Trusted Outsourced Computing Framework for Smart Contract of Permissioned Blockchain

Abstract

Permissioned blockchain can provide a decentralized, tamper-evident and trusted computing environment and has been widely deployed in e-government, finance and supply chain management. Smart contracts need to be developed using a specific programming language. Limited by the architectural design of blockchain systems, smart contracts still face many challenges in performing complex computing tasks. For example, there are many matrix operations in the inference process of deep neural network models. Loading deep neural network models directly in smart contracts not only requires relatively large development costs, the inference process of deep neural network models generally consumes more computing resources. We use Intel SGX to design a flexible architecture that can outsource complex computing tasks that need to be processed in smart contracts to off-chain computing servers, while being able to use Freivalds’ Algorithm to ensure the correctness of the outsourced computing results.