Features

• Efficient Consensus（>100,000 TPS in one shard）： Trias rapidly locates the most trustworthy nodes to enforce the consensus logic. This allows only a small set up node to participate in making consensus, and therefore significantly increase consensus speed;

• High Stability（Resistant to near 90% malicious collaborative）： Based on HCGraph, Trias builds a small-world network between nodes. This network ensures that even if 90% nodes are removed from the network, it is still connected, which means there still exist at least an attestation path between any two remaining nodes;

• TEE-based Graph Computation： HCGraph’s gossip protocols facilitates a small world network among the attesting nodes. It significantly increases any single consensus code’s cost to “lie”, and effectively reduces redundant calls to TEE capabilities. This will greatly increase the consensus efficiency while defending against high-degree malicious collusive.

• Heterogeneous TEE Technology： A single TEE technology would be centralized controlled by technology providers. For example, Intel SGX has a strong dependency on Intel’s centralised on-line verification service. HCGraph supports the attestations among different TEE technologies and creates a web-of-trust to constrain any single type of TEE-based system’s behaviour. It thus eliminates the single-point-of-failure in TEE technology.

• Heterogeneous TEE hosts： Remote attestation protocols cannot identify hardware attacks against TEE. This provides an opportunity for malicious-mining nodes to destroy the network by attacking their one TEE hardware. HCGraph considers the heterogeneity of the selected nodes' environmental characteristics, such as physical locations, affiliations etc. This will greatly increase the difficulties for malicious organisations to destroy all participating consensus nodes’s TEE hardware.

• Optimized Gossip Protocol： HCGraph utilizes a gossip about gossip protocol similar to the HashGraph to achieve efficient trustworthiness propagation. HCGraph further introduces gossip about reduced gossip and targeted Gossip to optimize the transfer traffic.

• Dynamic Election Based on Trust Relationship： HCGraph constructs the global ‘Conspiracy Breaching’ model according to the trust relationship among mutually attesting node. This embodies a real-time coded node selection in accordance to the nodes’ past behaviours. It eliminates the human factors as in the DPOS, while preserving high throughput and low latency.