Optical digital-to-analog converters (DAC) offer power-efficient amplitude multiplexing in the optical domain. They allow simplification to complex electronic circuits as the binary signals are directly fed to the Mach-Zehnder modulators (MZM). Drive circuits with a large effective number of bits (ENOB) and complex power-hungry electronic predistortion of the MZMs nonlinear transfer characteristic are eliminated. This results in a power advantage and smaller electronics, ultimately delivering an improved signal-to-noise and distortion ratio (SNDR). As oDACs can be operated by means of direct digital drives (DDD) with minimal signal processing, they are ideal for highest speed operation. When implemented with terahertz-bandwidth plasmonic modulators, they may provide highest symbol rates on a small footprint with the least power consumption. We demonstrate the first plasmonic oDAC implementation for IM/DD and coherent transmission, operating at up to 420 Gb/s line rate without SNR penalty. We show energy-efficient amplitude multiplexing in the optical domain at symbol rates up to 180 GBd.

ABSTRACT The continuous increase of data traffic demands necessitates evolution of optical communication systems towards larger capacities enabled by highly performing optical transmitters that generate multilevel and multidimensional ultrahigh data rate signals. Commercially deployed transmitters rely on high‐order electronic digital‐to‐analog converters (eDACs) that drive optical Mach–Zehnder Modulators (MZMs) under particular biasing/driving conditions. However, these conventional transmitters face bandwidth and power consumption limitations arising from the eDACs and the nonlinearity of MZMs transfer function among others. Recently, alternative transmitter architectures have been proposed to perform direct digital to optical conversion, by relying on the utilisation of lower‐order eDACs/drivers in combination with an increased number of MZMs operated with different biasing/driving conditions compared to the conventional transmitters, which are configured in either serial cascade or parallel stacking configurations. These novel transmitter architectures are referred as serial or parallel ‘Optical DACs’ (oDACs) enabling ultrahigh data rate and improved signal quality. In this article, we first discuss the issues of conventional optical transmitters and subsequently analyse selected oDAC architectures that solely rely on simple NRZ drivers or PAM4 eDACs. Simulation studies demonstrate their superior bit rate scalability (up to 3.2 Tbps) and analysis of power consumption reveals savings of about 40% comparing with conventional architectures.

The emergence of distributed Generative AI (GenAI) necessitates a paradigm shift in optical transport to accommodate periodic, massive “elephant flows” between edge GPU clusters and core “Giga-scale AI Super-factories”. Traditional ROADMs based on single-granular Wavelength Selective Switches (WSS) face significant scaling and SWaP-C limitations when handling such high-intensity workloads across Ultra-Wideband (UWB) and Space Division Multiplexing (SDM) layers. This work evaluates the performance of the FLEX-SCALE Multi-Granular Optical Node (MG-ON)—a three-layered hierarchical architecture capable of switching at fiber, flexible waveband, and wavelength granularities—specifically for distributed GenAI re-training. Utilizing Photonic Integrated Circuit (PIC)-based flexible WBSS, the architecture targets a net node throughput of 10 Pb/s. Our simulation results demonstrate that the multi-granular approach significantly outperforms legacy single-granular WSS architectures by substantially reducing total network cost and the number of required switching components. By routing AI-driven data bursts as coarse-grained wavebands, we maintain the Generalized Signal-to-Noise Ratio (GSNR) required for high-capacity transmission while enabling the millisecond-level latency vital for AI model synchronization. These findings validate multi-granular, multi-band networking as a critical enabler for the 6G-era AI infrastructure.

As bandwidth demands continue to rise, wavelength-level granularity in wavelength division multiplexing (WDM)-based optical networks becomes a limiting factor, as scaling such networks places increasing load on the WDM nodes. This motivates a shift toward coarser switching granularities. Node architectures will thus evolve from wavelength switching to waveband-selective switching and eventually to fiber cross-connects as capacity requirements grow. Multi-granular optical node (MG-ON) architectures support this transition by integrating wavelength, waveband, and spatial switching. However, during dynamic routing and resource assignment, MG-ON architectures may suffer from inefficient utilization of wavelength, waveband, and spatial resources. This can lead to increased connection request blocking. To address this, we propose a dynamic routing and resource assignment algorithm combined with an augmented optical grooming strategy for MG-ON networks. We evaluate the proposed approach in terms of connection accommodation, switch count, and per-switch port requirements across multiple network configurations. Its performance is compared with baseline WDM networks and existing MG-ON grooming strategies.

We investigate energy-efficient, traffic-adaptive resource provisioning for 6G x-haul networks by integrating a reconfigurable low-loss switching sub-system (referred to as an interlacer) into the TEFNET24 reference network topology from Telefónica. The interlacer combined with the digital subcarrier multiplexing (DSCM) transceivers forms a WDM-PON approach, enabling subcarrier-level switching at the network access, providing lower distribution loss than conventional power splitter (PS)-based TDMA-PON approaches. Moreover, under the practically low crosstalk levels reported for our interlacer implementations, the reduced distribution loss can be traded for increased modulation cardinality (e.g., 16QAM to 64QAM), allowing for reducing the number of allocated transceivers. We further examine a machine learning (ML)-driven framework for dynamic capacity provisioning and validate it using a real-world operator trace mapped onto the national-scale TEFNET24 topology, capturing 4G/5G load patterns with projections toward 6G. We compare static, semi-static, and fully dynamic strategies and show that ML-aided dynamic provisioning reduces transceiver overallocation by up to 63.7% relative to static methods, with the strongest gains during off-peak hours when traffic variability is highest. Finally, we quantify two complementary energy-saving mechanisms enabled by dynamic operation—subcarrier deactivation and transceiver sleep mode—with the latter yielding up to 21% total network energy savings while maintaining quality of service under time-varying demand.

In this article we study pilot-tone-aided all-optical polarization demultiplexing for short-reach IM/DD links. We evaluate how the pilot SNR and the measurement time determine the convergence time and the achievable SOP tracking speed. A simultaneous perturbation stochastic approximation (SPSA) controller is implemented to enable fast adaptation using only two metric measurements per iteration, independent of the number of tuning parameters.

We present a comprehensive system-level comparison of optical digital-to-analog converter (oDAC) architectures-serial, multi-parallel, and 2S2P-against conventional MZM transmitters for data-center links, accounting for realistic electronic DAC impairments including bandwidth limitations and timing jitter.

The evolution of passive optical networks (PONs) is rapidly progressing, with the next generation aiming to surpass 100Gb/s/ λ and target 200Gb/s/ λ —a typical fourfold increase over the current 50Gb/s/ λ power splitter (PS)-based PON standard. This paper outlines current research and development challenges in PONs and explores potential advancements using programmable photonic Nyquist-shaped Interleaver (NS-INT) filters paired with coherent point-to-multipoint (P2MP) digital subcarrier multiplexing (DSCM) transceivers. Specifically, we evaluated, based on analytical estimations and simulations, two NS-INT implementations for mobile transport: one based on ring resonators-assisted filters and a second utilizing an optical 18-tap cascaded asymmetric Mach–Zehnder Interferometer (aMZI). We compare their effectiveness in optimizing filtering and minimizing crosstalk (XT) across various coupling coefficient scenarios. Furthermore, we assess the feasibility of these designs by examining their performance under XT and signal impairment caused by filter narrowing effects, particularly in the presence of signal-to-filter center frequency misalignment. Bit error rate (BER) evaluations demonstrate a tolerance to frequency misalignments of up to 3 GHz across all architectures. Among them, our cascaded aMZI design exhibits superior performance, attributed to its inherent optical finite impulse response (FIR) filter characteristics as a minimum-phase filter.

We investigate the scaling potential and compute critical performance metrics of a three-tiered hierarchical optical node architecture. The optical node incorporates novel Photonic Integrated Circuit (PIC)-based WaveBand Selective Switches (WBSSs) to implement flexibly-defined band switching, across spatial lanes and degrees of connectivity, ranging from entire optical fibers, flexible-defined bands to individual wavelengths. We also develop a network simulator considering node's multiple switching layers, multiple/portion of bands and all key optical transmission parameters. Simulations reveal high optical Signal to Noise Ratio (OSNR) and low Bit Error Rate (BER) values, particularly for the Full Fiber Switching (FFS) scenario. Moreover, we analyze trade-offs among scalability, component number and complexity, and throughput across various configurations and traffic scenarios. Our results confirm high levels architecture adaptability and efficiency in addressing the evolving demands of future optical networks.

This paper focuses on the application of an analytical model for faster and reliable transmission characterizations of novel optical networks employing Digital Subcarrier Multiplexing transceivers. This transmission model is thought to be applied in metro-access networks, for which it is first demonstrated for single links. In this work, the model is applied to analyze the system-level performance of a link, including, as a device under test, a 2-Ring Assisted Mach-Zehnder Interferometer (RAMZI) interlacer filter used to perform optical routing at the subcarrier level, and to quickly identify the impact of theparameters of the device in the overall transmission.

Point-to-multipoint front-haul networks provide shared access to edge sites, with capacity and routing constraints. The front-haul capacity can be multiplied by introducing several spatial feeds and means for associating input feed to specific remote units at the edge via switching/routing. The front-haul is empowered using our spatially-diverse point-to-multipoint architecture, and we compare several technologies for its implementation. We introduce subcarrier-level rapid networking and spatial-feeder enhancements with an innovative four-port optical interlacer, increasing capacity and networking flexibility.

State-of-the-art optical transceivers are limited by DAC driver performance degradation at ultra-high data rates. To enable enhanced data rates in short-reach coherent communication, there is a growing need for energy-efficient, compact, low-loss, highly linear broadband modulators capable of generating higher-order formats such as QAM-m. Conventional MZM optical output suffers from modulation loss and nonlinear constellation distortion due to its nonlinear transfer characteristic and electronic driver noise. An emerging optical Digital-to-Analog Converter (oDAC) structure for generating PAM and QAM signals of any order mitigates conventional MZM limitations by nesting two or more MZMs in parallel, each driven by lower-order NRZ or PAM4 signals. To further improve oDAC performance, implementing modulators with enhanced linearity and other beneficial qualities in its parallel paths would be advantageous. Following this approach, we introduce and analyze a Ring-Assisted Mach-Zehnder Interferometer (RAMZI) optical DAC structure (RAMZI-oDAC), leveraging its ultra-linear and controllable transfer characteristics. The RAMZI-oDAC can be optimally tuned to generate signal constellations with reduced distortion while suppressing electronic driver noise. We present an analysis of the RAMZI-oDAC field transfer characteristic in lightly/strong Under/Over-Coupling (UC/OC) operating regimes. The proposed RAMZI-oDAC design demonstrates a 2 dB improvement in EVM compared to a single RAMZI at 50 Gbaud, extending also the supported baudrate to 90Gbaud.

We propose energy-efficient multi-parallel oDAC designs for Nyquist signal transmission in high-capacity data center links, using end-to-end offline learning and linear digital filters, effectively fully suppressing MZM-induced nonlinear ISI, achieving near-linear full-scale modulation swing.

Emerging services such as artificial intelligence (AI), 5G, the Internet of Things (IoT), cloud data services and teleworking are growing exponentially, pushing bandwidth needs to the limit. Space Division Multiplexing (SDM) in the spatial domain, along with Ultra-Wide Band (UWB) transmission in the spectrum domain, represent two degrees of freedom that will play a crucial role in the evolution of backbone optical networks. SDM and UWB technologies necessitate the replacement of conventional Wavelength-Selective-Switch (WSS)-based architectures with innovative optical switching elements capable of handling both higher port counts and flexible switching across various granularities. In this work, we introduce a novel Photonic Integrated Circuit (PIC)-based switching element called flex-Waveband Selective Switch (WBSS), designed to provide flexible band switching across the UWB spectrum (~21 THz). The proposed flex-WBSS supports a hierarchical three-layered Multi-Granular Optical Node (MG-ON) architecture incorporating optical switching across various granularities ranging from entire fibers and flexibly defined bands down to individual wavelengths. To evaluate its performance, we develop a custom network simulator, enabling a thorough performance analysis on the critical performance metrics of the node. Simulations are conducted over an existing network topology evaluating three traffic-oriented switching policies: Full Fiber Switching (FFS), Waveband Switching (WBS) and Wavelength Switching (WS). Simulation results reveal high Optical-to-Signal Ratio (OSNR) and low Bit Error Rate (BER) values, particularly under the FFS policy. In contrast, the integration of the WBS policy bridges the gap between existing WSS- and future FFS-based architectures and manages to mitigate capacity bottlenecks, enabling rapid scalable network upgrades in existing infrastructures. Additionally, we propose a probabilistic framework to evaluate the node’s bandwidth utilization and scaling behavior, exploring trade-offs among scalability, component numbers and complexity. The proposed framework can be easily adapted for the design of future transport optical networks. Finally, we perform a SWaP-C (Size, Weight, Power and Cost) analysis. Results show that our novel MG-ON achieves strong performance, reaching a throughput exceeding 10 Pb/s with high OSNR values ≈14–20 dB and BER ≈10−9 especially under the FFS policy. Moreover, it delivers up to 7.5× cost reduction compared to alternative architectures, significantly reducing deployment/upgrade costs while maintaining low power consumption.

This article analyzes the capacity and latency requirements imposed by various 6G envisioned scenarios and technologies over the fronthaul/ midhaul networks when considering radio access network (RAN) disaggregation. We also examine how emerging 6G technologies -- such as multi-connectivity, coordinated multi-point, and relays -- impact on these requirements. For the capacity requirements, we differentiate between basic and advanced 6G RAN scenarios, considering both single-carrier and multi-carrier setups, while extrapolating configuration parameters based on previous generation systems. For the latency requirements, our study is based on values already envisioned for 5G Advanced. Finally, we propose various transport network scenarios tailored to address such requirements in future 6G fronthaul/midhaul networks, focusing on two specific use cases: flexible function split (FFS) and cell-free multiple-input multiple-output (MIMO).

The European FLEX-SCALE project advances disruptive research on complementary optical x-haul network technologies in support of emerging requirements for 6G cellular systems. Employing a top-down strategy, it addresses the diverse requisites across all network levels. The architecture introduces optical nodes (ONs) for efficiently aggregating traffic, strategically placed to meet ultra-low latency demands at the network edge and varied capacities in the core. Utilizing optical ultra-wideband (UWB) and space division multiplexing (SDM) multiplexed transmission links, each ON routes traffic at multiple granularities (MG), from full fiber to flexibly defined bands, down to individual wavelength channels, as needed by the traffic flows. This enables the replacement of electronic routers by all-optical switches, significantly reducing energy consumption. The support of UWB/SDM boosts the network node aggregate capacity to 10 Pb/s, meeting the envisioned year-2030 capacity targets for 6G. With enabling technologies and components like the optical digital-to-analog converter (DAC), the plasmonic transceivers, or multi-granular optical nodes, FLEXSCALE seeks to make a reality key 6G network attributes like ultra-high capacity, flexibility, scalability, cost-effectiveness, low latency, and reliability. A sophisticated control and orchestration system ensures optimal operation, energy efficiency, and service provisioning, utilizing advanced traffic engineering, cloud-native architecture, and machine learning. In summary, FLEX-SCALE represents a visionary and comprehensive approach, ushering in a transformative era for 6G connectivity.

This paper presents a flexible data plane architecture designed to optimize switching and resource allocation in optical networks. By enabling multi-level switching spanning full fiber, spectral band, and wavelength levels the architecture demonstrates adaptability to varying traffic demands across different network segments or in end-to-end configurations. Its modular structure supports the integration of Ultra-Wide Band (UWB) and Spatial Division Multiplexing (SDM) technologies, enabling efficient use of spectral and spatial resources. This study explores how the proposed architecture increases network flexibility and scalability, particularly in high-capacity scenarios, and provides a scalable foundation for next-generation optical transport systems.

We present a scalable 6G X-haul optical network architecture integrating programmable photonic sub-systems, controlled by an advanced management and orchestration platform that enables intelligent, energy-efficient, and trusted coordination across disaggregated multi-domain transport network segments to support E2E-Network Slicing. (c) 2025 The Author(s)

Optical digital-to-analog converters (DACs) represent a new class of devices that offer power-efficient amplitude multiplexing in the optical domain. The combination with terahertz-bandwidth plasmonic modulators enables highest symbol rates and a low circuit complexity. (c) 2025 The Authors