【作者】 郭雨齐；

【导师】 钱志鸿；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2014， 博士

【摘要】 物联网是国家重点发展的战略性新兴产业之一，也是当今世界各国关注的一个话题，物联网产业的发展已被纳入我国十二五规划中。所谓物联网，简单地说是指物与物相互连接的互联网；从技术层面说是指通过信息传感设备，将需要进行信息传递的物体，按照相关协议与互联网相连接而形成的一种网络。其主要功能是对物体的实时状态进行感知，对物体进行实时智能控制。物联网技术发展迅速，其中无线传感器网络（WSNs）扮演着极为重要的角色。WSNs将物理世界与虚拟世界联系起来，感知周围物理环境的变化。因此，我们需要利用建模的方法来正确地描述WSNs的特征、适应变化的应用环境。若考虑到物联网的“海量数据”的特征，这个模型就需要能够适应于大尺度的WSNs网络。同时，一个极为重要的挑战就是因特网和WSNs的整合问题。因此，本文首先设计了一种Internet与WSNs的整合方法以及WSNs建模方法，文中所设计的模型能够更好地适应网络的动态变化，具有寻找多个优化路径的能力。提出的扩展报头压缩（EHC）方法可以对IPv6全局地址报头进行压缩，实现IPv6地址的自动配置。另外，针对网络中拓扑的动态变化、多路径数据传输机制和能量溢出等问题，提出一种细胞自动机模型，依据此细胞传输模型，文中提出了WSNs中信息多路径传输策略。仿真实验结果表明，文中所提EHC方案与6LoWPAN比较，吞吐量有了较为明显的提高，同时降低了能量消耗，更适用于大尺度物联网环境；实验也仿真了三种模型：以服务为中心的模型，簇模型和本文提出的细胞自动机模型，实验结果表明与其它两种模型相比，细胞自动机模型传输数据包的时间更少，发现的路径更多。物联网对数据的感知需要采用无线技术作为其传输手段。目前，在众多无线技术中，射频识别技术和蓝牙凭借自身的低功耗、高速、环境适应力强等特点脱颖而出，成为短距离传输物联网关键技术发展的中坚力量。蓝牙工作在全球统一开放的2.4GHz的ISM频段，因此在同频段可能会同时存在多个蓝牙设备，这些蓝牙设备之间会互相干扰。不同的蓝牙微微网之间是独立的，在微微网比较密集的室内环境，几个相邻的蓝牙微微网就有可能同时使用相同频点，而且由于各微微网使用的跳频序列是相互独立的，微微网之间也不会进行信息的交互，因此各微微网并不知道在同一时间对方使用的是哪个频点，如果它们同时使用了某个相同的频点，此时就会发生“频率碰撞”，并由此带来同频干扰，同频干扰的发生会严重影响蓝牙网络的性能，增加分组错误率，降低网络的载干比和吞吐量。这种“频率碰撞”的产生需要同时满足下述三个条件：1）参考微微网与干扰微微网同时传输数据；2）微微网的发射设备与接收设备之间的距离小于10m；3）干扰网与参考网之间发生同频时的载干比小于11dB。针对蓝牙微微网中的频率碰撞问题，本文提出了一种基于信道转换的频率碰撞修复方法。该方法在蓝牙微微网重传时进行信道转换，并采用MSK调制方式代替GFSK调制方式。为了使分析更加完善，方法根据载干比值判断微微网是否受到同频干扰，并且分析了多个蓝牙微微网之间的同频概率，分析同频概率时充分考虑了跳频保护间隔、返回分组与同频概率的关系、3种类型传输分组共存时对网络吞吐量的影响等多种情况。经过大量的仿真实验证明，采用本文所提出的频率碰撞修复方法后，参考网主从设备之间的距离在10m以内时，参考网的载干比有明显的改善，同时分组传输错误率也有了明显的降低；当参考网传输混合类型的分组时，采用本文提出的方法，也得到了较好的效果。因此，本文所提方法有效地减小了分组错误率、提高了参考网的载干比和吞吐量，同时主从设备间不受同频干扰的最大传输距离也有所增加，这在很大程度上减少了同频干扰发生的范围。射频识别技术（RFID）是一种非接触式的自动识别技术，在生产、生活、卫生医疗等各个领域发挥着越来越重要的作用，其最突出的特点即为多标签识别。但阅读器周围总是会有多个标签同时存在，当两个或两个以上标签同时向阅读器发送数据时将会产生数据冲突，即“数据的碰撞”，因此RFID技术的数据防碰撞算法应运而生。目前解决标签数据碰撞的算法主要有基于二进制树搜索的确定性防碰撞算法和基于ALOHA的不确定性防碰撞算法。这两种算法的思想都是标签主动方式，阅读器则需要根据相应碰撞规则对其识别范围内的标签逐个读取，当标签数量增大时，两种算法的复杂度都比较高，并且都会产生明显的系统开销及时延。因此，如何降低系统开销，降低时延，提高系统的吞吐率，是RFID技术需要解决的难题。针对传统防碰撞算法的不足，提高RFID系统中阅读器的标签读取效率，本文提出了一种RFID数据碰撞修复方法—阅读器主动式列表读取方式，这是一种基于UID的列表读取方式，将读取的主动权交予阅读器，通过将阅读器内既定标签群体唯一识别号（UID）事先存储于阅读器地址列表中，按照一定规则对地址列表逐个锁定式搜索，完成标签识别。文中详细阐述了列表式读取方式的三种读取模式：二次读取模式、循环读取模式和嵌套读取模式，分析了三种模式下阅读器寻呼次数、传输时延以及系统效率等三个重要性能指标。分析和仿真结果表明：与二进制搜索算法比较，本文所提的阅读器主动式列表读取方式下，阅读器的寻呼次数较为稳定，系统传输时延低，系统效率恒定。可见标签数量增大时，阅读器主动式列表读取方式的优势更为明显。另外本文提出的方法还具有对标签的主动查找功能，因此更具有优越性。更多还原

【Abstract】 The Internet of things (IoT) is a very popular topic in the world currently, and has beenincorporated into China’s12th five-year developing schemes, belong to one of the keynational strategic emerging industries. Internet of things simply refers to objects and objectsconnecting by virtue of the Internet; it’s one kind of network that some objects who is goingto transfer information must according to the relevant agreement to form a network throughinformation sensing device. Its main function is to sense and control the object in real-time.With the rapid development of Internet of things, the wireless sensor networks (WSNs) playa critical role. WSNs link the physical world and virtual world, perceive the change of thephysical environment around them. Therefore, we need to use modeling method to correctlydescribe the characteristics of WSNs and adapt to changing application environment. Ifconsidering the characteristics of the Internet of things “big data”, this model will need to beable to adapt to the large-scale WSNs network. At the same time, a very important challengeis the integration of Internet and WSNs. Therefore, this paper designed a kind of integrationof Internet and WSNs method and a modeling method for WSNs firstly, in this paper, thedesigned model can better adapt to the dynamic change of network and has the ability ofsearching for multiple optimal path. And proposed Extension Header Compression (EHC)method can compress the IPv6global address header and achieve the automatic allocationof IPv6address. In addition, in view of the dynamic change of network topology, themultipath data transmission mechanism and energy overflow problems etc, a cellularautomata model is put forward, based on the cell transmission model, this paper putsforward the information multipath transmission strategy in WSNs. The simulation resultsindicate that compared with6LoWPAN, the proposed EHC scheme has improvedthroughput obviously and reduced the energy consumption, and EHC scheme is moresuitable for the large-scale Internet of things environment; The experiment also simulatethree kinds of model: service centered model, cluster model and cellular automata model which is proposed in this paper, the experimental results show that compared with the othertwo models, cellular automata model has spent less time when transferred data and couldfind more paths.The Internet of things needs to adopt wireless technology to sense its front-end data. Inthe numerous wireless technologies, Bluetooth and Radio Frequency Identificationtechnology become to the key technical backbone of IoT depends on their characteristicssuch as low power, high speed, and environment adaptable stand out.Bluetooth works in a global unified open2.4GHz ISM band and it is susceptible tointerference with other device in the same spectrum. Because the Bluetooth piconet isindependent of each other, so the probability of using the same frequency will increasewhen in indoor environment which has more piconets. Because the FM series can be usedfor the piconet is also independent of each other, and information interaction also won’tproceed between piconets, therefore the piconet doesn’t know the frequency point which isused by other piconet at the same time, if several piconets have used the same frequency atthe same time then “frequency collision” will be occurred, namely the co-channelinterference, that will seriously impact on the performance of Bluetooth network and reducethe network throughput. This “frequency collision” needs the following three conditions:1)reference piconet and interference piconet to transmit data at the same time;2) the distancebetween interference piconet’s transmitting equipment and reference piconet’s receivingdevice is not more than10m;3) the same frequency carrier interference ratio (C/I) ofreference and interference piconet is less than11dB. The paper puts forward a frequencycollision repair method on channel switching for Bluetooth piconets. The method uses MSKmodulation instead of GFSK, and proceeds channel switching when retransmitting. In orderto analyze more completely, the method used the C/I value to judge whether the co-channelinterference accrued or not, and analyzed the co-channel probability between multipleBluetooth piconets by considering the return packets, frequency hopping guard time, threekinds of data packets coexistence etc, and analyzed the packet error rate (PER) andthroughput further. Simulation experiments show that using the frequency collision fixingmethods, when the distance of master-slave equipment is within10m in the referencepiconet, the C/I has improved obviously, at the same time, the PER has also reducedsignificantly; and the method also has good results when transmitting the mixed typegrouping. Therefore, the proposed method effectively reduced the PER, improved the C/Ivalue and throughput of reference piconet, increased the master-slave equipment’smaximum transmission distance free of co-channel interference, this is largely reduced therange with co-channel interference.Radio Frequency Identification (RFID) is a non-contact automatic identificationtechnology and playing a more and more important role in production, life, health and otherfields, multiple tags identification is the most outstanding character of RFID. But there will always be multiple tags around reader at the same time, when two or more tags send data tothe reader at the same time, it will be produce data conflict, namely “data collisions”,therefore the data anti-collision algorithm arises at the historic moment. Now the mainsolution to tag data collision are Binary Search algorithm and the ALOHA algorithm, thetwo algorithm ideas are tag active way, readers need according to the rules to read tags oneby one within its recognition scope, and the complexity of the two algorithm are all higher,and the overhead and time delay will increase with the tag number increasing. Therefore,how to reduce the system overhead and time delay, improve the system throughput is theproblem of RFID. Aiming at the shortcomings of the traditional anti-collision algorithm, toimprove the efficiency of reading tags, the paper puts forward a way of tag reading—UIDlisted reading mode, the reader reads tags initiatively, With putting the UID of tags to beidentified into the address list of a RFID reader，which was formed in advance，the addresslist would be tracked and searched until all of the tags were identified one by one，accordingto certain rules．The paper expounds on three reading modes of UID-listed reading: secondread mode, circulation mode and nested mode. To prove the strong points of the UID-listedreading mode proposed，query times，transmission delay and system efficiency weresimulated. Compared with the binary search algorithm, the method has relatively stablepaging number, the lower transmission delay and constant system efficiency. Theconsequences of simulation indicate that the UID-listed reading mode turns out to be betterthan the classical Binary Search algorithm．In addition, the proposed method also has activelookup function, so UID-listed reading mode has more advantages.更多还原