This PDF file contains the front matter associated with SPIE Proceedings Volume 12740, including the Title Page, Copyright information, Table of Contents and Conference Committee list.

In defense and security applications cost-intensive high-performance imaging sensors are common assets. Modern approaches in the field of Computational Imaging promise the cost-efficient construction of high-performance imaging sensors with additional features, like multitask or hyperspectral imaging capabilities. For example, the compressive single pixel imaging method we studied benefits from a low-cost architecture and multimodal sensor capabilities. The compressive single pixel imaging approach enables the adaption of resolution and compression at the cost of operational speed. The measurement process of an imaging system based on this approach involves a spatial modulation of the scene with a spatial light modulator utilizing different modulation patterns as well as a subsequent reconstruction step to obtain an image of the scene. The computational speed of the image reconstruction is still a problem, especially for the class of iterative optimization algorithms. For real-time applications like threat detection a fast image data acquisition and reconstruction are fundamental though. The fast reconstruction of the images via a fast and efficient transformation, e.g. the fast Fourier Transformation, is advantageous when using binary transformation-based modulation patterns. Therefore, we studied binary Fourier patterns and compared scene-based evolutionary and other sorting methods for transformation based single pixel imaging in order to maximize the compression while controlling the image degradation regarding relevant features to enable real-time applicability.

In the last years Unmanned Vehicles for different environments (UxVs) have been recognized as relevant game changers and key technologies for a wide range of military and civilian applications. Even parallel deployment of heterogeneous autonomous assets as swarms or cooperating teams is no longer science fiction, but a realistic operational scenario. Their effectiveness can be significantly increased by optimization of number, capabilities and application strategy of assigned assets, particularly in time-critical tasks (e.g. the search for missing persons, disaster relief and modern warfare operations). Especially, the movements of the vehicles must be carefully managed. However, the use of small UxVs is often accompanied by limitations like short mission time and limited sensor coverage, which in turn can be compensated by the intelligent assignment of vehicles with increased autonomy in cooperating groups. An important basis for efficient and effective autonomous reconnaissance, especially in swarming or teaming scenarios, is movement optimization considering the capabilities of the heterogeneous vehicles equipped with different sensor systems operating in a combined mission. For this purpose, algorithms have been developed at Fraunhofer IOSB that enable appropriate planning and dynamic processing in very different situations for heterogeneous groups of cooperating vehicles. In the following, two exploration methods for reconnaissance missions are described in more detail and their possibilities are discussed and compared.

By implementing a mosaic multispectral technology, Safran designed a compact and lightweight camera able to create a snapshot multispectral image of eight spectral bands and a panchromatic one. The camera is a silicon sensor sensitive both on the Visible and on the Near-Infrared wavebands, on which a spectral filters matrix is placed. The spectral bands where carefully chosen based on a large database of hyperspectral images of camouflage targets in realistic backgrounds. The multispectral images acquired with this camera are numerically treated and projected in an adapted space to maximize the visual contrast between each point and its surrounding. Then, a metric determines if the pointed area corresponds to a spectral anomaly, and potentially a camouflaged target. Using this snapshot ability; Safran designed a real-time camouflage detection device integrated in an armored vehicle sight system with the aim of demonstrating the benefit of a multispectral sensor on conventional imagers, such as daylight or thermal cameras. With this device, we conducted several test campaigns. During these campaigns, several types of relevant targets were evaluated, such as armored personnel carriers with or without camouflage nets, infantrymen and soldiers in camouflage Ghillie suits. For each multispectral acquisition, we also acquired images with conventional cameras to be clearly able to appreciate the gain provided by the multispectral channel and evaluate the operational interest of this technology.

The Ghost GR Vision 60 V5 robot is a walking robot from Ghost Robotics, which has extensive sensor and autonomy capabilities. The autonomy functions are provided by Fraunhofer IOSB’s Algorithm Toolbox (ATB) and enable the robot to perform localization, obstacle mapping, path planning, and path control. While Ghost autonomously moves from position A to position B, the environment is captured by laser scanners. Normally, the shortest path between the two points is taken. But what happens if this path can be seen by enemy observers and the walking robot is disturbed or, in the worst case, destroyed? At Fraunhofer IOSB, a fusion of the sensed environment based on laser scanners, existing height models and a coverage map was implemented on the walking robot’s system. This enables the robot to navigate tactically: It will – whenever possible – actively and autonomously seek cover in the shelter of houses and other objects on its way to the given target point.

Quantum Ghost Imaging (QGI) is a scheme which uses the temporal correlation of entangled photon pairs (signal and idler) to identify detections belonging to actual entangled pairs and separate them from uncorrelated detections. By also exploiting the spatial correlation of the photon pair, this identification allows to extract image information of the idler photons by imaging the signal photons. This can even be done for non-degenerate photon pairs, allowing imaging at wavelengths without sophisticated cameras in this regime. Recent advances in detector technology, especially single photon avalanche diodes (SPADs), allow the design for new cameras capable of single photon detection with both high spatial and temporal resolution. These new detectors are especially useful for application in QGI, allowing to detect each photon individually and identify the correlated pairs after the measurement. It further allows to analyse the time-of-flight of photons and thus enables 3D imaging, the first realization of which will be presented here.

We present a compact room temperature single photon source based on a color center in hexagonal boron nitride for future long-distance satellite-based quantum networks. The performance of this quantum light source is sufficient to outperform state-of-the-art laser-based decoy quantum key distribution protocols. The emitter is directly coupled to a photonic integrated circuit that routes the single photons to different experiments. This includes both a verification of the single photon source via measuring the photon statistics, as well as a fundamental test extended quantum theory in microgravity. The payload is currently being integrated on a 3U CubeSat and will be launched in 2024 as part of the QUICK³ mission.

We address the efficiency and robustness of quantum key distribution (QKD) based on entangled states in conditions of noisy quantum channels. We consider schemes based on either source or detection station located in the middle or the channel, the latter configuration allowing realization of measurement-device independent QKD. In our study we focus both on discrete-variable (DV) QKD protocols utilizing direct photodetection as well as continuous-variable (CV) ones, which are based on the homodyne detection. Equivalently parametrizing the channel excess noise for both DV and CV protocols, we use robustness of an ideal CV QKD protocol as a benchmark for the entanglement-based DV QKD. We show that DV protocols typically overcome the CV ones in both source- and detector-in-the-middle setup configurations. However, this advantage can be undermined by the practical aspects on the DV protocols. Considering quantum-dot type of the DV sources of entanglement, we show the levels of pair generation and collection efficiencies needed to overcome any CV protocols. The requirements are generally compatible with the existing technologies but may require experimental efforts to achieve. Our results are essential for the practical entanglement-based QKD protocols with modern efficient sources, aimed at potential scalability in the quantum networks.

The distribution of symmetric keys over optical networks is possible through the implementation of continuous variables quantum key distribution (CV-QKD) systems. Despite the rate of key extraction in CV-QKD being maximized for Gaussian modulation (GM), its practical implementation is demanding due to the finite extinction ratio of the electro-optic modulators. The use of discrete modulation (DM) allows for a simpler practical implementation. Provided that higher-order constellations can be used, DM can approximate the performances of GM. Higher-order DM has been shown to be feasible experimentally in CV-QKD. Notwithstanding, such implementations are usually obtained in offline without applying all the post-processing steps, particularly the information reconciliation, turning the validity of the results dependent on a null frame error rate after information reconciliation. Besides that, a fixed value is typically assumed for the reconciliation efficiency value, being the modulation variance optimized under such assumptions. This disregarding the dependency of the reconciliation efficiency parameter on the method for information reconciliation and on the modulation variance. In this work we study the impact that such assumptions have on the optimization of the modulation variance in a CV-QKD system using probabilistic shaped higher-order DM. Moreover, we analyse how optimizing the modulation variance can affect the extraction key rates of symmetric keys at the output of the CV-QKD system. Besides the assumptions of null frame error rate and of the reconciliation efficiency, we study how different code rates may also affect the optimization of the modulation variance. This considering M-symbol quadrature amplitude modulation and multidimensional reconciliation.

Dark-field techniques are successfully used in microscopy for increasing the contrast of almost transparent objects, and for edge detection by removing image components with low spatial frequencies. A removal of image components with little interesting information but high intensity yields a higher signal-to-noise ratio for the image components of interest. Here, we present a technique to enhance the signal-to-noise ratio of an image signature contained in a spatial asymmetry. While being an interferometric technique based on image inversion, it will work with incoherent light sources, and thus be applicable in many practical imaging scenarios. We experimentally demonstrate an increase of the signal-to-noise ratio in asymmetry detection by an order of magnitude in a proof-of-principle experiment.

Physical models play a fundamental role in the evolution of SPAD technology. However, as the majority of the detector properties strongly depends on the electric field, a thorough knowledge of the doping profile is mandatory. Conventional doping extraction methods proved to be not accurate enough, hence the need to develop a new profiling technique which allows us to reach the desired level of accuracy. To this aim, we adopted an inverse modeling approach based on the combined use of capacitance-voltage measurements and electrical simulations. We applied the method to several SPADs, which differ both in terms of doping profile and internal structure, and we used the extracted profile to simulate their breakdown voltages. Calculated results very close to the experimental data were obtained, providing us a convincing validation of the proposed extraction technique.

Imaging LiDAR (light detection and ranging) systems for sensing the surrounding environment are a key component for, e.g., reconnaissance and autonomous navigation. The physical basis of LiDAR systems is the time-of-flight measurement of the backscattered intensity of a pulsed laser beam. For a sufficiently high detection efficiency, the highest possible sensitivity of the photodetector is required as provided by the single-photon avalanche diode (SPAD). The maximum detection range can be even further extended by the usage of laser sources for the short-wavelength infrared (SWIR) spectral range with typical wavelengths around 1550 nm, allowing for increased optical intensities to be emitted. The InGaAs/InP material system is ideally suited for SPADs for the SWIR spectral range. The fabrication of InGaAs/InP SPADs into focal plane arrays as core component for future imaging lidar systems with a high spatial resolution are quite demanding. The key technology for the fabrication of InGaAs/InP SPADs is the planar process technology via zinc diffusion to produce spatially confined p-type regions. For the zinc-diffusion process, a novel method of selective epitaxial overgrowth was developed, achieving the intended double-diffusion profile. First experimental data of thus fabricated InGaAs/InP avalanche photodiodes show the expected dark-current, photo-current, and multiplication-gain characteristics in linear-mode operating, as well as breakthrough behavior in Geiger-mode operation at typical operating temperatures around 220 to 240 K, which are achievable by thermoelectric cooling.

Here, a high-performance and low SWaP-C octave-spanning frequency comb laser system operating in the visible and near-infrared spectral regions is reported. A highly-compact diode-pumped Kerr-lens mode-locked Ti:sapphire laser producing 37 fs pulses with an average power of 820 mW and a pulse repetition rate of 0.5 GHz was used as a pump source for supercontinuum generation. Octave-spanning supercontinuum (∼ 550-1100 nm) was produced with an average power up to 300 mW from an 8-cm long nonlinear photonic crystal fiber. The broadband nature of supercontinuum, its high power, and overall compactness of the set-up make the whole system to be viable for further development of self-referenced optical frequency combs for various portable optical clock applications.

Efficient guidance of physical experiments involving many control parameters presents a challenging optimization problem. In this work, we investigate how machine learning methods can be utilized to dramatically speed up the parameter tuning process pertinent to cold-atom sources with applications in quantum memories and atom interferometry. We compare the capabilities of several machine learning strategies in controlling the experimental process and report on the superior performance of the scalable Bayesian optimization algorithm, specifically tailored for this task.

The realization of next-generation communication systems requires fast and efficient THz signal control, enabling a data stream to be encoded into a THz-frequency signal. This requires modulators with both a high modulation depth and reconfiguration speed; both are important figures of merit for communication devices. Metamaterial modulators using a brickwork antenna structure have recently shown promising performance below 1 THz. In our work, we designed, fabricated, and characterised a solid-state optical intensity modulator operating at 2.2 THz using a single-layer metasurface loaded with CVD-grown graphene. We employ an equivalent circuit model to optimize the geometric parameters for the 2 – 3 THz range, accessible by quantum cascade lasers. The incident electric field is confined by the metallic antenna arrays to the narrow resonator gaps, where patches of commercially grown graphene are defined. The graphene conductivity can be continuously tuned by applying a back-gate voltage between the p-doped silicon substrate and the metamaterial on top, separated by a 300 nm SiO2 dielectric layer. By gating the graphene, the transmitted electric field is modulated. The transmission of the device has been studied with terahertz time-domain spectroscopy. Our single-layer solid-state device achieves an intensity modulation depth greater than 68% at 2.2 THz.

Lens are used to manipulate propagation of electromagnetic waves in various applications such as communication, imaging, and sensing. The paper describes the design and characterization of an ultra-wideband reflector lens operating at 300 GHz. The lens is constructed using a single-layer metasurface and the unit cell consists of a single metal ring with two openings. The design achieves an efficiency greater than 90% and a cross-polarization gain of 25 dB over the a bandwidth of over 160 GHz.

A novel, quantum mechanism, the in-plane photoelectric effect, which enables a large terahertz (THz) photoresponse in gated two-dimensional electron gases (2DEGs), was recently discovered. To develop the mechanism further, a systematic study is carried out in this work using a dual-gated photoelectric tunable-step (PETS) terahertz detector based on a GaAs-AlGaAs heterojunction. The electric field distribution within the 2DEG channel is simulated and integrated to obtain the radiation-induced ac potential difference as a function of the 2DEG depth and the gap size to investigate the parameter dependence of the THz photoresponse. Based on the simulation results, we design, optimize, and fabricate a dipole-antenna PETS THz detector with 2DEG depth d = 90 nm and gap size b = 190 nm, which shows a strong photoresponse to 1.9 THz radiation.

A compact receiver module comprising broadband silicon integrated dual-polarized antenna-coupled heterojunction bipolar transistor (HBT) detectors is implemented as THz polarization imager. Polarization imaging through the Stokes vector formulation captures the scattering of light due to the interaction of light-matter at the interfaces thereby providing the insights on optical activity of the object which otherwise are not observed in conventional spectroscopy or classical imaging. A typical polarization imager requires detection of orthogonal linear and circular polarizations simultaneously. As compared to visible wavelengths where the abundant natural radiation can be split and maneuvered using low loss optomechanical components to record different polarization states, terahertz radiation is weak and lossy optomechanical components pose challenges in performing polarization imaging at terahertz wavelengths. The dual-polarized antenna-coupled detectors packaged as a single pixel can detect orthogonal polarizations simultaneously with isolation below -30 dB thereby minimizing the required optomechanics in polarization imaging set-up. Each detector exhibits a peak responsivity of 5.6 kV/W and minimum NEP of 3 pW/√ Hz at 450 GHz with a broad 10 dB detection bandwidth of 750 GHz (from 250 GHz to 1 THz). These highly sensitive detectors offer a cost-effective alternative to perform frequency domain spectroscopy and imaging. Here, these detectors are implemented to demonstrate polarization imaging at 852 GHz on a diffusive Teflon object. The finer local physical fingerprints of an imprinted Teflon object are computed using the Stokes vector formulation on the orthogonally polarized transmission images.

This paper proposes a three-layer electromagnetics model for investigating the interaction of millimetre wave radiation with human skin and clothing as a means to improve security screening technology. Progress in this spectral band is only possible with a deep understanding of the interaction of this radiation with the human body, the clothing and threat items; the proposed model in this paper is implemented to provide us with this information. The model consists of three layers namely, 1) air (a semi-infinite layer and lossless layer), 2) clothing layer (finite thickness layer ranging from 0.0 mm-1.2 mm), and 3) skin layer (semi-infinite layer). Three types of clothing are investigated by the model, these being: 1) Fleece, 2) Denim, and 3) Leather. Simulation results indicate that a clothing layer, in direct contact with human skin, enhances the transmission of radiation between air and skin. This is due to the clothing layer acting as an impedance matching transformer. The simulation results indicate that the power transmission coefficient varies with the frequency and thickness of the clothing layer. For the Fleece, the power transmission coefficient was found to increase fractionally, ranging from 0.05 to 0.112 at 90 GHz when the thickness of the fleece varied between 0.5mm and 1.2 mm. For the Denim, the difference in the transmission coefficient was found to increase with the thickness of the Denim, up to a factor of 0.095, when the thickness of the Denim layer varied from 0.5 mm to 1.2 mm at 60 GHz. For the beige leather, the variation in the transmission coefficient between different thicknesses was substantial and it could be up ~10.0% when the thickness of the material varied from 0.5 mm to 1.0 mm at 60 GHz. However, when the thickness of the leather was 1.2 mm, the maximum value of the transmission coefficient of 0.83 was achieved at around 45 GHz. Simulation results obtained from the three-layer model presented in this paper are in good agreement with the measurements reported in the literature reviews at specific frequencies. The importance of having such a validated model is that security screening system parameters can be optimised with a minimal amount of costly trials and experimentation.

In recent years, low-cost millimeter-wave CMOS-IC radar have become available and particularly applied to automotive because of its superior environmental tolerance compared to cameras and LiDAR. On the other hand, it is inferior to other sensors in terms of spatial resolution(recognition), making target detection accuracy an issue. This fact limits expanding application like security and others where requires high resolution. MIMO Radar, also known as virtual array, is one of the promising technologies to improve spatial resolution. However, a commercial CMOS-IC generally has only a small number of antennas, which limits the performance improvement. To solve this problem, a recent trend is to increase the number of real antennas by cascading multiple CMOS-ICs on a single substrate, thereby increasing the aperture size of the virtual array and significantly improving spatial resolution. However, when using such a method, reflected waves from targets in short range cannot be regarded as plane waves, where target DoA estimation with conventional(far-field) mode vector degrades accuracy. This paper presents an approach to improve spatial resolution and suppress performance degradation in near-field areas simultaneously by concatenating plural modules, even if a module consists of a single CMOS-IC. Computer simulations and fundamental experiments show that the proposed approach can accurately identify close proximity multiple targets in the near field even by using the far field mode vector, which is low computational burden.

This work consists of the generation of free-space Laguerre-Gaussian (LG) modes at millimeter-wave frequencies by use of a 3D-printed spiral phase plate and measuring the transmission capability of LG beams with respect to conventional Gaussian beams in scattering media such as wood, drywall, and rocky soil in the 1-3mm wavelength range. When a Gaussian beam of light passes through the spiral phase plate, the phase of the light is modified according to the thickness of the etched spiral pattern, resulting in an LG mode. Transmittance of LG beams at both 90GHz and 270GHz is measured with rocky soil, wood, and drywall. The results show that the frequency dependent transmission of the beams is governed by the mode number as well as the anisotropy of the medium. A frequency-dependent improvement in transmission is observed for highly scattering media with respect to mode number. These results support recent studies that show that OAM modes are useful for measurements through anisotropic media. Their demonstration for lower orders, as shown here in the mm-wave frequency range, is exciting for applications such as the detection of concealed objects, wireless communication, and imaging in highly scattering media.

A noninvasive, painless, and sensitive approach to precise glucose monitoring for ever-increasing diabetics worldwide is critical to reduce the long-term severe complications of diabetes. Recently, wireless sensing in the sub-terahertz and terahertz (THz) bands between 100 GHz and 10 THz opens the door to biosensing and bioimaging for innovative medical applications beyond radar and localization. Here, we report the fingertip blood-free non-invasive glucose monitoring using three-dimensional (3D) eye diagram analysis of THz wireless data to replace current blood finger-stick glucose strips and further realize remote healthcare of THz joint communication and sensing. We demonstrate the accurate estimation of blood glucose concentrations by transmitting a 10 Gbps on-off keying-modulated THz wireless data to a fingertip and measuring a 3D eye diagram of the THz wireless signal reflected from the fingertip. In particular, the method addresses personal variation and sensing accuracy through multivariate analysis of eye parameters including eye height, eye width, Q factor, and probability difference. The experimental results clearly demonstrate that the data waveform is distorted and the probability difference between the logic one and zero levels in the histogram increases in proportion to the glucose concentration. The clinical trial results for THz glucose monitoring clearly show a strong linear correlation between the eye parameters of fingertip and blood glucose concentrations and the consistency between glucose dynamics in blood and fingertip in the oral glucose tolerance test. This approach provides an innovative and effective route for developing daily diabetes management and THz joint sensing and communication.