Domino – A Highly Scalable Blockchain Using AI for Mass Adoption

 Domino – A Highly Scalable Blockchain Using AI for Mass Adoption

It is widely recognized blockchain technology faces the trilemma of scalability, security, and decentralization. Developers are forced to make trade-offs. Bitcoin network trades scalability for security and decentralization to process about 7 transactions per second (TPS). Ethereum makes a similar trade-off to process about 15 TPS. Several projects can achieve 1000-3000 TPS by reducing the degree of decentralization. In comparison, a centralized network such as Visa can process over 25,000 TPS. To gain mass adoption and compete with centralized networks, blockchain needs to scale.

Many projects attempt to achieve scalability by sharding including Ethereum 2.0. Sharding requires partitioning of the network into multiple shards and each shard will achieve consensus on its own. By definition, sharding reduces the number of participating nodes and potentially increases the chances of a 51% attack.

A major reason that blockchain can not scale is transactions are received in a different order by different nodes. It is time-consuming for nodes to agree on the order. Bitcoin and Ethereum force mining nodes to use a lot of electricity to find solutions to a hard computational puzzle. Whoever solves the puzzle first gets to decide on the order of the transactions and create the next block in the blockchain.

Domino Blockchain uses the following approaches to solve the seemingly insolvable trilemma and create a blockchain that is scalable, secure, permissionless, and with unlimited participation and decentralization.

  1. Domino creates a cryptographic hash order for all received transactions. Transaction ordering is no longer determined by the block producers. Instead it is determined by the cryptographic hash algorithm and persisted to the blockchain. We call this Proof of Time (PoT). This is conceptually similar to Solana’s Proof of History (PoH, Reference 1), but with important differences in architecture and implementation.
  2. Domino uses a special Proof of Stake (PoS) algorithm called Fast Byzantine Agreement (FBA, Reference 2) as consensus for block production and validation to achieve transaction finality in 3.5 seconds.
  3. Domino introduces a hybrid validation model to enhance security while maintaining unlimited participation and decentralization. We introduce a new decentralized firewall, composed of mobile devices and laptops around the world, used to endorse transactions and surveillance block production and validation. Any device can join the network to perform these functions and get rewarded. Endorsement of transactions is done by sampling random nodes and obtaining sufficient staking. This process can greatly reduce the number of erroneous transactions. An unlimited amount of nodes can be used to perform this function (we call it surveillance nodes or decentralized firewall). Block production and validation is performed by a set of high-bandwidth and high-performance commercial-grade servers (we call them core nodes). The core nodes are distributed all over the world. Surveillance nodes are only involved with surveillance of the blockchain and endorsement of transactions. They do not produce blocks. By separating blockchain surveillance from block production/validation we enable unlimited decentralization, extensive participation, and enhanced security through surveillance. Any mobile device and computer can join the network without permission and hardware restrictions.
  4. Introduce a new consensus algorithm (Domino Consensus) that can quickly identify fraudulent nodes from a large number of surveillance nodes.
  5. Introduce a decentralized storage as part of the blockchain to offload large data or old data. This is particularly useful to store NFT image data with high resolution.

Domino surveillance nodes use AI technology to detect error patterns such as:

  • Double spending
  • Over spending
  • Dust attack or denial of service attack
  • Long range attack
  • Forged signature
  • Misbehaving block producers and validators
  • Parallel chains 

Any misbehavior will be marked and propagated to the whole network. Any producer or validator confirmed as the misbehaving node will be removed from the blockchain network and staked token will be slashed. Solana network was attacked by denial of service recently and the network was down for more than 12 hours. Domino blockchain leverages surveillance nodes and AI technology to detect such attacks and eliminate the attackers from the network at a very early stage to avoid significant damages or downtime to the network.

The world’s fastest blockchain deployed to date is the Solana network with a throughput of 50,000 TPS on 200 nodes with a block time of 400 ms. Domino blockchain prototype performs at 100,000 TPS on 360 nodes with a block time of 200 ms. Solana uses a variation of Practical Byzantine Fault Tolerance (PBFT) called Tower Consensus. Its throughput (TPS) may reduce when more nodes join the network, as in PBFT network traffic increases quadratically as the number of nodes increases. Domino utilizes a modified

FBA that can perform at the same TPS even when the number of nodes increases to 100,000. The scalability is achieved without relying on any layer 2 solutions or sharding. Domino will use the super-fast blockchain to build an ecosystem including a global peer-to-peer payment system, a decentralized exchange, Defi, NFT application or marketplace, games, and many other decentralized applications. For more information please email jack@dominochain.com.

About Author


Jack Ding is the Founder and CTO of Dominochain. This uniquely designed blockchain solves one of the biggest problems in crypto and blockchain space – scalability without compromising security and decentralization. He has 23 years of experience designing high-performance distributed software solutions including Nasdaq market real-time surveillance systems and webMethods B2B integration middleware. He also worked as a senior architect at the World Bank. Jack Ding received his Ph.D. degree from Dartmouth College and his M.S. degree from Johns Hopkins University.


References:

  1. Anatoly Yakovenko, Solana whitepaper, https://solana.com/solana-whitepaper.pdf
  2. Jing Chen, Sergey Gorbunov, Silvio Micali, Georgios Vlachos, Algorand Agreement, Super Fast and Partition Resilient Byzantine Agreement, https://eprint.iacr.org/2018/377.pdf

AITECHNEWS

Related post