毕业设计外文资料翻译-扩展的低功耗无线互联网架构的设计与实现

1、毕业设计外文资料翻译学 院： 专业班级： 学生姓名： 学 号： 指导教师： 外文出处：(外文)G JonathanW.Hui美 An Extended Internet Architecture for Low-Power Wireless Networks-Design and Implementation,2008 附 件：1.外文资料翻译译文； 2.外文原文指导教师评语：该同学的英文专业资料术语翻译基本准确，体现了一定的专业英语水平。翻译材料能够与原文保持一致，能正确表达出原文意思。翻译字、词数满足要求。翻译材料的格式基本符合要求。该同学较好的完成了外文文献翻译工作。签名： 2015年1

2、0月14日第第 页，共9页1外文资料翻译 译文扩展的低功耗无线互联网架构的设计与实现10.结论10.1.体系结构的回溯在过去的十年中，对无线传感器网络和互联网架构，尤其是IPv6的互联网架构的研究，都取得了极大的进步。因此，重新审视这个在大量传感器网络研究基础上形成的假设是有意义的。通过构建以IPv6为基础的高质量传感器网络，我们发现发展的两个领域实际上有很强的互补性。IPv6架构基本满足传感器网络s的独特需求，在很多情况下比分别进行的任何努力出不针对任何特定网络制式的关注更好。然而，在一些情况下，而结构是合适的，对特定网络RFC和实现是没有。扩展不得不被创建为包括特别需要和这些扩展自然由系统

3、使用的选项支持。传感器网络研究更侧重于网络协议的算法和机制，而比在更广泛意义上的网络。分层的IPv6形式，地址，标题格式，CON网络配给管理，路由和转发提供缺少的结构。并且，在许多情况下，在传感器网络空间开发到一些长期挑战问题提供了优雅的解决方案传统的IETF的方法。我们不仅ND网络连接的接近模拟了在传感器网络如此普遍的传播和收集的结构，但涓流机制提供达到一个优雅的手段。共识，快速响应变化，成为被动的稳定状态。需要特别指出的是，我们不必放弃分层，以满足严厉的许多优点传感器网络节点的资源限制。如果有什么事情，通过分层方法获得的重点往往产生比当许多自由度在一个特设的同时解决更好的解决方案方式。我们

4、能够获得极低的功耗，占地面积小，良好的吞吐量，低延迟，并具有层状溶液高可靠性。然而，这些层之间的界面不能不经意的限制和挑战的性质，为层，以提供足够的表现力有效的合作。无人使用的节点，需要以方便CON组和管理的绝对数量以及是通过ICMPv6的提供的机制对齐。而在IPv4中发现和反对网络连接的许多方面的外层像BOOTP和DHCP2服务依赖，这些都被系统地纳入具有增强发现IPv6的架构，是由较大的，较简单的命名空间，并支持组播功能。我们发现该ICMPv6的功能远远超过已经提出具体传感器网络。事实上，如果传感器网络不掺入的IP架构，多的这种功能将需要被彻底改造这些网络进入生产。然而，大多数这些IPv6

5、的解决方案假设所有节点从一个指定的单插槽。它们需要被扩展到服务于多插槽情况。或许令人惊奇的结果是，通过将每个节点作为路由器，因此德音响宁重叠链路本地作用域的网络体系结构，该现有的IPv6协议可以用简单的选项自然延伸。要在，它似乎是模仿经过多跳比较常见的单播域会保留更多的现有的IPv6的机制。我们的经验是，我们需要有2层的版本同样的功能，支持模拟，所以这是更自然，保持第二层非常简单，利用路由能力是通过内置于第三层。而它是否局部算法和在网络处理将变得仍不清楚普遍，而不是相对简单的数据收集，记录，警报和CON组配置中，使用的IPv6架构的，因为我们已经制定它是与支持这种算法的愿望是完全一致的。通过构

6、建架构重叠的链路本地范围提供全局路由，局部算法只是作为UDP数据报以各种组播地址或单播实现地址。它也仍不清楚命名是否会成为主要的数据中心，如果是这样，这是否将需要网络堆栈内或应用覆盖解决。但不管，通过提供对于传感器网络干净的IPv6架构，这场辩论从本质上变成一样的以数据为中心的争论在互联网的其他部分。我们所做的网络连接第二是大的，简单的地址空间，使用组播集团架构的德网络地址空间的角度定义，和规律性实际上使资源约束溶液比在IPv4的设置更加容易。压缩和首标信息省音在那里可以从第2层头部中的共享上下文的一些简单的假设存在下重建变得简单。虽然传感器网络实际上更多的应用程序特定网络连接温度比传统网络中

7、，我们网络ND的联网机制和结构都没有。什么是应用特定网络是如何使用这些的机制的优化和网络是如何的架构框架内举办。 是否这是通信或数据流的确定性调度，以减少开销，简单的机制这种支持多种应用可以利用应用特定网络C语言结构。然而，机制当网络的行为是不按特定网络连接模式还提供了更广泛的安全网。例如，它可以是在低功率与始终在线行为充分运作而节点通信，以确定时间表。或者，它可以回退到样监听时间表漂移的时候。这是典型的传感器网络的相对简单的应用程序特征可被利用来简化协议及其实施，实现小的资源需求。这样的解决方案可以是次优任意任意到任意这对于传统网络的主要设计点传输。10.2.影响研究本文广泛的架构问题都与社

8、会协议的最佳解决。在本工作中，我们一直在积极参与标准组织，如IETF和IEEE。我们当前与IEEE 802.15.4e工作组的工作，并提出了许多的想法，从第4章包含在IEEE 802.15.4标准。 IETF内部6LoWPAN的工作组发布了建议标准IEEE 802.15.4帧中进行通信的IPv6数据报。我们的工作初期工作导致头格式特定网络版在RFC 4944，它采用了适配层报头格式第5章中提出然而，仍然存在内6LoWPAN的显着的工作完成基于IPv6的网络架构，我们计划继续这一进程。我们的报头压缩机制在第5章介绍所服务作为报头压缩改进的基础上的。我们在第6章的工作是在发现的设计和特异性阳离子对

9、于6LoWPAN的网络。尽管6LoWPAN的工作组还远远没有完成，其成果已传播IETF的外面。其他标准组织，如ISA84，都采用了6LoWPAN的头用户资格垫和工业自动化应用基于IPv6的网络架构。学术努力已经实现了基于RFC4944基于IPv6的网络协议栈27，68。众多的商业实体也做同样的，包括日立76，Jennic公司88，Sensinode154和Sun Microsys-统160。我们在本文描述和评价执行情况的变种作为核心技术拱石公司。我们的产品质量实施一直在使用在实际客户部署的一年。我们也使用同样的技术在学术环境。 在加州大学伯克利分校，该技术被用来教标题为本科班道：“每天物联网“

10、26。通过支持熟悉的基于IP的通信机制，学生能够专注于利用广泛的现有的基于IP的工具，而不是构建基于Web的应用程序在非标准网络堆栈并将其纳入传统网络的复杂性。10.3.结束语支持IP提供与现有的IP设备的互操作性的宝贵，以及能够连接传感器网络到其他IP网络（即网络重新连接时利用现有的IP工具的广泛身体，代理，高速缓存等）。我们已经表明，基于IPv6的网络架构可以实现更在传感器网络比已被证明迄今没有遵守任何特定的标准。我们的实现是能够达到的0.65的平均占空比，平均每62ms跳延迟，和99.98，比在现实世界的家庭监控应用一个为期4周的数据接收速率其中每个节点产生每分钟一个应用程序包。基于广泛

11、的IPv6支持在同一个技术上下文之前更有限的存在，更奇特的解决方案意味着我们可以重新审视一个更为广泛的研究问题。该在网络处理，聚集，压缩和查询处理的问题没有被约束底层网络能力。这是相当简单的实现任何这些。其功效可以从问题的强烈分离是否应该应用水平叠加来实现或在某种程度上更深入地集成到网络堆栈。同样地，许多长期存在的传感器网络问题可以在一个更一般的设置进行检查，并且如果答案是“打开TCP连接”或“发送UDP数据报给需要的地方去”，这个问题的答案是可以接受的为好。在另一方面，一些新的问题出现关于互联网架构应该如何更改或现在演变，它是支持一类新的应用程序。这是难以忽视的普及UDP和TCP 它们必须被

12、实现为提供与现有的IP设备端至端的互操作性。然而，许多人认为，无论是UDP或TCP传输是周期性的读数，许多工业协议已经在常规遇到这些紧张关系有线链路，但他们采取的低功耗无线设置一个新的。另一个例子是存在实施小组的，特别是当致动涉及。楼宇自动化和家庭自动化的需要，例如，似乎是从音频和视频流，其中IP组播是常见大不相同。异质性的部署变得更加自然的，因为IP路由通过取消网络支持跨越各种链接。在这些或各种其它的研究，所述的IPv6体系结构提供了一个框架开发特定网络机制，如此有效地做，即使严重的资源限制，没有每个研究需要到尚未开发的另一个MAC，路由协议和传输。因此，它会似乎是发展两行不只是技术上的互补

13、性，合并可能加速在这两个进步。2.外文原文An Extended Internet Architecture for Low-Power Wireless Networks - Design and ImplementationChapter 10Conclusions10.1 Architectural RetrospectiveIn the past decade, wireless sensor network research and Internet architecture, especially IPv6, have both progressed substantially.

14、Thus, it makes sense to revisit the assumptions that have formed the basis of so much 传感器网络 research. By constructing a high quality 传感器网络 using IPv6 as a foundation, we nd that the two areas of development are in fact highly complementary. Most of the unique requirements of 传感器网络s are well served b

15、y the IPv6 architecture, in many cases better than by any efforts that were carried out without concern for any particular network standard. However, in several instances while the architecture was appropriate, the specic RFCs and implementations were not. Extensions had to be created to encompass p

16、articular needs, and these extensions are naturally supported by the systematic use of options.传感器网络 research has focused much more on network protocol algorithms and mechanisms, rather than on networking in the broader sense. The IPv6 forms of layering, addressing, header formats, conguration, mana

17、gement, routing, and forwarding provide the missing structure. And, in many cases the mechanisms developed in the 传感器网络 space provide elegant solutions to problems that have long challenged conventional IETF approaches. Not only do we nd a close analog to the structures for dissemination and collect

18、ion that are so common in 传感器网络s, but Trickle mechanisms provide an elegant means of reachingconsensus, responding quickly to changes, and becoming passive in the steady state. In particular, we nd that we need not forsake the many virtues of layering to meet the severe resource constraints of 传感器网络

19、 nodes. If anything, the focus obtained by a layered approach tends to produce better solutions than when many degrees of freedom are addressed simultaneously in an ad-hoc manner. We were able to obtain extremely low power consumption, small footprint, good throughput, low latency, and high reliabil

20、ity with a layered solution. However, the interfaces between the layers cannot be oblivious to the nature of the constraints and challenges, to provide enough expressiveness for the layers to cooperate effectively.The sheer numbers of nodes, unattended use, and need for ease of conguration and manag

21、ement are well aligned with the mechanisms provided by ICMPv6. While many aspects of discovery and conguration in IPv4 relied on external layer 2 services like BOOTP and DHCP, these have been systematically incorporated into the IPv6 architecture with enhanced autoconf and discovery that are enabled

22、 by the larger,simpler namespace and multicast support. We nd that ICMPv6 capabilities far exceed anything that has been proposed specically for 传感器网络s. Indeed, if 传感器网络s are not to incorporate the IP architecture,much of this functionality will need to be reinvented for these networks to go into pr

23、oduction.However, most of these IPv6 solutions assume that all nodes are a single hop from a designated agent. They needed to be extended to service the multihop case. Perhaps the surprising result is that by treating each node as a router and hence dening a network architecture of overlapping link-

24、local scopes, the existing IPv6 protocols can be naturally extended with simple options. Going in, it seemed that emulating the more common single broadcast domain over multiple hops would preserve more of the existing IPv6 mechanisms. Our experience was that we needed to have layer 2 versions of th

25、is same functionality to support the emulation, so it was much more natural to keep layer 2 extremely simple and utilize the routing capability that is by denition built in to layer 3.While it remains unclear whether localized algorithms and in-network processing will become prevalent, rather than r

26、elatively straightforward data collection, logging, alarms, and conguration, the use of an IPv6 architecture as we have formulated it is entirely consistent with the desire to support such algorithms.By constructing the architecture as overlapping link-local scopes that provide global routing, local

27、ized algorithms are simply implemented as UDP datagrams to various well-dened multicast addresses or to unicast addresses. It also remains unclear whether naming will become primarily data-centric and if so whether this will need to be addressed within the network stack or by application overlays. B

28、ut regardless, by providing a clean IPv6 architecture for 传感器网络s, this debate becomes essentially the same as the data-centric debate in the rest of the Internet. What we did nd was that the large, simple address space, use of multicast groups dened in terms of that address space, and regularity of

29、the architecture actually made the resource constrained solution easier than in the IPv4 setting. Compression and elision of header information where it can be reconstructed from the layer 2 header in the presence of some simple assumptions of shared context becomes straightforward and efcient.While

30、 传感器网络s are indeed more application specic than traditional networks, we nd that the networking mechanisms and the architecture are not. What is application specic is how the use of those mechanisms is optimized and how the network is organized within that architectural framework. Whether it is dete

31、rministic scheduling of communication or streaming data to reduce overhead, simple mechanisms that support a wide variety of use can exploit the application specic structure. However, the mechanisms also provide a more general safety net when the networks behavior is not following the specic pattern

32、.For example, it can be fully operational at low power with always-on behavior while nodes communicate to determine that schedule. Or, it can fall back to sample listening when the schedule drifts.The relatively simple application characteristics that are typical of 传感器网络s can be exploited to simpli

33、fy protocols and their implementations to achieve small resource demand. Such solutions may be suboptimal for the arbitrary any-to-any transfers that are the primary design point for conventional networks.10.2 Research ImpactBroad architectural issues are best addressed in agreement with the communi

34、ty. Throughout this work, we have been actively involved in standards organizations such as the IETF and the IEEE. We are currently working with the IEEE 802.15.4e working group and have proposed many of the ideas from Chapter 4 for inclusion in the IEEE 802.15.4 standard. The 6LoWPAN working group

35、within the IETF published a proposed standard for communicating IPv6 datagrams within IEEE 802.15.4 frames. Our work initial work led to the header formats specied in RFC 4944, which incorporates the adaptation-layer header formats presented in Chapter 5. However, there still remains signicant work

36、within 6LoWPAN to complete an IPv6-based network architecture and we plan to continue the process. Our header compression mechanisms presented in Chapter 5 are serving as the basis for header compression improvements to RFC 4944. Furthermore, our work in Chapter 6 is signicantly inuencing the design

37、 and specication of Neighbor Discovery for 6LoWPAN networks.Even though the 6LoWPAN working group is far from complete, its results are already propagating outside the IETF. Other standards organizations, such as ISA 84, have adopted the 6LoWPAN header format and an IPv6-based network architecture f

38、or use in industrial automation applications. Academic efforts have already implemented an IPv6-based network stack based on RFC 4944 27, 68. Numerous commercial entities have also done the same, including Hitachi 76, Jennic 88, Sensinode 154, and Sun Microsystems 160.A variant of the implementation

39、 that we described and evaluated in this dissertation serves as the core technology for Arch Rock Corporation. Our production-quality implementation has been in use for over a year in real customer deployments. We have also used the same technology in academic settings. At the University of Californ

40、ia, Berkeley, the technology was used to teach an undergraduate class titled: “The Internet of Everyday Things” 26. By supporting familiar IP-based communication mechanisms, students were able focus on building web-based applications using a broad set of existing IP-based tools, rather than on the i

41、ntricacies of a non-standard network stack and their integration into traditional networks.10.3 Last WordsSupporting IP provides invaluable interoperability with existing IP devices as well as being able to utilize the broad body of existing IP tools when connecting 传感器网络s to other IP networks (i.e.

42、, rewalls,proxies, caches, etc.). We have shown that an IPv6-based network architecture can be implemented more efciently in 传感器网络s than what has been demonstrated to date without adherence to any particular standard.Our implementation was able to achieve an average duty-cycle of 0.65%, average per-

43、hop latency of 62ms,and a data reception rate of 99.98% over a period of 4 weeks in a real-world home-monitoring application where each node generates one application packet per minute.The presence of broad-based IPv6 support in the same technological context as prior more limited,more idiosyncratic

44、 solutions means that we can re-examine a much broader set of research questions. The question of in-network processing, aggregation, compression, and query processing are not constrained by the underlying networking capability. It is fairly straightforward to implement any of these. Their effectiveness can be strongly separated from questions of whether they should be implemented as application level overlays or somehow more deeply integrated into the network stack. Similarly, many of the long-standing 传感器网络 questions can be examined in a much more general setting, and if th