Research On Collective Computing Architecture And Anonymous Incentive Technology

Various smart wearable devices and human-computer interaction devices have connected "people" to the Internet,and the Internet of Things has connected everything to the Internet.The Internet has already connected all kinds of software and hardware resources and massive data resources in the world.In this way,the interconnection of people,machines and things is truly realized.In 2016,Professor GREGORY D.ABOWD of the Georgia Institute of Technology in the United States proposed the concept of collective computing and regarded it as the fourth-computing-generation after Ubiquitous Computing.It is the extension and generalization of ubiquitous computing.By connecting people's abilities to the collective computing environment to enhance the computing ability,a distributed collaborative computing environment that is ubiquitous,omnipotent,highly intelligent,super-large-scale,and multi-participants is formed,providing humans with high-efficiency,Highly intelligent,low-cost computing services.Therefore,as an emerging research,collective computing has been paid extensive attention by domestic and foreign researchers once it was proposed,and brought many new open issues and challenges.This dissertation focuses on the key issue of the participation and participation rate of resources,especially which of "humanware",studies the basic theories of collective computing tasks and resource models,architecture,security,and incentive mechanisms,which are used to solve a series of problems of resource discovery,resource participation and use,resource protection and incentive respectively.The main contributions include:First,in order to clarify the characteristics and functions of tasks and resources in collective computing,by analyzing tasks and resources in ubiquitous computing generations,combining the characteristics of collective computing to classify and model various types of tasks and resources,In particular,an exploratory and in-depth study of humanware is presented,and the model and formal description of humanware resource model are proposed.Based on the human resource model,models of hardware,software,objects,and data resources are designed.These models ensure that various heterogeneous collective computing resources are discovered,scheduled,selected,and expanded by the system in a unified form.Based on the analysis of resource functions and collective computing applications,the task model for collective computing is designed to clarify the basic requirements and constraints of collective computing tasks.After participating humanware to collective computing,the existing ubiquitous computing architecture can no longer meet the requirements of collective computing.To address this issue,ELCC architecture(Extendable Layered Architecture for Collective Computing)is proposed for fairly and evenly dispatch distributed heterogeneous resources to efficiently execute large-scale,high-concurrency heterogeneous computing tasks.The ELCC architecture can be used to access or directly deploy ubiquitous computing or its generalized systems.At the same time,the architecture takes into account the expansion of custom components and functions,and supports distributed and large-scale resource and task scheduling.In order to ensure that various resources can be used uniformly by the system,the first collective computing prototype system was designed and implemented based on the ELCC architecture,and the rationality,practicability and efficiency of the ELCC architecture and the prototype system are analyzed and verified.The prototype system provides an experimental environment for the collective computing research,and provides an experimental verification platform for the follow-up research work of this dissertation.Humanwares have strong subjective,purpose and self-protection awareness.Therefore,privacy protection is a prerequisite for various resources to participate in collective computing stably,and incentives are the purpose of sharing their computing abilities.Providing privacy protection with incentive is the core of ensuring the participation rate of collective computing resources.Most of the existing privacy protection approaches are a compromise between data quality and privacy,and few studies consider both privacy protection and incentive mechanisms.This dissertation proposes ADR(anonymous data reporting protocol with ensure reverse auction incentive mechanism),which uses a verifiable shuffling mechanism to disrupt the transfer plan,and then uses bulk transfer to submit data according to the transfer plan.In this way,the relationship between the data and the identity of the data contributor is broken.The combination of a one-time pseudonym mechanism and a blind signature mechanism guarantees the anonymous signature of the token.This mechanism ensures that the token is unique,unforgeable,and untraceable.This token is bound to the identity of the data contributors which is used to exchange incentives.Finally,in order to pay dynamic pricing incentives to encourage various resources to provide higher quality data,this dissertation proposes OADR(optimized anonymous data reporting protocol with ensuring dynamic incentives).First,the bulk transfer is improved,which simplifies the generation and quantity of pseudorandoms.Using random slots selection in improved bulk transfer to establish a transfer plan greatly improves efficiency and reduces energy consumption,making it more suitable for the requirements of collective computing low-latency applications.And it uses the same strategy as the data submisstion cuts off the relationship between the incentive token and the identity of the data contributor,combined with the double blind signature mechanism to provide a dynamic price and ensure the anonymity of the incentive exchange process.