| In recent years, as a key enabler of the Internet of Things (IOT), radio-frequencyidentification (RFID) has been under rapid development both in research and in industry.RFID is becoming increasingly prevalent and is being applied more widely and in-depth.RFID systems are growing into a new phase with not only small-scale, standaloneapplications, but also with the opportunities and challenges in networked systems, andfocus on scalability, diversity and intelligence. To follow this trend, academia and industrypay more attention to RFID networked system. But so far, academia RFID networks stillstay in theoretical research level, lack of practicality and performance considerations,while industry has proposed a number of RFID network solutions, lack of systematicness,universality and extensibility.In this paper, we design a novel RFID network architecture, and study three coretechnologies which are distributed data processing, storage, queries and load balancingalgorithm. In network architecture research, apply centralized management and schedulingwith distributed processing, storage and queries, to ensure performance and scalability;abstract perception information, business information, hardware devices, networkconfiguration into resources, abstract upper application needs into services, to unifysupervision and scheduling, and enhance the universality of RFID network architecture;business modeling, scenario modeling and physical modeling of three-stage modelingtechniques map business logic design to physical implementation, and enhance pertinence.In distributed data processing, distribution of scheduling and execution achieves a betterresponse time and throughput, data-flow and control-flow pipeline is provided for an easilyassembly and secondary development. In distributed data storage and queries, take a different storage and addressing modes for different types of resources. Resourcesaddressing adds the mapping from hardware resource, classification, labeling, hierarchyand other attributes to IRS based on ONS and DS. In order to support advanced queries,the map can also include time, region or other dimensions. In distributed load balancingalgorithm, genetic algorithms and adaptive adjusted algorithms are applied to deploy andschedule the dynamic network topology in real-time, to ensure maximize performance.This paper first introduces the research background of RFID network, and presentsthe research content and innovations. Secondly, discuss the mainstream RFID networksolutions, analyze its characteristics and inadequacies, and propose the research ideas andkey technology in this paper. Next we analyze the challenges and opportunities of IoT,propose corresponding principles and design network architecture. Then study the threecore technologies which are distributed data processing, storage, queries and loadbalancing algorithm. Concrete system implementation, test and analyze results to show theeffectiveness of proposed method. Finally conclude this paper and look forward the futurework in the RFID network research. |