Research On IoT Overlay Network For Cloud Manufacturing

The “Made in China 2025”plan clearly proposes to promote the further integration between informatization and industrializationthe,and regards intelligent manufacturing as the main focus of the integration.It is important to develop new manufacturing models such as personalized customization and cloud manufacturing based on the Internet.Therefore,researching the Internet of Things(IoT)technology for cloud manufacturing is one of the key issues in intelligent manufacturing.The current IoT platforms and protocols are wildly different,communication and management cannot be unified.Their compatible communication with the traditional Internet is a big problem.The overlay network built on the IoT through virtual links is a technical way to realize the integration of cloud manufacturing resources,but it is still lack of the effective methods to work on it.However,there are a few relevant researches on building an application-oriented virtual group entity on top of the IoT overlay network to develop Internet personalized custom solutions.This thesis focuses on the architecture of cloud manufacturing IoT overlay network around IoT resource access and compatible communication,overlay network resource organization,distributed indexing and virtual group entity personalization.A three-layer cloud manufacturing IoT overlay network architecture based on REST architecture is proposed.It designs a standardized resource adaptation layer to implement heterogeneous subnet convergence and crossprotocol communication,shields the differences among all kinds of physical device resource access.Device network access protocol and device control transmission protocol are proposed to support Internet compatible communication,then we formalize and verify these two protocols.A distributed multi-ring structural peer-to-peer network is designed to realize resources multi-attribute indexing service.Methods of constructing virtual group entities for process execution,concurrency control,and data sensing are also proposed to customize production process of cloud manufacturing.We research group communication mode of overlay network,improve group entity communication methods to adapt to cloud manufacturing needs.A heuristic iterative neural dynamic programming method is designed for concurrency control group to satisfy concurrent schedule.The main research work and innovations are summarized as follows:(1)Aiming at the problem of heterogeneous IoT subnet convergence and compatible communication,two protocols are proposed to achieve cross-protocol communication.A equipment communication conversion framework is designed to support the integration of different subnets which are unable to Implement the TCP/IP stack.Then the Constrained Application Protocol is deployed to form a unified IoT resource web access interface on the network application layer and shield differences between different embedded device platforms on the underlying layer.The formal verification of the proposed protocol proves the correctness and no deadlock.We implement it on the communication transformation of the ZigBee network to test the actual transformation effect.The experimental results verify the reliability of compatible communication,the communication delay is small,and CoAP protocol communication can be completed normally.(2)For the cloud manufacturing IoT resource index and lookup problem,a distributed structured peer-to-peer network of Chord multi-ring is proposed.It is constructed on the resource directory nodes which are defined in the draft proposed by the IETF.It forms a decentralized directory management of IoT device resources,and introduces a group voting mechanism on the resource description to form a group decision for resource attribute definition.The multiring method overcomes the problem that the distributed hash algorithm only supports singledimensional attribute query,can support the effective query of multi-dimensional attributes,combining the uniform locality preserving hashing function to solve the range search problem.Compared with the existing multi-ring resource query methods,the effect of resource query is further improved by dividing the primary-second attribute mode.(3)For the customize problem of IoT overlay network under cloud manufacturing,the related methods of constructing virtual group entities are proposed,and solutions for group entity communication and multiplex problems are also given.a.A formal definition of the cloud manufacturing process is proposed.Defines of sequential execution,concurrent execution,and data sensing processes based on the characteristics of cloud manufacturing process are provided.A virtual group entity construction method on IoT overlay network for multi-process single task is proposed.b.A communication framework for virtual group entity communication problems is designed.It can better support virtual group object instantiation,group resource registration in the Distributed Resources Index Service Framework,communication between Internet clients and group entities and improving communication compatibility for heterogeneous IoT networks.It further improve the subscription method on group communication to adapt to the process perception of task execution in cloud manufacturing.Compared with the existing group communication,it can support multi-node linkage in the group,and the data filtering method is more flexible and the communication efficiency is higher.c.A multiplexing method for group members long-term concurrent execution is proposed to form the virtualization of manufacturing service capabilities.The group entity multiplexing scheduling problem is regarded as the optimal scheduling problem of nonlinear dynamic systems.A new iterative neural dynamic programming group scheduling algorithm is designed by using heuristic dynamic programming model combined with backpropagation neural network.The proof of the convergence of such algorithm is given,and the performance of the group scheduling is tested.Experiments show that the algorithm can effectively improve the cost of group multiplexing,and provides a good reference model for how to virtualize manufacturing capabilities to support cloud manufacturing.These research provide an architecture reference model for the design of a new generation of IoT overlay networks,which provides a flexible network definition method for various applications in cloud manufacturing,and promotes the research on virtualization of devices and networks.These findings provide theoretical support and implementation for the development and operation of cloud manufacturing systems based on the Internet of Things and will be helpful for further research.