This PDF file contains the front matter associated with SPIE Proceedings Volume 12027, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

The talk will first review the evolution of intra-data center interconnects over the past decade, then outline future opportunities of optics beyond networking. We will discuss optical and electrical devices, advanced modulation technologies and packaging innovations that deliver better performance and energy efficiency for applications including data center networking and machine learning. Requirements in 400G/800G short reach and long reach optical transceivers, coherent transceiver for intra-datacenter, fiber to machine, and die to die optical interconnects will be discussed.

The explosive growth in data analytics is driving an intensely growing need for compute performance. We will review our ARPA-E PINE ENLITENED experimental testbed and simulation results to motivate performance advantages of disaggregation through the use of flexible photonic interconnect networks for distributed deep learning applications.

Optical routing typically makes use of nodes located at various places around the optical network. However, many locations can be vulnerable to failure due to the potential for power to be interrupted at optical networks, leading to performance degradation, because the network routing node itself typically requires an electrical DC bias voltage and a control signal. In addition, another key challenge is monitoring the state of the node, especially due to bias and temperature drifts that are common in typical Mach–Zehnder interferometers (MZIs) or nonlinear waveguides such as periodicallypoled- lithium-niobate (PPLN) waveguides.

One approach could be to use bidirectional transmission of multiple laser sources over an optical link. Subsequently, (i) arrays of photodiodes that are driven from remote laser sources are used to enable biasing and controlling; (ii) pilot tones are transmitted and their ratios after backpropagation are observed in the transmitter side to monitor the state of operation and adjust the optical powers accordingly.

This paper highlights certain architectures for remotely biasing, controlling, and monitoring a network node that lacks local power, which might enable flexible optical networking. These architectures include: (i) an MZI-based optical switch with optically delivered control and data signals as well as a remotely controlled and monitored optical correlator based on either (ii) delay line interferometer (DLI) or (iii) nonlinear wave-mixing in a PPLN waveguide for a QPSK data.

With constant traffic growth network providers that operate Dense Wavelength Division Multiplexing networks need to quickly find solutions to upgrade their networks. A key challenge represents coping with increased traffic demands while utilizing minimal resources. Simultaneously, the availability of bandwidth variable transceivers offers network planners the possibility to flexibly allocate resources without excessively overprovisioning the network. However, this advantage increases the complexity of planning algorithms: varying data rate, channel bandwidth, modulation format, and corresponding receiver sensitivity must be accounted for in each transceiver configuration along with the available frequencies. Furthermore, the allocation of flexible grid channels requires the calculation of the non-linear interference penalty whenever channels are added, removed, or reconfigured.

We present an algorithmic approach that offers an efficient way of multiple lightpaths reconfiguration, either for increased traffic requests, or lightpath restoration. It permits the building of a feasible optical network design considering the available spectrum, traffic demand characteristics, network topology, equipment configurations, and engineering constraints. We show the employment of methods for calculating and provisioning multiple lightpaths. Our approach is flexible enough to accommodate optical network topologies of different types and sizes. The result is a lightpath configuration that is optimized for spectrum utilization and generalized optical signal to noise ratio (GSNR) degradation.

In the cabin of a modern aircraft, heterogeneous data from different systems are transported using separate and dedicated communication infrastructures. Moreover, the predominant copper-based networks in civil aircraft cabins have become more complex, heavy and expensive. Such architectures are not very scalable and require high maintenance costs with multi-skilled support teams. Therefore, leveraging fibre optic network technology for aviation applications represents one of the most innovative technological developments in the aeronautics industry to increase the network modularity, flexibility and scalability. Indeed, these systems implement all-optical scalable and reconfigurable components that can support very high-capacity point-to-multipoint architectures. In addition, optical networks technology provide a resilient infrastructure over the aircraft life (30 years) and offers the best guarantees to reduce the power consumption, weight1 and overall carbon footprint of the network. Within this frame, the main objective of this work is to identify, describe and analyze an aircraft intra-cabin optical network, based on combined PON and WDM architectures. This solution is relevant in terms of data rate, bandwidth, multiplexing of heterogeneous signals from a large number of data sources and offers flexible and protocol-independent communication links, making it a suitable alternative to aircraft systems. The pro- posed network configuration is analyzed through two options whose performance are evaluated using photonic simulation of realistic optical components, with data rates of 10Gbit/s and under environmental aeronautical constraints, e.g. [-40°C to + 85°C] temperature range and vibration bandwidth up to 2000Hz (DO-160 standard).

As network traffic increases continuously, advances in spectral efficiency with higher-order modulation formats, such as 64 and 256 quadrature amplitude modulation (64-QAM and 256-QAM, respectively), are expected. The device nonlinearity is a major obstacle in achieving a higher-order modulation which requires a high signal-power-to-noise ratio and a low waveform distortion. In this study, we analyze the characteristics of device nonlinearity and its impact on the performance of a coherent transceiver. Furthermore, we demonstrate a low-complexity nonlinear compensation that operates on a symbol rate basis and that can handle higher-order modulations such as 64-QAM or beyond.

To satisfy the ever-increasing traffic demands in a cost-effective manner, the transmission capacity per wavelength must be increased. The use of high-order modulation and high-baud modulation can increase the capacity per wavelength. However, such signal forms are susceptible to IQ impairments created in the transceiver since symbol distance and symbol duration are extremely short. The IQ impairments can be alleviated by adaptive finite-impulse-response (FIR) filters in the receiver. However, transmitter IQ phase imbalance and transmitter IQ power imbalance cannot be eliminated completely in the receiver-side digital signal processing (DSP). Furthermore, to eliminate transmitter IQ skew, additional filters need to be implemented in the receiver DSP. Therefore, transmitter IQ impairments should be calibrated at the transmitter side to enhance demodulation performance and cost efficiency. In this paper, we propose a novel method for measuring transmitter IQ impairments. In our proposed scheme, transmitter IQ impairments are calculated from the tap coefficients of the adaptive filters adopted in the measurement system. In addition, our proposed scheme can estimate transmitter IQ impairments even when other impairments contaminate the signal. With the estimated value, transmitter-side DSP can easily eliminate the transmitter IQ impairments with digital buffers. Simulations using 64 Gbaud 64-QAM signals show that the estimation error of IQ phase imbalance is less than 1.5 degree, that of IQ power imbalance is less than 0.2 dB, and that of IQ skew is less than 0.05 ps even if the signal is contaminated by other impairments.

Raman assisted phase sensitive amplifier has the potential to cover new bands beyond the C- and L-bands with better noise characteristics because its pump light is Raman amplified. In this amplifier, the phase matching condition is achieved for pump/signal/idler by phase tuning the fiber Bragg grating. In addition, since the propagation medium is the same, the phase relationship is maintained, thereby no phase lock loop is required. In this paper, we will report recent experimental progress with Raman assisted Phase sensitive amplifier prototype, which has >25nm gain bandwidth with about 20dB net gain and ~3dB net noise figure.

We review the progress of BDFA development for O-band amplification. Currently, BDFAs can provide up to 35 dB small signal gain, less than 5.5 dB noise figure, and up to 29 dBm output power over 1270-1360 nm. We discuss amplifier design issues including pump wavelength and bandwidth allocation, as well as components performance and availability. Recent transmission experiments including LAN-WDM and CWDM modules reach extension over the spooled fiber and installed cables will be presented. Future milestones in BDF/BDFA development would also be suggested. Since with the implementation of BDFAs the transmission over O-band would no longer be power budget limited, we will briefly discuss the options of chromatic dispersion compensation over O-band.

Due to the development of various mobile applications and various services for 5G and B5G communication environments, data traffic is steadily increasing explosively. In optical access networks, signals modulated on optical intensity have been used. It is difficult to expand the optical intensity modulation according to the increase in data traffic. Therefore, the modulations using other resources have been researched such as optical phase, optical polarization, and optical wavelength. Since modulation using polarization has a simple system compared to other techniques, modulation using optical polarization has widely studied as a multi-dimensional modulation resource. The conventional modulation technique using polarization modulates the signal according to the state of polarization (SOP). However, the technique is very vulnerable to SOP change, especially rotation SOP (RSOP), due to the characteristics of the optical fiber and the optical active devices. In this paper, we propose the polarized intensity rotational frequency shift keying (PIR-FSK) as a novel modulation technique robust to SOP change in optical access network. The PIR-FSK technique modulates the sine and cosine form of FSK RF signal to the intensity of each X and Y polarization of optical carrier, where the total intensity of the modulated optical carrier is in helical form. In addition, since the proposed technique does not affect the total optical carrier intensity, it can be accommodated with the intensity modulation at the same time. We verified experimentally that PIR-FSK offers better immunity to RSOP compared to the conventional PolSK.

The COVID-19 pandemic has greatly increased the internet traffic demand in trunk, edge, and optical data center networks (DCNs), due to large segments of the population staying at home and making use of services such as collaborative information technology (IT) tools, game streaming, and video on demand. This traffic growth is expected to continue after the pandemic due to the next-generation internet of things (IoT) and artificial-intelligence-driven massive machine-to-machine communications. Hence, there is a continuous need to optimize the existing capacity of optical links.

Recently, we investigated non-orthogonal multiple access-carrierless amplitude phase (NOMA-CAP) modulation format as a potential candidate for next-generation, beyond-5G optical short reach networks. NOMA-CAP has some interesting properties, including a high spectral efficiency, the use of the cost-effective intensity modulated-direct detection (IM-DD) scheme, a great degree of flexibility due to its multiband nature, and a resilience to channel non-linearities.

In this invited paper, we review recent experimental demonstrations of NOMA-CAP for three optical network scenarios in the optical C band. Firstly, we demonstrate a DCN in a split-enabled configuration using a 7-core multi-core fiber (MCF) and achieving 630 Gb/s with an electrical bandwidth of 25 GHz. Second, we show a NOMA-CAP-based passive optical network (PON) with a semiconductor optical amplifier (SOA)-based amplification, reaching 25 km of fiber transmission with power budgets of 23.6 and 18.9 dB for data rates of 50 and 90 Gb/s, respectively. Finally, we demonstrate a beyond- 5G optical front-haul network, were a 10 Gb/s PAM4-based PON is upgraded with a 15 Gb/s NOMA-CAP signal.

Next-generation datacenters must address specific space requirements imposed by the tight packing of processing units employing a high-capacity optical network with minimum footprint. We evaluate up to 400G transmission with four multiplexed PAM4 channels considering different carrier separation and optical filter BW to isolate each line of an optical comb. MCF media is proposed to reduce the footprint. The expected MCF crosstalk is calculated for different number of cores, core pitch and lengths. The resulting BER for different OSNR is also analyzed. The suitability of using two comb lines for sub-THz upconversion with optical heterodyning is confirmed by simulation and experimentally.

Future HPC and datacenter systems are expected to contain an increasingly heterogeneous set of compute and memory resources as a strategy to preserve performance scaling in the long term. This heterogeneity, combined with the low average resource usage of today’s systems, motivate resource disaggregation to allow applications to pool and compose no more than the fine-grain resources they require. In this paper, we start by motivating resource disaggregation by observing average utilization and rate of change of memory bandwidth and latency in NERSC’s Cori. We then perform an analytical analysis that quantifies if today’s photonic links and switches would meet key metrics to minimize overhead, and what that overhead will be. Finally, we perform preliminary experiments to demonstrate that even this minimal overhead penalizes application performance in some cases. Our study motivates future work on more aggressive photonic links and switches in order to make resource disaggregation more attractive in future systems.

Optical networks with high capacity, high reliability, and low latency are increasingly needed not only for conventional telecom carrier networks but also for 5G/local-5G/6G mobile networks and intra-/inter-datacenter networks, which can have significantly different network structures and requirements compared with conventional telecom carrier networks. For the optical fiber infrastructure to support such diverse network requirements across various areas, optical node structures must be differently yet precisely optimized. To support such diverse optical node structures in a practical network operation/construction scenario, a functional block-based disaggregation approach that handles optical component devices as a management unit will play a key role. This study presents a functional block-based disaggregation (FBD) model which provides one-to-one correspondence between the model and the actual hardware composition. That is, not only inter-node topologies, but also intra-node topologies can be precisely described. Furthermore, switching functionalities of the individual optical components is described using the integer linear programming (ILP) method in a machine-readable manner. With the ILP formulation, the optical paths between any two points on the network topology can be universally computed for any node structure. Such the detailed and accurate path computation is useful for automatic node function analysis or node configuration table generation, which is indispensable for realizing the functional block-based disaggregation approach in a practical and automated manner. The scalability of FBD model-based path computation considering computationally intensive flexible grid node cases is evaluated over multiple node structures providing different functionalities and node scales.

This paper summaries our recent research on digital radio over fibre (DRoF) based optical fronthaul links and experimentally demonstrates a novel last-mile wireless coverage system incorporating data compression, time-division multiplexing (TDM) based packetization, and wavelength division multiplexing (WDM) based optical transmission. Compression reduces the fronthaul data rate required per service by a factor of 3 when compared with the common public radio interface (CPRI) standard, enabling efficient radio resource distribution over optical fibre infrastructure. The new packetization mechanism and WDM architecture enable fully reconfigurable resource allocation in a fronthaul network for 20MHz-bandwidth RF inputs with 64x64 MIMO carried over an aggregated compressed optical data rate of 32Gbps using 4 wavelengths. The experimental results show over 40dB RF dynamic range with < 8% error value magnitude (EVM) for the 64 quadrature amplitude modulation (64-QAM) input signals across all the WDM channels while the lowest EVM is less than 2%. Meanwhile, this field-programmable gate array (FPGA) based DRoF system allows flexible, software definable and easy-scalable dynamic antenna resource allocation.

In this article, we review various fiber diagnostic methods based on distributed sensing such as optical timedomain reflectometry, optical frequency-domain reflectometry etc., fiber characterization schemes employing digital coherent receivers and various interferometric approaches. We also go through multiple industrialized fiber diagnostic instruments provided by leading manufactures in the installation and maintenance market.

We propose and demonstrate a multiple-input-multiple-output (MIMO) transmission under radio over multimode fiber (RoMMF) configuration. The mode multiplexer and demultiplexer have been configured for the 3x3 MIMO transmission over the GI50 fiber. Three radio signals at a frequency of 6 GHz with a bandwidth of 400 MHz can be transmitted over the 100-m-long GI50 optical fiber with a zero-forcing algorithm using Zadoff-Chu sequences for MIMO demultiplexing. This RoMMF configuration will be applied for the radio relay link in 5G and beyond systems to compensate for the radio dead zones such as in-vehicle and distributed antenna systems in homes and buildings.

A single input to multiple output optical concentrator is presented that can split a millimeter-scale input area into multiple sections and concentrate each section into a separate output. The outputs are designed to have a high coupling efficiency into a multimode fiber array or high-speed photodetector array, with individual output diameters in the range of 50μm to 250μm. The concentrator uses nonimaging tapers that can provide more ideal concentration than a lens. Ray trace simulations are used to evaluate the incidence angle tolerance for linear taper designs with a 3mm length and geometric concentrations up to 64x. The taper simulation results are as good or better than the performance of a ball lens, widely used for coupling to small-area photodetectors. We present simulations for a 1x7 taper design and the first fabrication results using fused silica as the cladding and Norland Optical Adhesive with n=1.51 as the core material. The optical throughput measurements indicate excellent splitting uniformity with a standard deviation of 0.26dB. The multiport concentrator provides an advantageous form factor for splitting, concentrating, and coupling to an output array and is scalable to larger areas and outputs. As such, it is a unique optical device with applications in high-speed optical wireless communication systems as a stand-alone concentrator or secondary concentrator to a primary concentrating lens or mirror.

We report our recent progress on analog radio-over-fiber technology for mobile fronthaul applications in this invited talk from two perspectives: capacity expansion and signal quality enhancement. We first introduce our high-capacity fronthaul transmission experiment achieving >200-Gbps using a commercial transmitter optical subassembly (TOSA) module. Then, we introduce high-fidelity fronthaul transmission using optical angle modulation. Using these key techniques, it becomes possible to implement high-capacity and high-fidelity fronthaul systems.

We propose a simple and yet effective compensation technique to address the potential back-reflected signal coming from the output port of the minimalist-designed, linear optical field frequency discriminator (FD) having a unique folded-back arrangement. Specifically, we consider the engineering problems associated with the combined effects of (i) back reflection signal coming from the output port of the Y-junction divider, and (ii) non-optimized coupling condition of the Y-junction power divider. Our proposed compensation technique solves these problems by using a reflector (or mirror) with variable reflectance positioned at one of the outputs of the single Mach-Zehnder interferometer (MZI) device. By tuning the value of the mirror reflectance to an optimal value, we can compensate for the effects of these above-mentioned problems.

The first transparent Optical-subTHz-Optical link providing record-high line-rates of 240 and 190 Gbit/s over distances from 5 to 115 m was recently demonstrated. The link has been based on a direct data-conversion from optical to subTHz using a > 500 GHz plasmonic Mach-Zehnder modulator. We discuss the potential of plasmonic devices in subTHz wireless links to efficiently bridge optical fiber networks.

For the past 50 years, mobile bandwidth requirements have evolved from voice calls and texting to ultra-high-definition video and a variety of augmented reality/virtual reality applications. Expanding machine-to-machine applications, such as smart meters, video surveillance, healthcare monitoring, connected drives, and automated logistics, contribute in a major way to device and connection growth and push the expansion of data center infrastructure. The popularity of pluggable modules will continue as they take advantage of 100G single-wavelength optics already proven in 400GbE systems and thus can be technically and cost-effectively implemented in new form factors for 800G modules. However, current form factors will be limited in their ability to support 1.6T and higher capacities in terms of the required electrical and optical densities, thermal issues, and power consumption. As a result of discrete electrical device implementation, power dissipation and thermal management are becoming limiting factors for future pluggable optics. Co-Packaged Optics (CPO) is a new approach that brings the optics and the switch ASIC close together and aims to overcome the challenges. Furthermore, CPO technology is considered a new deployment model of the whole ecosystem and an alternative to pluggable optics. The industry is working on heterogeneous integration of InP lasers directly onto silicon chips enabling scalable integration and elimination of the cost and complexity. Revenue generated by the CPO market reached around $6M in 2020 and is expected to reach $366M in 2026 at a 101% Compound Annual Growth Rate for 2020-2026. This growth is driven by substantial energy (<30%) and capital expenditure ($/Gbps) savings over pluggable optics.

In order to meet the recent vast demands of ICT bandwidth, next -gen 1 Tb/s-era high speed optical transceivers, such as 800Gbps and 1.6Tbps, began to be discussed in principal standards bodies in early 2021. Especially IEEE802.3, OIF and some specific MSA organizations have played important roles in the industry so as to lead the proper concepts and designs of the leading-edge high speed optical transceivers. Reviewing the recent activities of these standardization bodies, the emerging technical solutions such as Co -Packaging Optics, in addition to the transceiver pluggable solutions, are also to be examined and reported.

Laser-based visible light communication (VLC) system has been confirmed as a vital part of current VLC technology due to its high data rate and extended range. However, the application of the conventional laser VLC system is very much limited by its high directivity and thus requests a complex tracking mechanism. In this work, we present a laser-based white light VLC system with ultra-wide field-of-view (FOV) exceeding 150°, which paves the path for future deployment of laser-based VLC systems.

This system consists of two blue laser diodes pumping YAG ceramic phosphor plate in a reflection configuration as the transmitter and a silicon photomultiplier (SiPM) as the receiver. An engineered diffuser is integrated into the transmitter end.

The system modulation bandwidth is measured to be 200 MHz, which is limited by the frequency response of the SiPM. When the forward error correction (FEC) threshold bit error rate (BER) of 3.8 × 10^-3 is considered, data rates up to ~400 Mbps have been achieved within 88° using the adaptive bit loading discrete multi-tone (DMT) modulation scheme. When extending the FOV to 152°, the transmission data rate remains exceeding 200 Mbps. Our work confirms the possibility of constructing the wide-angle and ultra-high-speed laser-based MIMO VLC system.

Increasing interest in indoor navigation has recently been generated by devices with wireless communication capabilities that enabled a wide range of applications and services. The rise of the Internet of Things (IoT) and the inherent end-to-end connectivity of billions of devices is very attractive for indoor localization and proximity detection. Other fields, such as, marketing and customer assistance, health services, asset management and tracking, can also benefit from indoor localization. Different techniques and wireless technologies have been proposed for indoor location, as the traditional Global Positioning System (GPS) has a very poor, unreliable performance in a closed space. The work presented in this research proposes the use of an indoor localization system based on Visible Light Communication (VLC) to support the navigation and operational tasks of Autonomous Guided Vehicles (AVG) in an automated warehouse. The research is mainly focused on the development of the navigation VLC system, transmission of control data information and decoding techniques.

As part of the communication system, trichromatic white LEDs are used as emitters and a-SiC:H/a-Si:H based photodiodes with selective spectral sensitivity, are used as receivers. Through the modulation of the RGB LEDs, the downlink channel establishes an infrastructure-to-vehicle link (I2V) and provides position information to the vehicle. The decoding strategy is based on accurate calibration of the output signal. Characterization of the transmitters and receivers, description of the coding schemes and decoding algorithms will be the focus of discussion in this paper.

We explore the use of 785 nm near-infrared laser source for the giga-bit class indoor visible light communication system compatible with the lighting fixture. The NIR laser light can easily hide behind the white LED light. For the proof-of-concept, a hybrid source is used as light source which consists of a white LED ring and a NIR laser diode. To avoid the non-uniform distribution of color and to improve the illumination performance, light is diffused using diffuser plate with full-width half-maximum (FWHM) of 20^o. A glass lens is used to direct the light towards the receiver. White light from the LED with and without NIR is demonstrated with International Commission on Illumination (CIE 1931 XYZ color space) (.3186, .3293), a correlated color temperature of ~6100 K, and a color rendering index of 74.2 to provide an illuminance of 29.7 lux. To check the eye safe point-to-point communication capacity over a link length of 2- meters, 16QAM-OFDM format is used to directly modulate the NIR laser diode with the data rate of 3 Gbps (Bit-errorrate (BER) < 3.8×10^-3). Such a high-speed LED-LD mixed VLC system without any channel interference can be used to simultaneously provide data transmission and white lighting in an indoor environment.

To support people’s wayfinding activities in crowded buildings this paper proposes a method able to generate landmark route and alert instructions using Visible Light Communication .The system is composed of several ceiling luminaries transmitters which send the map information, alerts and the path messages required to wayfinding. The system informs the users, in real time, not only of the best route to the desired destination, through a route without clusters of users, but also of crowded places. Data is encoded, modulated and converted into light signals. An architecture based on a mesh cellular hybrid structure was used. The luminaires are equipped with one of two types of nodes: a “mesh” controller that connects with other nodes in its vicinity and can forward messages to other devices in the mesh, acting like routers nodes in the network and a “mesh/cellular” hybrid controller, that is also equipped with a modem providing IP base connectivity to the central manager services. Each luminaire for downlink can forward messages to other devices or to the central manager services. Mobile optical receivers, collect the data, extracts theirs location to perform positioning and, concomitantly, the transmitted data from each transmitter. Uplink transmission is implemented and the best route to navigate through venue calculated. Buddy wayfinding services are also considered. The results show that the system makes possible not only the self-localization, but also to infer the travel direction and to interact with information received optimizing the route towards a static or dynamic destination.