Research On The Improvement And Application Of Verified Computation Protocol

Outsourced computation enables customers with limited computational resources to outsource large-scale computational tasks to the service provider. For instance, in cloud computing, clients outsource expensive computational tasks to the cloud. However, clients face some security issue like they cannot make sure whether the server executed the program correctly. Using verified computation protocols is an efficient way to check the correctness of the returned output of the outsourced computation. In recent years, with the tremendous growth of cloud computing and peer-to-peer computing, a considerable amount of research was devoted to verified computation protocols. However, the existing verified computation protocols show obviously different applicability and performance when execute different types of computational tasks, it cannot achieve general applicability and satisfying cost when multiple types of computational tasks are submitted. This limits the widely use of verified computation protocols. A verified computation protocol model with analysis and prediction module was proposed in this paper to improve the applicability and overall performance.Firstly, relative theory of verified computation protocol was deeply studied in order to analyze the applicability and cost of representative verified computation protocols. Secondly, a verified computation protocol model with analysis and prediction module was proposed in this paper based on the original protocol model structure. The introduced analysis and prediction module which includes computational type analysis and cost analysis can choose the more appropriate protocol to execute the computational task. The proposed protocol model was applied into practical verified computation protocols after studying the theories of the two protocols. Lastly, the proposed protocol model was realized in experimental system, the results of executing actual computational tasks showed satisfying improvement of applicability and overall performance compared with the original protocols.