【作者】 宋洪治；

【导师】 易波； 石磊；

【作者基本信息】 中国科学技术大学 ， 微电子学与固体电子学， 2011， 博士

【摘要】 石油是重要的战略物资,与国家的经济发展、政治安全和军事安全息息相关,是制约国家发展的重要战略要素。因此,为了保障国家的石油安全,需要减少我国对国外石油资源的依存度,需要建立自己的石油勘探产业链。而石油勘探仪器是这个产业链中的重要一环。目前地震勘探仪器为了满足高分辨率勘探的需求,普遍具有大道数、小道间距、单站单道的特点,其数据传输系统本质上就是一种特殊的分布式传感器网络。传感器网络的系统架构、运行的操作系统、使用的网络协议,以及网络内的分布式数据处理方式都与实际应用紧密相关。而且传感器网络负责感知的实际物理世界也将一定的不确定性引入运行的网络。所以对于传感器网络,需要对实际的应用进行全面的研究,包括使用的操作系统以及协议栈。通常单纯模拟算法或协议是不够的。因为算法往往在网络仿真平台中实现,使用的都是特定的网络仿真平台提供的特性,并没有考虑到运行节点的实际硬件特性和节点上运行的操作系统的特性。由于忽略了这些与应用紧密相关的因素,在网络仿真平台中很难模拟与操作系统或硬件紧密相关的特性,尤其是中断处理、任务调度、内存管理这些基本的底层功能,而这些基本功能极大地影响了节点的行为与效率。所以,有时候网络平台上行得通的算法或协议,在实际节点中的运行效果并不理想,不利于整个系统仿真的完整性与逼真度。而且由于网络仿真平台本身和实际系统的差异性,导致了算法或协议在仿真研究过程结束之后,必须结合实际的应用重新实现,不利于代码的重用,加长了开发周期与代码维护的工作量。研究人员不得不在对算法或协议进行改动之后,在不同的平台上都实施一次同样的改动。受到实际节点的软硬件平台的限制,同样的目的往往需要使用不同的方法。尤其对于传感器网络协议跨层设计的应用,要求可以直接操作底层硬件,以便高效率地完成数据的传输。这使得与硬件平台紧密相关的部分除了算法思想,具体实现几乎与网络仿真器中的实现完全不同,不利于代码的复用。基于以上考虑,本文使用多线程模拟多任务,信号模拟硬件中断的方法实现了FreeRTOS模拟器,并将FreeRTOS模拟器与网络仿真器集成,构成了网络仿真平台。在兼顾网络仿真器本身具有的扩展性的基础上,弥补了单纯使用网络仿真器的不足,对传统网络仿真平台的完整性、逼真度以及代码的复用性方面进行了增强。本论文以大型地震传感器网络仿真平台的研究和实现为核心,总体分为六大部分:绪论、高分辨率地震勘探、地震传感器网络、多任务嵌入式OS模拟器的实现、FreeRTOS模拟器与网络仿真平台的集成、总结与展望。第1章作为绪论,介绍了油气勘探的方法,地震勘探仪器的发展历史,以及当前地震勘探仪器用于数据传输的几种拓扑结构。结合当前第五代和第六代地震勘探仪器使用的网络结构,给出了地震传感器网络的概念,并简要讨论了网络的研究方法,以及采用仿真方法研究网络时,网络仿真器应该满足的特性。为了克服当前网络仿真的不足,提出了论文的研究内容。第2章从理论的角度讨论了地震分辨率的概念,不同的分辨率标准,分辨率的定量计算方法,以及各种影响分辨率和地震传感器网络规模的因素。本章也简要介绍了一些高分辨率地震勘探的方法,并定量计算了这些方法对道间距的需求。最后,讨论了这些地震勘探方法对地震勘探仪器的影响,以及地震勘探仪器的平均道数和数据传输方式的发展趋势。第3章讨论了地震传感器网络的特点和设计时需要注意的问题,并简要介绍了当前已经存在的很多用于传感器网络的路由算法和协议的跨层设计方法。本章还讨论了目前第五代和第六代地震仪器系统采用的数据传输方式,并给出了使用这些数据传输方式的代表仪器实例。在现有无线技术的基础上,结合全无线地震勘探系统需要满足的特性,讨论了一种可能的全无线地震勘探仪器,即全无线地震传感器网络的结构和使用的技术路线。最后指出,适用于地震传感器网络的算法或协议的研究,以及地震传感器网络新结构的研究,都需要结合网络仿真的方法。第4章主要研究用于大型地震传感器网络仿真平台的嵌入式操作系统模拟器的设计与实现。本章在讨论了适用于仿真平台的嵌入式操作系统的特点及需求的基础上,介绍了适用于仿真平台的FreeRTOS系统,及FreeRTOS模拟器在系统中所处的层次。在比较了不同的嵌入式操作系统模拟的实现方式之后,选择了使用进程模拟的方式。采用线程模拟任务,信号模拟中断实现了FreeRTOS核心的模拟。采用异步I/O和层次化的设计实现了硬件外设的模拟。这种设计方法还适用于其它轻量级多任务嵌入式操作系统的模拟。最终实现的FreeRTOS模拟器基本系统只消耗了120KB的内存,平均每个测试任务消耗了10KB左右的内存。这样对一个代表地震传感器网络节点运行的FreeRTOS模拟器,若运行有4个用户任务,只需要消耗不到200KB的内存。第5章主要讨论将FreeRTOS模拟器与网络仿真器集成的方法。在比较了不同的网络仿真器后,综合考虑了网络仿真器应具有的特性,选择ns-3作为仿真平台的网络仿真器。使用了同步点完成ns-3与FreeRTOS模拟器的同步和数据交换,并给出了真实仿真系统的三个基本条件。采用了网格状的拓扑结构,泛洪的方式对仿真平台进行了测试。测试结果表明,对于运行FreeRTOS模拟器的仿真网络,仿真1000个节点时,大约使用了170MB的内存。第6章是整篇论文的总结与展望。归纳总结了采用的技术路线及工作成果,阐述了论文的创新点,提出了未来工作的目标。更多还原

【Abstract】 Petroleum is an important strategic material of the country. And it is closely linked with economic development, political security and military security of the country. To protect petroleum security of the country, dependence on foreign petroleum resources should be reduced, and our own petroleum exploration industry chain should be established. And petroleum exploration instrument is one important link of the chain.Nowadays, in order to meet the requirements of high-resolution exploration, small trace intervals, more channels, one unit one channel are characteristics of modern petroleum exploration equipments. And the exploration system is essentially a specified sensor network.The architecture, OS, protocol and distributed data processing method of a sensor network are closely related with a specified application. And the sensor network is connected to the physical world. This adds uncertainty to execution. So, for a sensor network, the application should be studied thoroughly, including its operating system and network stack. Usually, it is not possible to reveal actual performance of algorithms and protocols by simulation only, without considering characteristics of real platforms. So it is very difficult to simulate behaviors that relate to operating systems or hardwares, especially low level operations, such as interrupt handling, task scheduling and memory management. But these low level operations effect system performance a lot. And some algorithms and protocols run well in a simulation environment, but perform poorly when running in a real node. So such simulations can not meet the requirement for completeness and fidelity. And because of differences between a simulation platform and a real node, algorithms and protocols have to be reimplemented for different ploatforms. And the reusability of the code is limited, the development cycle is lengthened, and the code maintenance workload is increased. Once algorithms or protocols are changed, researchers have to implement the same changes for different platforms. And for different platforms, the same functionality has to be implemented with different approaches, especially for a cross-layer protocol design whose applications can access low level hardware directly for efficiency. So for codes that tied to hardware, only algorithm ideas are the same, implementations are different for different platforms. And the code reusability is not possible.Based on the above considerations, a FreeRTOS simulator is implemented with its tasks and interrupts simulated by threads and signals respectively. And the simulation platform is implemented by integrating the FreeRTOS simulator with a network simulator. And the completeness, fidelity, and reusability of the sensor network simulation platform are enhanced without losing the scalability of the platfrom.In chapter 1, oil and gas exploration methods, seismic exploration instruments development history, and several topologies used by current seismic exploration instruments are introduced. And the concept of a seismic sensor network is given based on topologies of instruments. Methods for researching networks and requirements of a network simulator are analysed. To overcome the shortage of current network simulators, research of the dissertation is proposed.In chapter 2, the concept of seismic resolution and resolution criterions are introduced. And quantitative calculation of seismic resolution, and factors that impact seismic resolution and scale of a seismic sensor network are analysed. Some high resolution exploration methods and their impacts on trace intervals are presented. And the trend of seismic exploration instruments are analysed.In chapter 3, characteristics and design guide of a seismic sensor network are introduced. And some routing protocols and cross-layer protocol design methodologies for sensor networks are presented. Transmission technologies adopted by the 5th and the 6th seismic exploration instruments and typical instruments that use these technologies are demonstrated. On the basis of existing wireless technologies, a wireless seismic sensor network is analysed. In order to construct suitable topology, and find out suitable algorithms and protocols for a seismic sensor network, network simulation should be adopted.In chapter 4, design and implementation of a multi task embedded operating system simulator are presented. On the basis of characteristics and requirements of a embedded operating system simulator for our seismic sensor network simulation platform, FreeRTOS is chosen as the candidate. After introducing FreeRTOS, and its level in the system, methods for simulating embedded operating systems are compared, and the simulator should be implemented by a process. Tasks and hardware interrupts are simulated by threads and signals respectly, and they compose the core of FreeRTOS simulator. Hardware peripherals are simulated by asynchronous I/O with layered design. And the method for implementing FreeRTOS simulator is also suitable for other light weight multi task embedded operating systems. The memeory consumption is 120KB for FreeRTOS simulator core, and about 10KB for each task under test. So, if a FreeRTOS simulator that runs on a seismic sensor network node contains 4 tasks, it only consumes less than 200KB memory.In chapter 5, the method for integrating FreeRTOS simulator with a network simulator is presented. After comparing some network simulators, ns-3 is chosen for its high performance and low resource consumption. At synchronization points, ns-3 and FreeRTOS simulator are synchronized, and data are transmitted. And the simulation is considered to be real when 3 conditions are satisfied. The platform is tested with grid topology by flooding the entire network. The test shows that, for a simulation network with FreeRTOS simulator running on each node, memeory consumption of 1000 nodes is about 170MB.In chapter 6, ideas, methods and results of the entire work are summarized. Innovation points of the dissertation are described. And the goal of future work is pointed out.更多还原