This paper presents a deep learning approach to predict wireless channel parameters for noisy synthetic log-distance path loss models. We trained a deep neural network on data from 100 transmitters at different locations, generating radio coverage heatmaps to evaluate its ability to predict received power across receiver grids. We investigated network architectures and input features, focusing on the role of spatial correlation in the path loss exponent and the impact of additive Gaussian noise due to shadowing. Without spatial correlation, the model failed to generalize, with a model root mean square (RMSE) of ~18 dB, while correlated data enabled effective learning, achieving a model RMSE below 0.5 dB without noise and ~6 dB with significant noise (STD of 5 dB). These results on a simple synthetic model provide critical intuition on spatial correlation and noise, guiding the design of deep neural networks for sophisticated wireless channel prediction tasks, including future ray-tracingbased urban modeling.
The growing deployment of aerial vehicles across various domains, including unmanned drone delivery and urban air mobility, demands robust identification mechanisms capable of adapting to a rapidly evolving landscape, especially within congested airspace and urban environments. Existing identification systems primarily rely on visual tracking or wireless communication, each with inherent limitations: the former is constrained by line-of-sight and environmental conditions, while the latter depends on continuous connectivity and GPS data, making it vulnerable to spoofing. To address these challenges, we propose a novel radar-based identification method that introduces artificial micro-Doppler flashes generated by coded scatterers mounted on rotor blades. This approach creates unique, repeatable Doppler signatures, analogous to a Vehicle Identification Number (VIN), or Aerial VIN (AVIN), that can be interrogated by standard surveillance radars. Through a combination of theoretical modeling and experimental validation in a controlled indoor environment, we demonstrate the generation and detection of structured micro-Doppler signatures with high signal-to-noise contrast. The resulting code space reliably supports more than 264 uniquely distinguishable configurations, excluding axisymmetric redundancies, thereby confirming the feasibility of micro-Doppler-based identification for aerial platforms. Extrapolation to outdoor scenarios indicates a feasible detection range of up to 30 km under typical X-band radar conditions. These results suggest a promising path toward secure, passive identification of aerial vehicles, eliminating the need for onboard power or communication modules, and providing a reliable alternative in case of system failure.
In this work, an asymmetric double quantum well (ADQW) structure is investigated to evaluate the feasibility of its unique operating principle, based on a precise design of the third energy level. Unlike many ultrafast photodetectors that rely on thermally-driven mechanisms−which are inherently sensitive to ambient temperature variations - the proposed ADQW demonstrates intrinsic thermal stability. Although the device exhibits pronounced temperature-dependent effects, the underlying mechanism remains robust and functional over a wide temperature range from 20 K to 293 K. The operation of the structure relies on purely quantum mechanical inter-sub-band transitions (ISBT) and a built-in self-induced electric field (SIEF) generated via doping modulation (DM), enabling a photovoltaic effect with an ultrafast response. Consequently, by overcoming the limitations of thermal instability, this nanoscale structure represents a highly adaptable and promising electro-optic component for effective integration with broader electronic and optoelectronic systems.
Remote monitoring of drones is important for airspace management and security, including assessment of unauthorized payloads. Prior radar studies have established that payload state can affect drone signatures, but wind can alter rotor speeds and confound a payload estimate. Here, we examine a hovering DJI Mini 2 under controlled anechoic-chamber and wind-tunnel conditions. A balanced, top-mounted payload shifts the dominant blade-related micro-Doppler peak, whereas axial wind produces two spectral branches. We interpret the splitting as consistent with controller-mediated body tilt and thrust redistribution; rotor speed and attitude were not independently measured. At a 7 GHz carrier, increasing payload from 0 to 120 g shifts the dominant peak from approximately 340 to 430 Hz, giving a descriptive sensitivity of about 0.75 Hz/g for this platform. In the wind tunnel, the branch separation follows Δf = (16 Hz/(m/s))v, is the axial wind speed in m/s, and reaches approximately 52.8 Hz at 3.3 m/s while remaining nearly independent of the tested payload. The separation can therefore provide a model-specific wind estimate before payload correction. The resulting relations are presented as a platform-specific empirical calibration under controlled hovering. The specific contribution is the observation and empirical modeling of wind-induced branch splitting, which complements prior payload-classification methods.
This paper presents a method for passive radar detection and localization in a multistatic configuration. Conventional Cross-Ambiguity Function (CAF) processing faces two key limitations: the challenge of noncoherently summing signals across different geometries for detecting low Signal-to-Noise Ratio (SNR) targets, and the challenge of associating detections in multi-target scenarios. Direct Position Determination (DPD) methods, which operate in the position-velocity (PV) domain, resolve these issues by enabling direct noncoherent integration and bypassing the target association problem. However, the performance of DPD is limited by the high computational cost of an exhaustive multi-dimensional search. To mitigate this, we propose extending the support of the correlation peak, enabling an efficient PV sampling scheme based on a novel voxel representation. Our key contribution is the design of this voxel representation, which maps a single point in the PV search grid to a deliberately extended region in the Delay-Doppler (DD) domain. This approach significantly reduces the number of required states, thereby lowering computational complexity while ensuring full coverage of the search space.
