认知无线网络中分布式盲信道汇合算法研究

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 工业控制网络与系统研究室

题名	认知无线网络中分布式盲信道汇合算法研究
作者	杨博
学位类别	博士
答辩日期	2017-05-27
授予单位	中国科学院沈阳自动化研究所
授予地点	沈阳
导师	肖杨 ; 梁炜
关键词	认知无线网络信道汇合跳频序列分布式对称算法抗干扰
其他题名	Research on Distributed Blind Channel Rendezvous Algorithms in Cognitive Radio Networks
学位专业	控制理论与控制工程
中文摘要	认知无线网络是目前解决无线频谱资源稀缺问题的关键技术，其特有的动态频谱接入机制允许非授权用户（即次用户）机会式地与授权用户（即主用户）共享频谱资源，从而提高授权频谱的资源利用率，已成为未来无线网络发展的必然趋势。信道汇合是次用户在通信之前选择公共接入信道并与邻居用户建立通信链路的过程，是认知无线网络通信的前提。受主用户活动影响而无法预留公共控制信道，次用户之间信息的局部性以及无法协商等问题使得认知无线网络的盲信道汇合面临着极大的挑战。本文针对现有研究在次用户之间时钟异步、可用信道异构、无先验标识信息辅助汇合以及抵抗恶意用户攻击等方面存在的挑战性问题，从全局信道/局部信道、单天线/多天线、两用户/多用户等角度对信道汇合问题开展了系统性研究，取得如下创新性成果： 1. 面向全局信道的两用户单天线场景下的信道汇合算法。针对次用户在时钟异步、信道异构、次用户角色对等、次用户信息匿名等条件下难以保证信道汇合的问题，提出三种信道汇合算法。所提基于Galois域的信道跳频算法，利用Galois域的和两种素数运算定义接入信道，各周期序列依次按升序遍历信道、在校验信道等候和按降序遍历信道以确保汇合。针对前述算法仅在信道数为素数时才可保证汇合的问题，提出了无此约束的基于衬底Dyck路径的跳频算法，依据迂回折返的衬底Dyck路径生成跳频序列，可在迂回停留信道提供更多的重叠机会从而加速汇合。针对前述算法在每个跳频周期内仅能在少数信道上汇合的问题，提出了可同时在所有公共信道上汇合的跳频算法，利用动态跳频与静止等候相结合的方式以及欧拉五边形数对素数求模后的均匀分布性质，可使汇合机会最大化从而更快实现汇合。分别从理论分析和仿真角度与现有算法的汇合延迟进行比较，所提出的3种算法在汇合时间上均具有明显优势。 2. 面向局部信道的多用户多天线场景下的信道汇合算法。针对全局信道信息在分布式环境中不可知的问题，提出了面向局部信道的全分布式跳频算法。该算法将跳频序列转化为葵花集合元素的排列映射问题，采用动态跳频与静止等候相结合的方式并将葵花集合元素按不同顺序交错排列，从而变换访问局部信道的顺序，可使次用户在分布式环境下较快搜索到邻居。此外，进一步将该全分布式跳频算法扩展至多天线、多用户的场景，分别通过多根天线无冗余的高信道容量以及邻近中继用户的报文合作传递机制来加速汇合。分别从理论和仿真角度验证了该算法的有效性，结果表明在单天线/多天线、两用户/多用户的不同组合场景下，所提算法在汇合时间上均具有一定优势。 3. 以抗干扰为目标的高可靠信道汇合算法。针对现有多数汇合算法的固定性跳频规律易于被恶意干扰用户检测并阻塞通信的问题，提出了跳频规律难以预测的高随机、高可靠的信道汇合算法。该算法将次用户随机选取的守护信道转化为二进制序列模板，该模板可在时钟有偏移的条件下重叠并引导汇合。依据该二进制模板生成包含0/1不同类型列的随机扩展矩阵（其中0/1两种类型列中的遍历信道数互质），将该矩阵按行连接而成的跳频序列可由互质性质确保汇合。从理论上分析了所提算法的汇合时间复杂度和抗干扰能力，仿真结果也表明所提算法在有限强度干扰场景下仍可较快实现汇合。
英文摘要	The cognitive radio network (CRN) is a key technique to address the current scarcity problem of the wireless spectrum resource. The characteristic dynamic spectrum access (DSA) scheme of the CRN paradigm allows the unlicenced or secondary users (SUs) to opportunistically share the spectrum with the licenced or primary users (PUs). As a result, the efficiency in the usage of the licensed spectra is improved, which indicates that the CRN technique has become an inexorable trend for the future development of wireless networks. Channel rendezvous, as a prerequisite for data communications in CRNs, refers to the procedure in which SUs select a common access channel and establish communication links with their neighbors. It is impossible to reserve a common control channel in the CRN due to the influences of PUs’ activities, the problems of local information among SUs and inability to negotiate also further challenge the blind channel rendezvous issue in CRNs. In view of the challenging problems existing in the state-of-the-art works, such as the asynchronous clocks, heterogeneous available channels, no assistance of a priori identification information and resistance to adversarial attackers, we investigate the channel rendezvous issue systematically from the perspectives of global-channel-oriented vs. local-channel-oriented, single-radio vs. multi-radio and pairwise rendezvous vs. multi-user rendezvous. The contributions of this dissertation are summarized as follows. 1. Global-channel-oriented rendezvous algorithms in the pairwise and single-radio scenarios. As it is difficult to guarantee rendezvous in the holistic scenarios of asynchronous clocks, heterogeneous channels, symmetric SUs’ roles and anonymous SUs’ identifiers, we propose three channel rendezvous algorithms. The proposed Galois-field-based channel hopping (CH) algorithm defines the accessed channels by performing the operations of plus and multiply in the Galois field. The CH sequence in each period sequentially accesses all the channels in a clockwise direction, accesses the parity channel and accesses all the channels in a counter-clockwise direction so as to ensure rendezvous. However, the Galois-field-based CH algoritm requires that the number of global channels must be a prime number. To remove this constraint, we propose a padded-Dych-paths-based CH algorithm, which generates CH sequences according to the roundabout padded Dyck paths. This algorithm can create more overlapped opportunities when SUs access the roundabout channels, and thus can accelerate the rendezvous. However, the above two algorithms can only achieve rendezvous on a small number of channels in each CH period. To this end, we propose another CH algorithm that can simultaneously rendezvous on all common available channels in each period. This algorithm adopts the combination scheme of dynamic hopping and static waiting. Moreover, as the Euler’s pentagonal number modulo a prime integer can yield a balanced permutation of channels, the proposed algorithm leverages this property to maximize the rendezvous opportunities and can achieve rendezvous faster. We compare the proposed algorithms with the state-of-the-art rendezvous algorithms in terms of rendezvous latency in theories and simulations, respectively. Results reveal that the proposed three algorithms have obvious superiorities in terms of rendezvous latency. 2. Local-channel-oriented rendezvous algorithms in the multi-user and multi-radio scenarios. Considering that the information of global channels may not be available in the distributed environment, we propose a fully distributed CH algorithm by utilizing only local available channels. We transform the design of CH sequences into the permutation and mapping problems of the elements in a sunflower set. We adopt the combination scheme of dynamic hopping and static waiting. We also leverage the interleaved permutations of elements in a sunflower set to vary the order of accessing the local available channels. This method enables SUs to efficiently discover neighbors in the distributed CRNs. Furthermore, we extend this fully distributed algorithm to the multi-radio and multi-user scanarios. We utilize the irredundant hign channel capacity in the multiple radios and the cooperative message passing scheme among the nearby relaying users respectively to accelerate the rendezvous. We validate the effectiveness of the proposed algorithms in theories and simulations, respectively. Results show that the proposed algorithms can achieve superior rendezvous latencies in all the combinations of single-radio/multi-radio and pairwise/multi-user scenarios. 3. Highly-reliable rendezvous algorithms with the aim of anti-jamming. Considering that the CH patterns of most existing rendezvous algorithms are easy to be detected and blocked by adversarial jammers, we propose a highly-random and highly-reliable rendezvous algorithm whose CH patterns are difficult to predict. In this algorithm, each SU converts a randomly selected guardian channel into a binary sequence template, which ensures to overlap in the asynchronous scenario and to guide rendezvous. Guided by the binary template, the CH sequence is constructed by concatenating each row in a randomly generated extention matrix, which includes the one-type columns and the zero-type columns. As the numbers of channels enumerated in the one-type and the zero-type columns are co-prime, the rendezvous is guaranteed by the co-primality property. We analyze the theoretical rendezvous latency and the anti-jamming capability, respectively. Simulation results also reveal that the proposed algorithm can achieve rendezvous faster than the existing ones even under a proper intensity of jamming attacks.
语种	中文
产权排序	1
内容类型	学位论文
源URL	[http://ir.sia.cn/handle/173321/20568]
专题	沈阳自动化研究所_工业控制网络与系统研究室
推荐引用方式 GB/T 7714	杨博. 认知无线网络中分布式盲信道汇合算法研究[D]. 沈阳. 中国科学院沈阳自动化研究所. 2017.