison of consensus algorithms” returns many research papers. Among them, [1] has included a very informative table categorizing evaluation metrics based on three dimensions: Effectiveness, Decentralization, and Security.

The truth is metrics of all the above dimensions matter. As an example, Bitcoin has proven to be both decentralized and secure. Still, its lack of effectiveness due to its high energy consumption has sparked intense debates around its overall sustainability as a cryptocurrency.

There might be an inherent trade-off here, similar to many other multi-dimensional engineering problems, that prevents achieving effectiveness, decentralization, and security simultaneously (if you happen to know about any, please let me know).

However, if an AI solution can improve one dimension while maintaining the others, I will certainly appreciate it (Won’t you?). For instance, blockchains may become more widely accepted if their effectiveness metrics (such as throughput) are boosted without sacrificing decentralization or security. So, let’s explore the research space of this mindset.

AI models (and deep learning in particular) are uniquely able to discover patterns in massive amounts of data of any kind, so why not use them to:

After reviewing research papers with both AI and blockchain keywords, I found some critical points:

If I want to name one industry — other than Fintech — where blockchain will flourish, I name the IoT industry (maybe I am a bit biased toward networking!). Simply put, blockchain technology can provide a secure platform for collaboration among things.

However, the IoT has some special considerations to be met, one being intolerance of high delays as IoT needs a rapid consensus. Another issue is the high probability of IoT devices being compromised (because of their minimal security measures and ease of access to outsiders), making the security dimension much more critical.

In this vein, a team of researchers at the University of Central Florida has proposed “A new framework for a secure and robust consensus in blockchain-based IoT networks [3]”.

Their proposal is rather complex, but in short, they have added an outlier-detection learning mechanism at the front door of their consensus mechanism. As can be seen in the figure, both uncompromised and compromised data enter the consensus algorithm, but the outlier detection only accepts the healthy data (although there will be false positives and false negatives, too).

The Proof of Artificial Intelligence (PoAI) consensus mechanism [4] is another research proposal worth mentioning. This mechanism uses a neural network to select the super nodes (i.e. nodes with excessive computational power) and random nodes for mining transactions.

I think their proposal is somewhat ambiguous (although the paper has nearly 30 citations), but the reason for introducing this paper was to show the vast potential of AI in selecting nodes for voting based on their history, the graph of connections, etc.

AI-powered consensus algorithms appear to be a relatively new area of research, and they have enormous potential to improve blockchains in many ways, including:

Just keep in mind that we are at the beginning of the blockchain-native era, which means any advancement in consensus algorithms will be crucial to achieving a 20-billion-dollar market by 2025.

[1] Fu, Xiang, Huaimin Wang, and Peichang Shi. “A survey of Blockchain consensus algorithms: mechanism, design and applications.” Science China Information Sciences 64.2 (2021): 1–15.

[2] Ynag, Qiinglin, et al. “Fusing Blockchain and AI with Metaverse: A Survey.” arXiv preprint arXiv:2201.03201 (2022).

[3] Salimitari, Mehrdad, Mohsen Joneidi, and Mainak Chatterjee. “AI-enabled blockchain: An outlier-aware consensus protocol for blockchain-based IoT networks.” 2019 IEEE Global Communications Conference (GLOBECOM). IEEE, 2019.

[4] Chen, Jianwen, et al. “An AI-based super nodes selection algorithm in blockchain networks.” arXiv preprint arXiv:1808.00216 (2018).

Related posts:

When it comes to selecting NetSuite consulting services, it’s crucial that you hire professionals with the necessary technical and theoretical knowledge. ERP, CRM, and database solutions require…

I wait. I deepen in the room. Fed lions, glowing, congregate In corners, sleep and fade. For whom It may concern I, tawny, wait. Time flowing through the window; day Spilling on the board its bright…