Poster Sessions
Fibre Optical Parametric Amplifiers for QAM signal amplification with optimised SBS suppression technique
Mariia Bastamova, Vladimir Gordienko, Nick Doran, Andrew Ellis, Aston Institute of Photonics Technologies, Aston University
Fibre optical parametric amplifiers are promising technology for future fibre optic communications for their virtually wavelength unrestricted abilities for ultra-broadband amplification, phase-sensitive and transient-free amplification. Furthermore, modern FOPAs are polarization-insensitive, with low noise and low crosstalk. Currently, the Stimulated Brillouin Scattering (SBS) suppression techniques (commonly pump dithering) mainly contribute to signal degradation.
Whilst the systems impact of the pump dithering on a QAM signals is enhanced, general guidelines for the design of the dithering waveform remain to be minimizing both power spectral density to maximize SBS threshold and minimizing spectral width to reduce required-OSNR penalty.
The demonstrated required signal OSNR penalty has been decreased by a third with only a modest reduction in SBS threshold by using the dither signal with the largest number of the most narrowly spaced tones by operating with a minimum tone spacing close to the SBS linewidth rather than a conventional spacing of 100MHz.
ML-aided mitigation of inter-channel impairments in WDM systems
Nelson Castro, Andrew Ellis, Stylianos Sygletos, Aston University
Most machine learning (ML) models for nonlinearity mitigation in transmission systems have been developed for processing single-channel signals. Unfortunately, the improvement margins they offer to practical WDM scenarios are limited since they cannot address the inter-channel impairments affecting these systems. A potential alternative are Multiple-Input-Multiple-Output (MIMO) schemes, which can effectively estimate the nonlinear phase shifts caused by neighboring channels. However, the viability of MIMO approaches presents serious obstacles. For instance, traditional MIMO schemes need to operate with a very large number of steps, resulting in a prohibitive computational complexity. Nevertheless, it is possible to design ML-aided MIMO models which, through the optimization of their parameters, can potentially require a significantly lower computational effort compared to traditional schemes.
Moreover, this parameter optimization can inform the design of optimized architectures. Through numerical simulations, we show that our models significantly improve the performance of WDM systems while operating with a reduced computational load.
Hybrid-Screens for Simulating Optical Turbulence in Complex Environments
Ultan Daly, PhD Student, University of Glasgow
Proposed urban free-space optical communication channels, which can have complex turbulence distribution profiles, can require expensive site surveys to determine expected system requirements. Alternatively, system requirements can be determined by simulating optical propagation through CFD simulations of a proposed channel. However, the required CFD resolution to capture all relevant optical turbulence length scales makes this approach unfeasible for many channels. In this work we create Hybrid-screens, which utilise a low-resolution Large Eddy Simulation to capture the effect of low-frequency eddies, and the turbulence profile distribution of a channel. Statistically representative high-frequency components of the screens are then extrapolated under the assumption of Kolmogorov turbulence. We investigate the power spectral densities of the generated screens and find good agreement with the Kolmogorov power law. Additionally, we analyse the results of optical propagation through a simulated channel using Hybrid-screens.
Quantum Radio for future network applications
Daniel Gilks, Senior Manager, Reserch, BT Group Plc
We present RF to optical conversion using a ‘Quantum Radio’ receiver. By exploiting two photon processes in highly excited atomic vapours, we can embed a narrowband digital encoding onto an optical transmission path. Introducing a radio to optical converter which can be physically separate from optical to electronic/DSP componentry we can anticipate a range of novel network architecture options with radio to optical components used to geofence industrial and transport sites or to follow roads or tunnels. Furthermore, by selecting appropriate atomic electronic states this system could operate over a wide and rapidly reconfigurable frequency range.
Fibre Optic Parametric Amplifiers for Communications
Vladimir Gordienko, Research Fellow, Aston University
Fibre optical parametric amplifiers (FOPAs) have a great potential to improve transmission capacity of future optical communications due to theoretically unconstrained gain bandwidth not confined to any band, ability of virtually noiseless phase-sensitive amplification and lack of transients. Recent advances of the FOPA technology have resolved its key issues, whereas the most important advancement was to enable a practical polarisation-insensitive amplification. Consequently, we have demonstrated a robust fully automated (black-box) FOPA operation in the C and L bands simultaneously with gain of ~20dB and output power over 23dBm when amplifying polarisation-multiplexed WDM QAM signals as well as bursty traffic. Additionally, we have demonstrated FOPA to amplify WDM signals in the S band and across continuous bandwidth of 40nm. Finally, we have demonstrated a power budget improvement of a transient-sensitive link by up to 8dB when employing a FOPA with noise figure of ~6dB as a drop-in replacement of an EDFA.
One-Dimensional and High-Dimensional Non-Coherent Detection Methods for Ultraviolet Communications
Wenxiu Hu, Cenqin Jin, Mark S. Leeson, Tianhua Xu, School of Engineering, University of Warwick; Zhuangkun Wei, School of Aerospace, Transport, and Manufacturing, Cranfield University; Qiankun Li, School of Physics, University of Electronic Science and Technology of China,
Ultraviolet communication (UVC) has strong particle character, which benefits its non-line-of-sight (NLOS) transmission but leads to strong inter-symbol interference (ISI). Existing coherent signal detection schemes are over-reliance on the accuracy of the channel estimation, and non-coherent schemes cannot provide high detection accuracy in the presence of inter-symbol interference. We propose a novel non-coherent paradigm by extracting the UV signal features which are irrelevant to the ISI. Leveraging such features, two types of non-coherent schemes are developed, namely one-dimensional (1D) and high-dimensional (HD) detection. One-dimensional non-coherent scheme combines the signal features linearly via optimal weights; and high-dimensional non-coherent scheme leverages a HD construction of the UV signal features, transforming the ISI caused sequential detection into an ISI-released HD detection framework. Our results shown that these novel schemes have merits of high detection accuracy and low computation complexity, compared to commonly used maximum-likelihood sequence detection (MLSD).
Extended Kalman filter in optical fibre communications for joint compensation of phase and amplitude noise
Cenqin Jin, Wenxiu Hu, Mark S. Leeson, Yunfei Chen, Tianhua Xu, School of Engineering, University of Warwick; Mingming Tan, Aston Institute of Photonic Technologies, Aston University; Nikita A. Shevchenko, University College London; Qiankun Li, School of Physics, University of Electronic Science and Technology of China
Morden long-haul optical fibre communication systems can be heavily degraded by linear and nonlinear transmission impairments, e.g. chromatic dispersion, polarization mode dispersion, laser phase noise, equalization enhanced phase noise, amplified spontaneous emission noise, fibre nonlinear interference. In this work, extended Kalman filter has been investigated to improve transmitted signal quality in long-haul Nyquist-spaced optical fibre communication systems. Strong capability of the extended Kalman filter on joint compensation of the phase and the amplitude noise has been observed and demonstrated. The performance of linear Kalman filter, Viterbi-Viterbi estimator and pilot-aid carrier phase estimator is also studied as benchmarks. Electronic dispersion compensation and digital nonlinearity compensation algorithms have also been applied in considered systems with reasonable laser phase noise from transmitter and local oscillator sources taken into account, to investigate the performance of extended Kalman filter in practical transmission scenarios.
Optimal Detector Collocation in QKD Networks using Twin-Field QKD
Vasileios Karavias, PhD Student, University of Cambridge; Andrew Lord, BT; Mike Payne, University of Cambridge
We developed a mixed integer linear program to optimise the detector placement cost for collocated detectors using TF-QKD and showed that the model can deal with large graphs of circa 100 nodes in under a minute. We considered networks where switches allow detectors to be shared between users. We used this model to investigate the effects of increases in switch loss and switching calibration time and showed that the cost of building a quantum network increases rapidly with these parameters. We showed that you can reduce the cost of the network by more than a factor of 3 by reducing the detector dark count rate at the expense of efficiency. We investigated the benefit of using cooled SNSPD detectors compared to SPADs and showed that on larger graphs the use of SNSPDs is favoured, even when the cooling cost of the node is 10 times higher than using a SPAD.
Optical Technology: The Unsung Hero Meeting Next Generation Infrastructure Requirements
Raza Khan, Semtech Corporation, Senior market manager for Semtech’s Signal Integrity Products Group, Semtech Corporation
All infrastructure must start with a strong foundation, including our telecommunication networks. The pandemic ushered in new connectivity requirements, as remote work became a new normal for many. Furthermore, the convergence of new IoT applications and 5G use cases—such as AI, cloud computing, autonomous driving, precision farming and more—continue to demand a fast and reliable network capable of supporting vast amounts of data.
The success of 5G will be dependent upon its framework, which starts with optical technology. Optical technology is often overlooked as the key to making fast and reliable 5G a reality, yet, it will play a crucial role in delivering the high-bandwidth and low-latency requirements needed to support 5G, 5.5G, 6G and beyond. Additionally, the high-bandwidth and low-latency characteristics of optical technology will facilitate future key applications in smart cities, including smart grids for enhanced energy efficiency, sensory networks to improve public services, smart mobility for optimizing traffic and available city-wide network connections.
Machine learning enabled compensation of phase-to-amplitude distortion due to imperfect pump-dithering in optical phase conjugated transmission systems
Long H. Nguyen, Sonia Boscolo, Andrew D. Ellis , Stylianos Sygletos, Aston Institute of Photonic Technologies, Aston University
We propose a machine learning-based digital signal processing technique to mitigate the impact of imperfect counter-phasing pump dithering in optical phase conjugated transmission systems. Contrary to state-of-the-art approaches that can deal only with the residual phase distortion, our scheme also tackles the corresponding phase to amplitude transformations that have occurred in the dispersive channel. With the use of an adaptive configuration, we first track and compensate the dither induced phase deviations on the received signal and subsequently extrapolate and remove their amplitude impact. Through extensive numerical we explore the operational margins of our approach in terms of system transmission distance, constellation order and pump-phase mismatch level, and demonstrate significant performance improvement against current schemes.
Optical Amplifier Optimisation in Ultra-Wideband (UWB) Systems using Reinforcement Learning
Shabnam Noor, Research Associate, Aston University
One of the key challenges in UWB systems is the development of optical amplifiers that can provide acceptable gain over wide bandwidths in a controlled and rapid manner. In the case of All-Raman amplification, multiple pumps, and thus, several input parameters need to be controlled. Moreover, in hybrid amplification, different types of amplifiers have different requirements. This becomes an even bigger challenge in dynamic optical networks under, e.g., various channel loading conditions. To this end, this work investigates a Reinforcement Learning framework, which uses real-time training data and interactive feedback, to achieve the optimum gain and NF by learning the best possible combination of input parameters. Unlike existing approaches, this does not require large amounts of training data and is not limited to a particular type of network scenario or amplifier. The approach will be verified in VPI-Python co-simulations and on All-Raman, as well as hybrid EDFA-Raman amplifier experimental testbeds.
UKQNtel and Modelling multiplexed QKD networks
Joseph Pearse, PhD Student, University of York
I will describe the UKQNTel network, a QKD network between Cambridge University and Adastral Park with 3 trusted nodes and co-propagated classical and quantum data.
I will then describe the development of a model for QKD systems where classical and quantum data is co-propagated. By inputting parameters of the QKD links such as fibre lengths, powers, detector statistics, and QKD protocols, as well as details of the network structure, the model generates the QKD performance for each link in the network; including QBER, Visibility, and maximum Secret Key Rate. It can then determine the maximum Secret Key Rate between any two nodes in the network.
The model includes greater detail than previous simulations, especially in regard to Raman scattering, visibility, and detection. This greater detail allows the model to be applicable over a greater range of quantum and channel wavelengths, allowing channel wavelengths between 1200 nm and 1900 nm.
To corroborate the model, I will present data from real QKD systems, including UKQNtel.
Random Forests for Equalizers for Coherent Optical Transmission lines
Egor Sedov, PhD student, Aston University
Machine learning techniques are rapidly increasing their influence in the telecommunications sector due to their ability to effectively mitigate transmission impairments. In our study, we demonstrate the possibilities of using Random Forest to reduce the nonlinear distortion of a signal after processing at the receiver. Random Forest is successfully used in classification and nonlinear regression problems where standard methods do not provide the required accuracy. The ability to increase the number of Decision Trees helps to reduce overfitting and increase the final accuracy of the algorithms. Moreover, the addition of new Trees can be used to adapt to changing system parameters and reduce the computational costs of transfer learning.
Micro-LED Projector Based 100-Mb/s PAM4 Optical Camera Communications
Yingjie Shao, Researcher, Centre for Applied Photonics, Fraunhofer UK Research Ltd
In recent years, optical cameras have increased significantly in volume worldwide, not only on consumer mobile devices but also in industrial applications and smart infrastructure. By employing LEDs or display screens as transmitters, optical camera communication (OCC) can be easily established using existing hardware for various scenarios such as indoor positioning and vehicle-to-infrastructure communications.
We present an OCC system based on a 128×128 array CMOS-driven micro-LED projector with 4-level pulse amplitude modulation. Using a high-speed global shutter camera, a 100-Mb/s OCC system is achieved to support high-speed applications. The demonstration also indicates the ability to utilize the grayscale brightness control capability of the micro-LED projector chip and predicts in excess of 1-Gb/s potential data rate OCC system at full resolution of the chip. Moreover, practical considerations are addressed. Over 10-Mb/s demonstrations using the smartphone camera and behind-the-pattern communication link are achieved.
Practical Laser Stabilization Outside The Lab – Two Photon Systems
Qiushuo Sun, Research Manager, BT Group PLC
Locking and stabilizing laser sources to absolute atomic references is essential to support cold atom and neutral ion-based technology applications (quantum radio, quantum gravitometers, quantum sensors). Whilst frequency combs and high finesse cavities offer excellent laser stabilisation in a lab setting it remains challenging to imagine such tools making their way into mainstream fibre networked applications, due to both prohibitive costs and the limitations of stimulated Brillouin scattering. Ground state atomic transitions are generally well addressed with off-the-shelf saturation absorption spectroscopy systems. However, this limits the lockable optical frequencies (based on available atomic vapours). We present 3 methods for two-photon optical laser locking to an excited transition using a narrow electromagnetically induced transparency (EIT). We implement AC Zeeman saturation EIT locking, Dichroic atomic vapour spectroscopy EIT locking, and electro-optical modulator (EOM) sideband-locking to implement scalable laser locking systems which could be embedded within mass market network technology.
Quantum Random Number Generator Based on On-Off-Keying Encoding
Hamid Tebyanian, Department of Mathematics, University of York, Heslington
This paper presents a semi-device independent protocol for quantum randomness generation constructed on the prepare-and-measure scenario based on the on-off-keying encoding scheme and homodyne and heterodyne detection schemes. The security estimation is based on lower bounding the guessing probability for a general case and is numerically optimized by utilizing semi-definite programming. Additionally, we introduce a straightforward optical setup which can be implemented via commercial off-the-shelf components.
Optical Atomic Clocks for Time Dissemination in Telecom Networks
Lakshmi Rajagopal, Early Stage Researcher, Optical Networks Research, Applied Research, BT Group
According to the growing end user demands, time and frequency need to be distributed more precisely across the networks. The growing demand on stringent timing requirements make us think of the next best solution that is better than existing. This can either be with respect to new technologies for precise time source or new technologies to transport time precisely across the networks. The aim of the model is to understand how time can be disseminated across a network. Conventional architecture and protocols used to deliver time and frequency is reaching its limits, as demands from the evolving networks raise to below sub-nanosecond levels. Proliferation of devices across the network is a challenge to distribute time through satellite technologies. The model helps to understand how time can be distributed across the network from highly precise optical atomic clocks. Applications requiring precise time emerge every day. A hierarchical topology of the network is implemented in the model, which gives an impression of the core and tier nodes that exists within a real timing network architecture. The model is built upon a UK national level network, giving an idea on how good the optical atomic clocks can be for time dissemination at the national level. With the model, it becomes easy to understand the flexibility of the architecture with time error requirements. Evaluating the impact of different types of asymmetries on time transfer, lifts the model to fit best with a practical scenario.