不确定环境下ThingML设计量化分析与寻优框架

发布时间：2018-04-17 14:41

本文选题：物联网 + 不确定建模　；参考：《华东师范大学》2017年硕士论文

【摘要】：随着传感器技术、通讯技术、嵌入式计算等的发展,物联网如今已经能够被广泛应用于各个领域。ThingML作为一种建模与规约语言,可以对资源约束的物联网应用进行快速高效地开发,因而被受到关注。由于物联网应用通常被部署在开放的物理环境下,因而其执行会受到一些环境中的不确定因素影响(比如:网络延迟)。然而在现有ThingML建模中缺乏一种有效的机制对这些不确定因素进行建模,并且有效地量化由不确定性而产生的性能影响。因此,在不确定环境下由ThingML生成的物联网应用的质量和性能很难得到保证。如何高效地对在不确定环境下的基于ThingML的物联网设计的服务质量进行评估并从中快速获得最优物联网参数配置成为如今面临的一个挑战。针对这些问题,本文的主要创新如下:1.针对ThingML建模机制,本文对其语法语义进行扩展,使得能够精确地对由外部不确定环境引起的性能变化进行精确刻画。这是因为现有的ThingML模型假设物联网事物工作在理想环境中,因而最终生成的物联网应用通常很难保证服务质量,所以必须在ThingML中对这些行为进行有效建模。2.本文采用时间价格自动机,作为扩展ThingML的计算模型。通过使用提出的映射规则,扩展的ThingML设计可以转化成价格时间自动机网络模型从而进行相应的量化分析。通过对服务质量的量化分析的,物联网开发者能够对ThingML模型在不确定环境下的性能进行评估、调整以及保证。3.基于所产生的价格时间自动机网络模型,本文提出了一个全新的框架能够有效地评估物联网应用的服务质量,并且采用机器学习中神经网络算法从而能够寻找出最优目标参数配置。这是因为如果对系统参数组合进行逐一验证,将消耗大量时间。本文提出的学习方法能够大幅减少物设计者所消耗的物联网系统参数寻优时间。实验结果证明本文提出的评估与寻优框架能够有效地在不确定环境下对基于ThingML物联网设计进行服务质量分析,并且对系统参数进行快速寻优,从而方便ThingML开发者的设计决策。
[Abstract]:With the development of sensor technology, communication technology, embedded computing and so on, the Internet of things has been widely used in various fields. ThingML as a modeling and specification language,Resource constrained Internet of things applications can be developed quickly and efficiently, so it has attracted much attention.Because Internet of things applications are usually deployed in open physical environments, their execution is affected by uncertainties in some environments (such as network latency).However, there is a lack of an effective mechanism to model these uncertainties in the existing ThingML modeling, and to quantify the performance effects caused by uncertainty effectively.Therefore, the quality and performance of Internet of things applications generated by ThingML in uncertain environments are difficult to guarantee.How to efficiently evaluate the QoS of Internet of things (IOT) design based on ThingML in uncertain environment and quickly obtain the optimal IOT parameter configuration has become a challenge.In view of these problems, the main innovation of this paper is as follows: 1.Aiming at the modeling mechanism of ThingML, this paper extends its syntax and semantics to accurately depict the performance changes caused by the external uncertain environment.This is because the existing ThingML model assumes that the things in the Internet of things work in an ideal environment, so the resulting IoT applications are often difficult to guarantee the quality of service, so these behaviors must be effectively modeled in ThingML.In this paper, time price automata is used as the computational model of extended ThingML.By using the proposed mapping rules, the extended ThingML design can be transformed into a price time automaton network model for corresponding quantitative analysis.Through the quantitative analysis of QoS, Internet of things developers can evaluate, adjust and guarantee the performance of ThingML model in uncertain environment.Based on the generated price time automata network model, this paper proposes a new framework to effectively evaluate the QoS of Internet of things applications.And the neural network algorithm in machine learning is used to find out the optimal target parameter configuration.This is because it takes a lot of time to verify the combination of system parameters one by one.The learning method proposed in this paper can greatly reduce the time consumed by the object designer in optimizing the parameters of the Internet of things system.The experimental results show that the proposed evaluation and optimization framework can effectively analyze the quality of service (QoS) of Internet-based design based on ThingML in uncertain environments, and quickly optimize the system parameters, thus facilitating the design decision of ThingML developers.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.44;TN929.5

【相似文献】