面向资源优化管理的物联网服务组合及自适应研究

Service oriented Architecture can be used to achieve the resource share in the Internet of Things (IoT). However, existing service computing techniques are mainly focus on web services, which cannot be directly applied to the resource constrained IoT. With the resource management as the starting point, this project will explore a series of key research issues, which consists of service construction, service composition and composed service adaptation. The proposed schemes can basically meet the theory and application requirements of IoT. Firstly, according to the dynamic and untrusty characteristics of sensed data, we will research the special features of service oriented IoT; Secondly, aiming at the constraint resource of the sensing layer, this project will propose the QoS model for IoT. Then, based on this model, we will bring the service reuse technology into the service selection and composition, in order to achieve the optimal QoS performance and resource efficiency; Then, considering of the resource status, we will explore the trigger condition of service adaptation and propose the corresponding adaptation mechanism. Finally, oriented to the smart park, we will implement the validation of our research and bring the application demonstration.

面向服务架构可以有效解决物联网资源共享的问题，实现全面物联，但现有成熟的服务计算主要针对全资源Web服务，不适用于资源受限的物联网。本项目以资源优化管理为出发点，探索物联网服务构建、服务组合及自适应中的系列关键问题，使之基本满足物联网的理论和应用需要。首先，开展物联网服务计算特性研究；其次，针对感知层资源受限的特点，提出面向物联网的组合服务扩展QoS模型。以此QoS模型为指导，研究在满足用户QoS约束前提下资源利用效率与组合服务QoS多目标优化的服务发现及组合算法；接下来，结合资源使用状况，探索组合服务自适应触发条件及相应的动态自适应算法。在此基础上，提出面向多用户多任务的资源优化管理机制。最后，面向智慧园区应用，实现面向服务物联网的平台验证，为物联网走向大规模实际应用奠定理论及技术基础。

本项目立足于资源受限物联网环境下的服务选择及组合，重点研究在满足QoS约束前提下的物联网服务计算中其有限资源的优化合理利用，划分为三个方面的研究内容：开展了物联网服务计算特性研究，进而引入资源效率等，面向资源优化管理，提出了面向物联网服务计算的扩展QoS模型；针对部分服务存在QoS缺失或超过时效的问题，提出了协同过滤的缺失QoS预测方法；将资源受限物联网场景下多用户的服务选择及组合问题建模为多任务多目标优化的数学模型，通过数学模型的最优化求解，解决了多用户请求下服务资源的动态优化调度问题。最终，提出了资源有效的物联网服务计算方案，解决从感知数据获取到上层应用构建所面临的资源优化管理系列问题，为大规模物联网应用提供了一定的理论参考和技术支撑。