Physical Layer Security of Multimode Optical Fiber Transmission Systems

单模光纤网络促进了现代互联网的出现。应用于多个域（时间、偏振、波长）的复杂多路复用技术实现了快速数据调制。在需要不断增加网络容量的数据中心时代，多模光纤（MMF）通过额外利用空间域提供了增加可能的数据速率的潜力。此外，通过适当的发射机侧波前整形，可以利用MMF内的复杂光传播来提高信息安全性。这种方法被称为物理层安全（PLS），并首次在我们的实验室对MMF进行了实验研究。与高层的经典密码技术相比，PLS生成代码是基于物理信道特性的。此外，PLS与光学数据传输的经典组件（放大器、中继器）兼容。如果两个节点（Alice和Bob）之间的通信信道是已知的，例如通过测量传输属性（传输矩阵TM），则可以通过随后的逆预编码来实现相对于潜在窃听者（Eve）的SNR优势。当信道被校准到Bob时，Eve遭受模式相关的损耗，这允许Alice通过合适的编码算法（人工噪声，“窃听编码”）降低Eve的信道质量以实现对Bob的优势。对PLS的第一次基本调查表明，10米阶跃指数光纤的方法的潜力。它们表现出强烈的模式混合。然而，在传统的MMF通信网络中，在长得多的距离上使用具有相对较少混合的渐变折射率（GRIN）MMF。在长距离时，会发生时间变化，这就是为什么校准和通信必须在信道的有限相干时间内发生的原因。因此，我们的目标是一个实时校准的神经网络的基础上，在通常的传输距离调查PLS在MMF。我们也在调查偏振对PLS的影响，以及偏振串扰的操纵是否会导致信息安全的增加。使用第二光路和copula理论，将创建统计模型来研究正交偏振态之间的随机依赖关系。这可以用于确定保密性的增加，例如通过“密钥生成”。

Singlemode optical fibre networks have facilitated the emergence of the modern internet. Sophisticated multiplexing techniques applied to several domains (time, polarisation, wavelength) enable fast data modulation. In the era of data centres, which require a continuous increase of network capacities, multimode fibres (MMF) offer the potential to increase possible data rates by additional exploiting the spatial domain. In addition, it is possible to exploit the complex light propagation within MMF by suitable transmitter-side wavefront shaping to increase information security. This approach is called Physical Layer Security (PLS) and was first investigated experimentally on MMFs by our lab. In contrast to classical cryptographic techniques on higher layers, PLS generates codes are based on physical channel properties. Further, PLS is compatible with classical components of optical data transmission (amplifiers, repeaters). If the communication channel between two nodes (Alice and Bob) is known, for example, by measuring the transmission property (transmission matrix, TM), an SNR advantage over a potential eavesdropper (Eve) can be achieved by subsequent inverse precoding. While the channel is calibrated to Bob, Eve suffers mode-dependent losses that allow Alice to degrade Eve's channel quality to achieve an advantage for Bob via suitable coding algorithms (artificial noise, "wiretap coding"). The first fundamental investigations towards PLS demonstrated the potential of the approach on 10 m step-index fibres. These exhibit strong mode mixing. However, in conventional MMF communication networks, graded index (GRIN) MMFs with comparatively less mixing are used over much longer distances. At long distances, time variance occurs, which is why calibration and communication must occur within the finite coherence time of the channel. Therefore, we aim for a real-time calibration based on a neural network to investigate PLS in MMF at usual transmission distances. We are also investigating the influence of polarisation on PLS and whether manipulation of polarisation crosstalk can lead to an increase in information security. Using a second light path and copula theory, statistical models will be created to investigate stochastic dependencies between orthogonal polarisation states. This can be used to determine the increase in secrecy, e.g. through "secret key generation".