During the last few years the blockchain technology was gradualy improved and eventually left the infancy stage. Now corporations, banks and other institutions find benefits in this technology and actively implement it in their solutions. While such network consensus algorithms as Proof-of-Work slowly fade away, Proof-of-Stake solutions methodically secure the majority in this industry.
IOST Blockchain is one of those networks which offers a variety of useful perks, starting from improved sharding, ending with disrupting consensus protocol called Proof-of-Believability. This network resolves well-known problems related to transaction throughput, scalability, and decentralization through such solutions as Efficient Distributed Sharding, TransEpoch, Micro State Blocks and other.
Consensus Algorithm
IOST aims at the development of a platform that is capable of providing decentralized services to software development, gaming, fintech, media, and entertainment industries.
The project has developed a platform that differs from existing Proof-of-Stake solutions that are working on the sharding principles. Mainly, the IOST uses an intra-shard Believable-First approach which implies division of all validators into two groups called a believable league and a normal league.
The first group processes transactions quickly, while the second group only ensures verifiability and provides finality. Believable league nodes are chosen based on a unique metric called believability score which is assigned based on the number of services provided by the node, contributions made, reputation and its behavior. In other words, this score is called Servi and it gets destroyed once a node validates a block. By implementing such a complex metric the network became more secure because it’s hard to forge all of those parameters artificially.
Specifics
Distributed Randomness Protocol
In order to avoid different types of network attacks, IOST proposes the Distributed Randomness Protocol (DRP) which allows generating an unforgeable, uniformly random value via publicly verifiable secret sharing (PVSS) and non-interactive zero-knowledge proof (NIZK). This protocol is used to unite nodes into shards, assign nodes into those shards and choose leaders in each shard.
The DRP mechanism is divided into two phases, the randomness generation, and randomness verification. First of all, a client broadcasts a message with a randomly generated balanced grouping, then servers generate a random input value and create shares for members of the same group via PVSS. Once encrypted shares are received by all servers, a client extracts a subset of inputs from each group. That’s how a client gets the output of the protocol and each group’s secret. During the second phase, servers send decrypted shares to a client, and the client interlinks recovered group secrets, that allow them to get the random output.
Leader Election with Back-up Protocol
Back-up Protocol aims at preventing malicious nodes to have behaviors that violate the correct network processes. The protocol pushes all nodes to choose one of two behaviors or to cooperate by running the DRP or pass this action, and in case a node passes the DRP run it gets excluded from the consensus process.
Inter-Shard Transactions
This functionality is realized through the usage of proprietary Atomix protocol which allows conducting atomic cross-shard transactions. It’s important to emphasize that the protocol is similar to Omniledger and designed in a way so that it prevents double-spending. Moreover, the Atomix allows smart-contracts as well even on the inter-shard transaction level.
