Research On Heterogeneous Resource Sharing For Edge Computing

With the breathtaking development of interactive Internet technologies and services,the cloud computing paradigm arises at this historic moment.Due to the influences of transmission delays,communication bandwidths,and other factors,the centralized cloud platform with strong computing power and storage capability cannot respond well to the demand of massive Internet of Things(Io T)terminals requesting computation,storage,and data resources.Bridging the gap between cloud platforms and terminal devices,edge servers provide efficient,accessible,and reliable services for local Io T sensors,within an advantageously close distance to Io T devices,and reasonably schedule edge services and data resources.However,the heterogeneity of edge resources,the disorder of resource sharing behavior,the unpredictable user demands,and the lack of proper incentive mechanisms make edge resource sharing extremely difficult,facing challenges such as low resource utilization,serious security risks,and reluctant sharing will.In light of this,this dissertation aims to overcome the shortcomings of state-of-the-art edge resource sharing and thereby establish a cloud-edge-device collaborative heterogeneous resource sharing architecture,incorporating the reliable heterogeneous edge resource sharing framework,to-business service resource sharing scheme,and to-consumer data resource sharing scheme,to provide complete,reliable,efficient,and proactive edge resource sharing technology solutions.The concrete research contents are as follows,(1)A reliable heterogeneous edge resource sharing framework is established.Due to the anonymity security,public transparency,and pervasive liquidity advantages,blockchain smart contract technology is employed to design a virtual resource coin,serving as a medium of resource exchanges,to solidly support heterogeneous service resource sharing,e.g.,edge computing and edge caching,and heterogeneous data resources sharing,e.g.,Mobile Crowd Sensing(MCS)and user-generated content(UGC).Based on the historical behavior information of resource interactions between users,this dissertation studies the trust management mechanism for heterogeneous edge resource sharing and exploits deep reinforcement learning intrusion detection algorithm to minimize the impact of malicious nodes on heterogeneous edge resource sharing.The proposed architecture and algorithm solve the resource barriers between different stakeholders,screen the heterogeneity between service resources and between data resources,address the problem of trust management and the lack of a medium of exchange in resource sharing,and provide a reliable architecture for all stakeholders to realize heterogeneous resource sharing.(2)A to-business service resource sharing scheme is designed.Considering the two typical edge computing services,i.e.,computation offloading and edge caching,a resource density based differential edge service resource pricing strategy is proposed to provide the underlying decision support for the cloud computing data center based Internet service enterprise to purchase edge service resources,improving the utilization of heterogeneous edge service resources.In addition,the edge task offloading strategy based on the Stackelberg game is proposed for computation offloading and edge caching services,which also provides decision support for Internet service enterprises to achieve computation offloading and content caching service enhancement,and improves the user experience of edge computing services.(3)A to-consumer data resource sharing scheme is designed.Focusing on MCS services driven by the demand side,this dissertation proposes a differential task assignment scheme based on user attributes and task attributes.The Gaussian mixture model is employed to cluster the participants for MCS tasks,and the optimal participants are recruited for task publishers.Considering UGC services driven by the supply side and MCS services driven by the demand side,a double auction based incentive data resource pricing strategy is proposed to provide decision-making support for UGC producers to price their contents,boost the enthusiasm of mobile users in UGC producing,provide the underlying support for mobile users to sell MCS data,reward and encourage mobile users to actively participate in data resource sharing.