无线物理层安全传输关键技术研究

发布时间：2018-05-29 17:07

本文选题：无线物理层安全 + 人工噪声　；参考：《国防科学技术大学》2014年博士论文

【摘要】：无线物理层安全技术利用发送端到合法用户的信道和到窃听者的信道之间的随机性、差异性和互易性等特性从物理层上实现信息的安全传输。它从信息论意义上确保窃听者无法获取信源发送的信息,在抗窃听、低截获和无线通信系统设计等领域具有广阔的应用前景。多天线技术通过利用丰富的空间资源,不仅能够获得传输的可靠性和有效性两方面的优势,而且自然增加了信道间差异性,为无线物理层安全技术研究提供了广阔的空间。然而,多天线系统的物理层安全传输性能很大程度上依赖于发送端获取的信道状态信息(Channel State Information,CSI),不同信道信息条件和应用场景下的安全传输成为物理层安全技术研究中的关键和难点问题,论文围绕这些关键和难点问题展开研究。首先,针对多输入单输出(Multiple-input Single-output,MISO)系统中发送端已知窃听信道统计信息的情况,论文研究了人工噪声辅助安全传输的功率分配问题,提出了一种最小化系统保密中断概率的最优功率分配算法。已有研究证明,在仅有窃听信道统计信息的情况下,人工噪声的设计应当正交于主信道,而且系统的安全性能取决于有用信号和人工噪声间的功率分配。论文推导了保密中断概率的闭合表达式,并以最小化保密中断概率为目标设计了最优功率分配算法。针对单个具有多天线的窃听者情况,提出了一种基于Golden搜索的最优功率分配算法。针对多个具有单天线的窃听者情况,提出了一种闭合的最优功率分配算法,并将最优功率分配算法转化为一个确定性方程求解,从而推导得到其闭合解。针对上述两种场景的仿真结果表明所提出的最优功率分配算法性能明显优于传统波束成形算法和等功率分配算法。其次,针对多输入多输出多窃听(Multiple-input Multiple-output Multi-antenna Eavesdropper,MIMOME)模型中发送端未知窃听信道信息的情况,论文提出了一种发送天线选择(Transmit Antenna Selection,TAS)和最大比合并(Maximal Ratio Combining,MRC)相结合的安全传输算法。在这种场景下的安全传输设计中,发送端需要得到合法接收端反馈的最优天线系数。在理想反馈时,系统能够获得与未考虑安全性传输的传统多天线系统相同的分集阶数,且该分集阶数与窃听者天线数目无关。在实际系统中,由于信道反馈延时、信道估计误差等因素影响,理想反馈几乎是不可能的。论文分别研究了时延反馈、错误反馈以及联合时延和错误反馈对发送天线选择和最大比合并算法安全性能的影响,推导了获得非零保密传输速率的概率和保密中断概率的闭合表达式。通过分析高信噪比(Signal-to-Noise Ratio,SNR)下的渐近保密中断概率发现,当出现时延或错误等非理想反馈时,来自天线选择的分集增益将消失,系统仅仅获得接收端的最大比合并分集增益。所得分析结果更具有一般性,可直接扩展到已有文献中研究的理想反馈情况。论文通过仿真验证了理论分析结论的正确性。再次,针对多输入单输出的感知无线电网络(MISO Cognitive Radio Network,MISO-CRN)中发送端未知窃听信道信息的情况,论文提出了一种人工噪声辅助安全传输算法。假定网络中主用户和次用户工作于Underlay模式。该算法从用户服务质量(Quality of Service,Qo S)的角度,在保证次用户(Secondary User,SU)的接收信干噪比(Signal-to-Interference-plus-Noise Ratio,SINR)和主用户(Primary User,PU)的干扰温度约束下,最大化人工噪声的发送功率。当次用户发送端已知主用户信道信息和次用户信道信息时,理论证明了波束成形是最优的有用信号传输策略,此时输入方差矩阵的秩为1。然而,当次用户发送端获取的信道信息存在误差时,理想信道条件下设计的人工噪声策略将产生噪声泄漏,严重影响合法用户的接收性能。论文假定次用户发送端获取的主用户信道信息和次用户信道信息均存在误差,分别将信道误差建模为信道矢量误差模型、信道方差误差模型和统计信道误差模型。在信道矢量误差模型和信道方差误差模型下,论文提出了一种基于最差性能的鲁棒人工噪声算法,相应的优化问题是一个NP-hard(Nondeterministic Polynomial hard)问题。基于S-Procedure引理和凸优化理论,论文推导了等效的约束条件,并将原始问题转化为一个半正定规划(Semidefinite Program,SDP)问题,从而得到了最优的人工噪声输入方差设计。针对统计信道误差模型,论文提出了一种基于中断概率约束的鲁棒人工噪声算法,相应的优化问题简化为一个机会约束的非凸问题。借助于Bernstein型不等式,论文将复杂的概率约束条件转化为一系列确定性不等式,从而获得了人工噪声输入方差的一个近似解。仿真结果表明,提出的鲁棒人工噪声算法在三种信道误差模型下均有效地降低了信道误差的影响。最后,针对单向中继窃听系统、双向中继窃听系统和MIMO非信任中继窃听系统中发送端已知理想的信道信息情况,论文分别提出了相关的最优中继波束成形算法、中继聊天安全传输算法和目的端干扰辅助安全传输算法。在出现单天线窃听者的单向中继系统中,目前并没有合适的算法来计算放大转发(Amplifyand-Forward,AF)模式下最大化保密传输速率的中继波束成形矢量。论文首先提出了一种基于分支定界(Branch-and-Bound)的中继波束成形算法,并理论证明该算法能够获得全局最优解。考虑到最优算法的计算复杂度高,进一步提出了一种基于广义功率迭代(General Power Iteration,GPI)的次优中继波束成形算法。次优算法在每次迭代过程中仅需要计算三个矩阵值,有效地降低了计算复杂度。在双向中继窃听系统中,已有文献提出的联合中继和干扰节点选择算法存在着保密中断概率随着信噪比增加而并不趋近于零的问题。为了解决这一问题,论文提出了一种中继聊天(Relay Chatting,RC)安全传输算法。该算法结合了机会中继选择和协作干扰两方面的优势,通过选择多个中继使用分布式波束成形能够消除对合法接收端的干扰影响,而仅降低窃听者的接收性能。理论分析和仿真结果表明在高信噪比下,所提中继聊天安全传输算法的保密中断概率将趋近于零。进一步,在实际系统中,中继可能是非信任的(即中继试图破译信源发送信息),此时没有直传路径的MIMO中继系统并不能进行安全传输。针对此场景,论文提出了两种目的端干扰辅助的安全传输算法,即联合信源、中继和目的端的预编码算法和基于非信任中继天线选择算法。在联合信源、中继和目的端的预编码算法中,提出了一种基于交替迭代的最优信源、中继和目的端预编码矩阵设计方案来最大化系统保密传输速率。其中,最优的信源和目的端预编码矩阵分别通过求解一个凸优化问题获得,而最优的中继预编码矩阵具有闭合解。在基于非信任中继天线选择算法中,根据不同的应用场景,提出了最优策略(Optimal)、最大化非信任中继SINR策略(Suboptimal I)和最小化非信任中继SINR策略(Suboptimal II)等多种天线选择策略。仿真结果表明,提出的联合预编码算法和非信任中继天线选择算法极大地提高了系统安全性能。
[Abstract]:The wireless physical layer security technology uses the randomness, diversity and reciprocity between the channel of the legitimate user and the channel of the eavesdropper to secure the safe transmission of information from the physical layer. It ensures that the listener can not obtain the information sent by the source from the information theory, in the anti eavesdropping, low interception and wireless communication system. Design and other fields have broad application prospects. By using rich space resources, multi antenna technology can not only gain the advantages of two aspects of transmission reliability and effectiveness, but also naturally increase the difference between channels and provide a wide space for the research of wireless physical layer security technology. However, the physical layer security of multi antenna systems is also available. The performance of full transmission depends largely on the channel state information obtained by the transmitter (Channel State Information, CSI). Different channel information conditions and secure transmission in the application scene are the key and difficult problems in the research of the physical layer security technology. The thesis focuses on these key and difficult problems. In the single output (Multiple-input Single-output, MISO) system, the information of the eavesdropping channel is known by the transmitter. The paper studies the power allocation problem of the artificial noise assisted security transmission, and proposes an optimal power allocation algorithm to minimize the probability of the system's secrecy interruption. In this case, the design of artificial noise should be orthogonal to the main channel, and the security performance of the system depends on the power allocation between the useful signal and the artificial noise. The paper derives the closed expression of the secrecy interruption probability, and designs the optimal power allocation algorithm for minimizing the secrecy interruption probability. In the case of the eavesdropper, an optimal power allocation algorithm based on Golden search is proposed. A closed optimal power allocation algorithm is proposed for multiple eavesdropper with single antenna. The optimal power allocation algorithm is converted to a deterministic equation and its closed solution is derived. For the two scenarios mentioned above, the optimal power allocation algorithm is derived. The simulation results show that the performance of the proposed optimal power allocation algorithm is obviously superior to the traditional beamforming algorithm and equal power allocation algorithm. Secondly, the paper proposes an unknown eavesdropping channel information for the transmitting terminal in the Multiple-input Multiple-output Multi-antenna Eavesdropper (MIMOME) model. A secure transmission algorithm combined with Transmit Antenna Selection (TAS) and the maximum ratio combination (Maximal Ratio Combining, MRC). In this scenario, the transmitter needs to obtain the optimal antenna coefficients of the legitimate receiver feedback. In ideal feedback, the system can obtain and unconsider security transmission. The traditional multi antenna system has the same diversity order, and the diversity order has nothing to do with the number of eavesdropper antennas. In the actual system, the ideal feedback is almost impossible because of the influence of channel feedback delay, channel estimation error and so on. The effect of antenna selection and the maximum ratio combination algorithm on the security performance is derived. The closed expression of the probability of obtaining non zero secrecy transmission rate and the probability of secrecy interruption is derived. By analyzing the asymptotically secrecy interruption probability under the high signal to noise ratio (Signal-to-Noise Ratio, SNR), the antenna selection is derived from the antenna when there are non ideal feedback such as time delay or error. The diversity gain will disappear and the system only obtains the maximum ratio of the merge diversity of the receiver. The results of the analysis are more general and can be extended directly to the ideal feedback in the existing literature. The correctness of the theoretical analysis is verified by simulation. Thirdly, the perceptual radio network with multiple input and single output (MISO In the case of Cognitive Radio Network, MISO-CRN), an unknown eavesdropping channel information is transmitted at the sending end. This paper proposes an artificial noise assisted security transmission algorithm. It assumes that the primary and secondary users in the network work in the Underlay mode. The algorithm guarantees the secondary user (Secondary User) from the Perspective of the quality of service (Quality of Service, Qo S). Under the interference temperature of the Signal-to-Interference-plus-Noise Ratio (SINR) and the main user (Primary User, PU), the transmission power of the artificial noise is maximized. When the secondary user has known the main user channel information and the sub user channel information, the theory proves that the beamforming is the best useful signal transmission strategy. At this time, the rank of the input variance matrix is 1., however, when the channel information obtained by the secondary user is error, the artificial noise strategy designed under the ideal channel condition will produce a noise leakage, which seriously affects the receiving performance of the legitimate user. The channel error is modeled as the channel vector error model, the channel variance error model and the statistical channel error model respectively. Under the channel vector error model and the channel variance error model, a robust artificial noise algorithm based on the worst performance is proposed, and the corresponding optimization problem is a NP-hard (Nondeterministic Polynomial hard). Based on the S-Procedure lemma and convex optimization theory, the paper derives the equivalent constraint conditions, and transforms the original problem into 1.5 positive definite programming (Semidefinite Program, SDP), thus the optimal artificial noise input variance design is obtained. The robust artificial noise algorithm with the interruption of probability constraints, the corresponding optimization problem is simplified as a non convex problem with a chance constraint. With the help of the Bernstein type inequality, the paper transforms the complex probability constraint conditions into a series of deterministic inequalities, thus obtaining an approximate solution of the input variance of artificial noise. The simulation results show that the proposed method is an approximate solution. The robust artificial noise algorithm can effectively reduce the influence of channel error under three channel error models. Finally, the optimal relay beamforming algorithm is proposed for one way relay eavesdropping system, two-way relay eavesdropping system and MIMO untrusted relay eavesdropping system. In the unidirectional relay system of single antenna eavesdropper, there is no suitable algorithm to calculate the relay beamforming vector maximizing the secure transmission rate in the Amplifyand-Forward (AF) mode. In the branch and bound (Branch-and-Bound) relay beam forming algorithm, it is proved that the algorithm can obtain the global optimal solution. Considering that the computational complexity of the optimal algorithm is high, a suboptimal relay beamforming algorithm based on the generalized power iteration (General Power Iteration, GPI) is proposed. In the process, only three matrix values are calculated, and the computational complexity is effectively reduced. In the bi-directional relay eavesdropping system, there is a problem that the joint relay and interference node selection algorithm proposed in the literature has the problem that the probability of secrecy interrupts is not close to zero with the increase of signal to noise ratio. (Relay Chatting, RC) secure transmission algorithm. This algorithm combines the advantages of two aspects of opportunity relay selection and cooperative interference. By selecting multiple relays, distributed beamforming can eliminate the interference effect on the legitimate receiver, and only reduce the receiver performance of the eavesdropper. In the actual system, the relay may be untrustworthy (that is, the relay is trying to break the message from the source), and the MIMO relay system without direct path can not be transmitted safely at this time. In this scenario, two kinds of destination interference assistance are proposed in this paper. The secure transmission algorithm, namely, the precoding algorithm of the joint source, relay and destination and the algorithm based on the untrusted relay antenna selection. In the precoding algorithm of the joint source, relay and destination, an alternative iterative optimal source, relay and destination precoding matrix are proposed to maximize the secure transmission of the system. Rate. Among them, the optimal source and destination precoding matrix are obtained by solving a convex optimization problem, and the optimal relay precoding matrix has closed solutions. Based on the different application scenarios, the optimal strategy (Optimal) and the maximization of the untrusted relay SINR strategy (Suboptim) are proposed. Al I) and the minimization of a variety of antenna selection strategies such as the untrusted relay SINR strategy (Suboptimal II). The simulation results show that the proposed joint precoding algorithm and the untrusted relay antenna selection algorithm greatly improve the system security performance.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN915.08

【相似文献】