Recent publications

Identifying surgical phases is a crucial component of surgical workflow analysis, facilitating the automated evaluation of surgical procedures’ performance and efficiency. A significant challenge in developing neural networks for surgical phase...

This paper addresses the challenge of real-time rendering of dynamic 3D scenes using Instant Neural Graphics Primitives (iNGP), training a separate radiance field for each frame. To enable interactive VR playback, efficient data storage and...

Spatial-diversity schemes are applied to improve signal quality of coherent free-space optical transmission systems with uncorrelated phase noise. We compare the performance of conventional schemes (MRC, SDC) to a newly proposed one (X-MRC).

Researchers from the University of L’Aquila, NICT, Finisar, Prysmian, Nokia Bell Lab and Fraunhofer HHI Space-Division Multiplexed Transmission technologies from the Lab to the Field and present a summary for their work in this invited...

Single-photon avalanche detectors (SPADs), which operate without cryogenic cooling, are a core component for practical QKD systems. We demonstrate 13-hour continuous key generation with our timebin-phase BB84 QKD system using our InGaAs/InP SPAD,...

To enable QKD in various scenarios, we developed a 625 MHz real-time autonomous timebin-phase QKD system with custom clock synchronization. We demonstrate the system’s performance and robustness under unstable link conditions, a variety of fiber...

We investigate high-symbol rate SDM transmission using 16 to 128 GBd QPSK, 16-QAM and 64-QAM modulation over a 3-mode fiber. An aggregated net rate of 2.56 Tb/s was obtained after 54 km transmission using 128 GBd 16-QAM signals.

We introduce iQSync, a clock-offset recovery method for quantum key distribution. iQSync achieves sub-linear computational complexity, allowing for the implementation on low-level hardware. We demonstrate the method for channel attenuations...

A narrow linewidth tunable laser based on ring resonators is presented. It uses low loss passive waveguides implemented in a monolithic InP photonic integrated circuit. The laser can be tuned over a wavelength range from 1559 nm to 1602 nm. The...

3D Gaussian Splatting (3DGS) is a powerful technique for real-time radiance field rendering but is limited by its high storage and memory requirements. This survey provides the first comprehensive analysis of existing compression and compaction...

This study introduces a novel approach for precise, radiation-free intraoperative measurements using 3D stereoscopic imaging integrated into a digital surgical microscope for stapedotomy. By optimizing the calibration of stereoscopic zoom-focus...

We address digital self-interference cancellation (SIC) in simultaneous transmit-receive systems using data-driven approaches. We propose kernel-based learning methods -- adaptive projected subgradient method (APSM) and support vector regression...

We present the design and fabrication of more than 2000 km of a 7-core multi-core fiber (MCF) drawn from a single large-scale MCF preform. The fiber was fabricated without any online fiber breaks and exhibits axially excellent geometrical...

The EU-funded project COMPAIR designed advanced monitoring and simulation tools that aggregate data from multiple sources to provide a well-rounded view of air pollution and related challenges e.g. traffic, waste management, energy use. Built-in...

We successfully achieved MFI using a neuromorphic PIC with a photonic reservoir and digital readout, both in a laboratory environment and with commercial network equipment.

We address system trade-offs, e.g. between resolution and speed of a coherent scanning LiDAR. We analyze the performance of our real-time FMCW LiDAR prototype with a high-quantumefficiency coherent receiver, two-dimensional...

In cell-free networks, system throughput is enhanced by allowing more than one access point (AP) to serve each user, which may lead to high energy consumption. To circumvent this, this paper presents a surrogate optimization technique using...

This research compares the local- and global modeling ways of XGBoost in the geospatial studies. Two experiments based on simulation data and German COVID-19 infection data were conducted to compare the two modeling ways from multiple aspects....

We introduce SPNerf, a NeRF based zero-shot 3D segmentation approach that leverages geometric priors. We integrate geometric primitives derived from 3D scene into NeRF training to produce primitive-wise CLIP features, avoiding the ambiguity of...

A novel 3D photonic integrated interposer is developed for coupling the different layers of Multi-core-fiber (MCF) cores with arrays of conventional 2D electro-absorption modulated lasers (EMLs) and photodetectors (PDs). The 3D photonic...

Extended Reality (XR) is evolving rapidly, offering new paradigms for humancomputer interaction. This position paper argues that integrating Large Language Models (LLMs) with XR systems represents a fundamental shift toward more intelligent,...

Recent deepfakes preserve subtle heart rate signals, challenging the assumption that rPPG-based detection remains effective. Our study demonstrates that modern deepfake generation techniques encapsulate the heart rate dynamics of source videos,...

We propose AT-GS, a novel method for reconstructing high-quality dynamic surfaces from multi-view videos through per-frame incremental optimization. To avoid local minima across frames, we introduce a unified and adaptive gradient-aware...

This paper presents an automated drywall analysis method for construction progress and quality assessment using on-site cameras. It integrates a deep learning-based segmentation model to detect and classify drywall elements, with a module to...

NeRFs struggle with low-textured areas, causing "floaters" that reduce realism, especially on featureless surfaces. We propose a method for computing dense depth priors for these areas, using a novel depth loss function and patch regularization...

We demonstrate blind 6x6 adaptive equalization of 32 GBd DP-QPSK 3-mode transmission over a 54 km FMF based on the singularity-free convergence ability of the correlation-avoidance CMA (CA-CMA). The CA-CMA is used for robust pre-convergence in a...

We report on a wide-beam LiFi transmitter using multiple VCSEL arrays with digital drivers. We transmit a 3.75 GBd PAM-3 signal at a gross data rate of 5.625 Gbit/s. Single-driver OOK operation yields 7.75 GBd.

We propose a novel solution using the Eclipse Dataspace Components (EDC) Connector to enable regulated and policy-driven telemetry data sharing in telco ecosystems. We present its architecture, several compatible use-cases, and results validating...

This paper presents a dual-polarized magneto-electric dipole antenna with orthogonal Gamma and inverted-Gamma probes, fabricated via 3D metal printing. It offers a 3-4 GHz impedance bandwidth, with excellent cross-polarization isolation (>50...

This paper compares different raytracing tools in the sub-THz area, to analyze the weight of several different effects, e.g. reflections, diffractions, scattering or atmospheric attenuation, during raytracing simulations for high frequencies. For...