【作者】 王军平；

【导师】 马严； 朱其亮；

【作者基本信息】 北京邮电大学 ， 计算机科学与技术， 2013， 博士

【摘要】 随着物联网技术迅猛发展,物联网日益形成一个巨大的商业价值网。这样激励了服务提供者、内容提供商参与到物联网服务开发中。这种趋势需要开发一种合适的、可扩展的基于物联网的服务提交平台SDP,以便多个服务提供者能够快速、灵活地提供满足用户需求的创新性服务。但是受物联网环境的动态性、网络能力有限、能量受限、分布性、多样性和开放性等影响,如何基于物联网来提交服务就成为服务计算领域中面临的一个重要挑战。本课题围绕基于物联网的服务提交关键技术展开研究,内容包括基于物联网的服务提交框架模型及其形式化描述、物联网能力抽象框架及其相关算法、基于生命周期的分布式服务描述和动态组合模型及其算法和面向多个服务提供者的服务可信暴露模型及其算法。取得的研究成果和创新点包括：1)在TMF的SDF、SensorLogic SDP和L.Atzori的IoT中间件模型的基础上,提出了一种基于物联网的服务提交框架模型。针对当前物联网服务环境的混乱、“烟囱”式发展服务模式和缺乏可持续性发展等问题,提出了这种模型。首先该模型借鉴TMF的SDF的分层理念将基于物联网服务提交环境划分为服务提供者、服务消费者和物联网基础设施提供者。其次、该模型采用抽象和动态协作方法实现了物联网资源共享。最后、采用服务动态组合技术实现了复杂服务快速提交。同TMF的SDF、SensorLogic SDP和L.Atzori的IoT中间件模型相比,有效解决了物联网服务稳定性、动态组合和可信暴露问题。2)参考了欧盟FP7的SENSEI、G.Fortina的物联网抽象和动态协作的研究成果,提出了一种具有动态协作、自组织和容错能力的物联网能力抽象框架及其相关算法。针对当前物联网中每个节点易发生故障、不可靠、无质量保证和动态可变等问题,提出了这种模型。首先、该框架针对物联网中非结构化数据的实时性和复杂性,提出了一种将物联网中每个Smart Object(CHN节点)抽象成为PRA的方法。然后、针对每个PRA计算能力有限,提出了一种基于RCT的动态协作和自组织的算法。最后、针对每个PRA内部结构动态可变、提出了一种基于TFA和负载的自演化算法。与欧盟FP7的SENSEI、 G.Fortina的物联网抽象和动态协作方法,该模型及其相关算法不仅有效保持了物联网服务能力的稳定性和可靠性,而且能支撑复杂服务的组合。3)根据E.G. da Silva的动态服务组合方法和S.C.Geyik的传感服务动态组合方法的研究成果,提出了一种基于生命周期的分布式服务描述和动态组合模型及其算法。针对物联网原子服务分布性强、以非结构化数据驱动服务特性和无法以集中式方式组合这些服务能力等问题,提出了该模型及其算法。该模型首先采用元数据将每个PRA封装成语义化服务组件,解决了基于数据驱动服务语义描述问题,增强了服务健壮性。然后,按照服务生命周期将物联网服务动态组合划分为规划、发现、选择和执行等阶段。在规划阶段,提出了一种面向PlanningDomain, UserRequirement和Workflow的服务组合策略的生成算法。在发现阶段,提出了一种基于Input/Output语义匹配的服务类动态发现算法。在选择阶段,提出了基于Fuzzy Logic和PSO融合的服务选择算法。在执行阶段,采用服务组合脚本语言实现设组合服务的元数据与服务实例元数据动态映射。与E. G. da Silva的动态服务组合方法和S. C. Geyik的传感服务动态组合方法相比,有效解决了非结构化数据驱动的物联网服务动态组合问题。4)借鉴S. Alam的服务暴露和T.Finin的SDP服务暴露的解决思路,提出了一种面向多个服务提供者的服务可信暴露模型及其算法。该模型首先采用基于Trust Policy的会话访问控制策略,给每个服务提供者分配虚拟的身份证。然后通过直接或间接信任推理建立互信机制,并采用服务级别协议SLA构建双方可信的暴露协商机制。最后通过服务综合评估计算策略确定为服务提供者配置SLA服务等级,并暴露相应级别的服务资源。与基于S. Alam提出的服务暴露方法和T.Finin的SDP服务暴露方法相比,从而解决了服务提交能力的可信暴露问题。更多还原

【Abstract】 With the rapid development of Internet of Things technology, a huge business value network is formed. This resuts would encourage multiple service providers to develop services for IoT. Multiple services providers will require the development of a suitable, scalable service delivery platform, which enables the fast and cost-effective creation of new IoT services. However, the network capacity of IoT has some characteristics of errpor-prone, unreliable, limited energy, and constrained resources, so how to deliver services over IoT becomes a big challenge in service computing. The research topic discussed in this paper is about IoT-based service delivery, which includes the service delivery framework model, IoT capabilities abstract framework and related algorithms, distributed service description and dynamic combination based on life cycle model and its algorithm, and trustful service exposure and negoziation mechanism and its algorithm for multiple service providers. Research results and innovation points are summarized as below:1) Based on TMF SDF, SensorLogic SDP, and L. Atzori IoT middleware, a IoT-based service delivery framework model is proposed for chaos of the IoT service environment, service development of "chimney", and lack of sustainability. The first, IoT-based service delivery environment is divided into service providers, service consumers, and IoT infrastructure providers by referring to TMF SDF. Then, the model realizes IoT resource share using dynamic coordination and abstraction technology. Lastly, complex service is quickly created by service dynamic composition. Compared with TMF SDF, SensorLogic SDP, and L.Atzori IoT middleware, the model effectively solves problem of IoT services stability, service dynamic composition, and trustful service exposure.2) Refered to EU FP7SENSEI, G. Fortina object abstraction and dynamic coordination, this paper proposes a novel IoT capabilities abstraction framework with dynamic collaboration, self-organization and fault tolerance, and its related algorithm. For errpor-prone, unreliable, limited energy, and constrained resources of node communication. This framework firstly abstracts Smart Object in the Internet of things (CHN node) into PRA agent for the unstructured data real-time and complexit. This has intelligence formalization, self-management ability, and autonomic computing ability for each PRA agent. Then, a RCT-based coordinate algorithm is proposed for each PRA limited computing power. Lastly, a TFA and Load-based self-envalution algorithm is proposed for each PRA internal structure dynamic variable. Compared with EU FP7SENSEI, G. Fortina object abstraction and dynamic coordination, the model and some algorithm not only maintain IoT services stability, and reliability, but also support the service dynamic composition.3) According to the E.G. da Silva dynamic service composition, and S.C.Geyik robust dynamic sensor service composition, this paper proposes a life cycle-based distributed service description and dynamic composition model and its algorithm for a large heterogeneity, unstructured data-driver service, and decentralization service compostion of atomic sensor service. Each PRA is abstracted into semantic service by metadata in sensor service description, which enhances robustness of service. According to service life-cycle, the dynamic sensor service composition is divided into four phases:service planning, discovery, selection and execution. In the planning phase, it has proposed generation algorithm of dynamic service composition for PlanningDomain, UserRequirement and Workflow. In the discovery phase, it proposes based on Input/Output semantic matching service discovery algorithm. In the selection phase, it has proposed based on Fuzzy Logic and PSO service selection algorithm. In the execution phase, dynamic service composition script is deployed on service broker, which service composition metadata is dynamically mapped on IoT network capabilities. Compared with E.G. da Silva dynamic service composition, and S.C.Geyik robust dynamic sensor service composition, the model and some algorithm effectively solves problem of unstructured data-driver service dynamic composition.4) Referencing S. Alam service exposure of IoT services and T. Finin service exposure on SDP, This paper proposes the trustful service exposure and negotiation mechanism for multiple service providers. The mechanism firstly assignes virtual ID card to each service provider by trust-based access control. Then, the mutual trust is established by the direct or indirect trust reasoning, and the trustful service exposure mechnism is established by the trustful service level agreement negotiation. Finally through comprehensive SLA evaluator, service capabilities are assigned to service provider. Compared with S. Alam service exposure of IoT services and T. Finin service exposure on SDP, service delivery capabilities is exposed for multiple service providers by the trustful service exposure and negotiation mechanism.更多还原