Research On Secure Outsourcing Algorithm Based On Edge Computing

With the development of Internet of Things(Io T)and 5G,edge computing,as a new computing paradigm,has been widely popularized in academia and industry.An important application of edge computing can allow the resource-constrained client to outsource the complicated computation task to the nearby edge servers.The user can reduce computational overhead and improve efficiency.Outsourcing computation in edge environment faces some security challenges.At first,it is hard to ensure all edge servers are fully trustworthy.Therefore,a secure outsourcing algorithm should protect the privacy of users' inputs and outputs.Then,the edge servers may return a random value to the user due to various reasons.The user should have ability to verify the correctness of results with high probability.In addition,an outsourcing algorithm based on multiple edge servers needs all edge servers to collaboratively work towards solving computation tasks.The user needs to ensure that multiple edge servers securely and correctly compute the results.Finally,the local computation complexity should be much lower than that by locally solving the original problem alone by the user.Otherwise,it is meaningless for the user to run the outsourcing algorithm.In this paper,for the time-consuming modular exponentiation in cryptographic algorithms,the least squares solution to the overdetermined system of linear equations and the convex optimization problem with equality constraints in scientific computation,we design the secure outsourcing algorithms based on edge computing,specifically including:(1)In the scenario of multiple edge servers,we propose two distributed and secure outsourcing algorithms for modular exponentiation.The first algorithm solves modular exponentiation with constant base and variable exponent.The second algorithm solves modular exponentiation with variable base and exponent.The two algorithms can protect the privacy of data by logically splitting base and exponent,and reduce the computational overhead of each edge server by splitting the binary form of exponent,and verify correctness of results by comparing the two outsourcing results of the same modular exponentiation.The experimental evaluations show that the two proposed algorithms are efficient on the user side and the server side.(2)In the scenario of multiple edge servers,we propose a distributed and secure outsourcing algorithm for solving the least squares solution to the overdetermined system of linear equations.In the algorithm,the coefficient matrix is divided into several blocks according to the rows.All edge servers obtain the correct solution through interactive computation.Random blinding technology is used to protect the privacy of input parameters and returned results.The mutual verification between the servers can ensure the correctness of results.The experimental evaluations show that the proposed algorithm outperforms the previous ones in terms of the computational overload on the edge server side and the efficiency on the user side.(3)In the scenario of single edge server,we propose a secure outsourcing algorithm for solving the convex optimization problems with equality constraints.The algorithm uses a series of sparse matrices to reduce the computational overhead on the user side and protect the private data.Meanwhile,we design an efficient algorithm to verify the correctness of results.The experimental evaluations show that the proposed algorithm is efficient on the user side and the server side.