【作者】 韦云凯；

【导师】 毛玉明；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2016， 博士

【摘要】 无线传感器网络、物联网、M2M[网络、移动社会网络等新兴网络的迅猛发展与广泛应用,将进一步推动信息技术渗透到人类活动的方方面面。然而,这些网络都面临着两个共同的问题：(1)通常工作在无线环境下,链路稳定性较差,具有较高的误码率：(2)网络中的无线节点通常使用电池供电,是能量受限的低功耗设备。具有以上特点的网络,可以统称为低功耗有损网络(Low-power and Lossy Network, LLN)。高误码率和低功耗给LLN网络带来了可靠性与节能之间的矛盾。一般而言,为了提高可靠性,数据传输过程中通常会增加数据冗余、增强差错恢复机制；而考虑到节能需求,则需要降低数据冗余,减少数据传输次数。考虑到LLN网络常见的应用领域,这两者不能做简单的取舍。数据传输可靠率过低,可能导致业务中断,这在某些应用领域(如军事侦察、实时监测、医疗看护等)可能带来严重的后果。另一方面,能耗过高,可能导致节点和网络提前失效,使得业务彻底中断。因此,研究兼顾节能需求的LLN网络可靠传输技术,具有重要的意义。LLN网络中的数据,根据所携带信息的特点可以分成多种类型,其中最具典型性的有两种：(1)数量众多的常规数据。这类数据往往包含诸如温度、湿度、血压、短文本等短信息,数据数量众多,但每个数据所封装而成的报文长度都较短。这类数据对可靠性有较高的要求,但是对时延的容忍度较为宽松。(2)体量巨大的实时多媒体数据。LLN网络中另一类非常重要的应用是实时多媒体应用,如实时音/视频监控、视频会议等。这类数据信息量大、占用网络资源多、对时延要求高,但是对数据的丢失容忍度高于普通常规数据。这两类数据特征鲜明、需求相反,需要分别考虑。目前已经存在大量关于节能和增强数据可靠性的研究。但是,这些研究中很大一部分是单独针对节能或者可靠性进行的,不能直接应用于LLN网络；还有一部分研究将节能与可靠性进行了结合,但往往基于某种具体场景,无法适应LLN网络应用类型多变化、网络架构差异化、数据类型多样化的特征。因此,本论文结合不同数据类型的特点,以提高数据可靠性、降低能耗为目标,兼顾网络架构与应用特点,进行了如下研究：(1)针对数量众多的常规数据,结合其数据短、报文多的特点,采用自动请求重传(Auto Retransmission reQuest, ARQ)机制保证其可靠性需求。分别针对树状网络结构和平面网络结构,对传统的ARQ机制进行了改进,提出了低开销高能效ARQ算法与分段ARQ算法,在保证高可靠性的同时降低了ARQ机制带来的能量开销。本文对ARQ机制的改进,打破了传统ARQ中相关功能节点固化的限制,形成了可以灵活选择差错检查节点、重传请求发起节点及处理节点的动态ARQ系统,并且从理论和应用效果上进行了分析与验证。(2)针对体量巨大的实时多媒体数据,考虑到其实时性要求高、报文的丢失容忍度互不相同、占用网络资源多等特点,研究如何利用最少的信道资源(给定信道带宽),在给定时延约束下保证实时流的有效传输。在不同的擦除模型下,提出了其信道传输能力上限,形成了一套比较完整的RST (Real-time Stream Transmission,实时流传输)容量理论体系,并设计了高效的纠错编码机制,以最少的信道资源和传输代价,达到理论传输能力上限。(3)为了进一步提高数据发送成功率,改善网络中报文转发的效率,并减少网络中冗余报文的数量、降低数据冲突概率,本文使用费马点理论建立了r-费马域,设计了基于费马点与费马域的联合路由算法,同时提出了事件发生预测与修正模型,形成基于事件预测的报文减缓机制,在提高数据传输成功概率的同时,降低了网络能耗与节点能耗。与此同时,本论文对所有新提出的理论模型和算法/协议都进行了理论分析和/或仿真比较,对其正确性和有效性进行了验证。更多还原

【Abstract】 The emerging wireless networks such as Wireless Sensor Networks, Internet of Things, Machine to Machine Communications, Mobile Social Networks, etc., are further permeating information technology into all aspects of human beings. Although these net-works are different in application areas and patterns, they have two common challenges: (1) High bit error rate. These networks usually have relatively high bit error rate, low data rate and unstable links. (2) Limited energy supply. The wireless nodes in these net-works are usually energy limited devices, which are mostly powered by batteries. Such networks can be classified into Low-power and Lossy Networks, or LLNs.The high bit error rate in LLNs leads to the contradiction between energy saving and data reliability. Generally, in order to improve the data reliability, we should increase data redundancy and enhance the error recovery schemes. On the contrary, to save the energy, we should decrease the data redundancy and reduce the transmitted data packets. Considering the usual application areas of LLNs, this is not a simple trade-off. If the data reliability can not be guaranteed, the service may fail and result in serious consequences, such as applications in real-time monitoring, health care, etc. At the same time, if too much power is used to ensure the data reliability, the energy of the nodes (or the network) may exhaust ahead of schedule, leading to the interruption of the services. Consequently, it is important to study efficient data transmission methods that can balance data reliability and energy saving.The data in LLNs can be classified into two kinds:(1) A large number of short data. These data, such as temperature, humidity, blood pressure, text message, etc., are usually huge in amount but short in length, and strict in data reliability but loose in data latency. (2) High-volume real-time multimedia data. Another important application in LLNs is real-time multimedia service, such as real-time monitoring, video conference, etc. These data are content rich and resource hungry stream data, with strict data delay. But they are not as sensitive to data loss as the first kind of data. In a word, the above two kinds of data have opposite requirements and need to be considered separately.There have been many work done about reliability and energy saving. Whereas, most of these researches treat reliability or energy saving separately, which are not suit-able for LLNs. The others usually consume a special application scenario, without con- sidering the differences in applications, network architecture and data types. Therefore, this thesis studies the following issues:(1) Ensuring the reliability of short data. Considering the characters of short length and huge amount, we use Auto Retransmission reQuest (ARQ) to guarantee the reliability of short data. In tree structured LLNs and ad hoc structured LLNs, we develop low-power high efficiency ARQ and stage ARQ respectively, which can ensure high data reliability and reduce the energy consumption caused by ARQ simultaneously.(2) Guaranteeing the reliability of real-time stream data. Considering the characters of high volume, strict delay requirement and loss insensitive (for some packets), this thesis studies efficient Forward Error Correction (FEC) schemes to ensure the reliability of the real-time stream data with minimum resources needed. Consequently, we present the capacity upper bound in different erasure models for real-time streams, and designed efficient FEC code, which can reach the capacity upper bound with minimum channel resource and transmission cost.(3) Improving the successful rate in a single transmission. In order to improve data relay efficiency and decrease the total amount of packets in the network so as to reduce the collision probability, this thesis provides a Fermat Point base Energy-saving Route algorithm and a Variation based Energy-saving scheme. The presented algorithm and scheme can improve the data reliability while decreasing the energy consumption in the nodes and the network.At the same time, all the theoretic models, practical algorithms and protocols pre-sented in this thesis have been analyzed mathematically or simulated practically, validat-ing their correctness and effectiveness.更多还原