| Edge computing utilizes devices which close to the data source for distributed computing and provides service for user.Since its advantages of saving bandwidth and low latency,it is widely concerned by academia and industry.However,edge devices are usually distributed at the edge of networks and lack of unified management,they may be untrusted.Therefore,the computing data are likely to be leaked and the leaking is not easy to be found.Thus,how to achieve efficient and secure edge computing has become a hot research spot.A large number of studies show that linear coding has advantages in ensuring data security.Previous studies have proposed an effective linear coding scheme to resist the independent attack of edge devices,achieve the confidentiality of computing data,and minimize the total cost of resource consumed by edge devices.On this basis,we use linear coding technology,under the cooperative attack of edge devices,take matrix multiplication as an example,and design an effective coded edge computing scheme.The designs aim to:(1)complete the computing tasks,(2)ensure the confidentiality of computing data,(3)reduce the total cost of resource consumed by edge devices,including computing,communication,and storage costs.The details are as follows:In this paper,we study the secure matrix multiplication problem in homogeneous edge computing.Firstly,we elaborate the research significance of the problem.Secondly,we present a homogeneous edge collaborative computing model,and define the attack model and security requirements.Then,the necessary conditions for the existence of feasible solutions are analyzed.Next,we design a secure coded computing scheme composed of storage scheme and coding design,and give the optimal parameter selection algorithm to determine the optimal values of the parameters in the proposed scheme,so as to achieve a smaller total cost.Then,we analyze the decodeability,security and feasibility of the proposed scheme.Finally,we verify the effectiveness of the proposed scheme through a large number of experiments.In this paper,we study the secure matrix multiplication problem in heterogeneous edge computing.Firstly,we elaborate the research significance of the problem.Secondly,we present a heterogeneous edge collaborative computing model,and define the attack model and security requirements.Then,the necessary conditions for the existence of feasible solutions and the range of related parameters are analyzed.Next,we design a minimum cost scheme composed of coding design and task allocation,and give a task allocation algorithm to determine the optimal values of the parameters in the proposed scheme,so as to achieve the minimum total cost.Then,we analyze the decodeability,security and feasibility of the proposed scheme and proves that the solution returned by the proposed scheme is the optimal solution of the problem.Finally,we verify the effectiveness of the proposed scheme through a large number of experiments. |
