Saeed Gazor

Spoofing attacks (SAs) in Global Navigation Satellite Systems (GNSS) involve transmitting counterfeit timing and location information and pose serious threats to many critical infrastructures. Target GNSS receivers are vulnerable to lowpower SAs, which may lead to unavailability, navigation errors, or entirely spurious position/time solutions. Traditional SA detectors typically assume a single SA originating from a specific direction with higher power than the authentic signals. These methods rely on eigendecomposition of the spatial covariance matrix to identify the spoofer's direction and often fail to detect lowpower spoofers or multiple simultaneous spoofers. In this paper, we introduce a spatio-temporal method to distinguish spoofers from authentic signals using a sparse array antenna in the pre-correlation stage of the GNSS receiver. We employ high-resolution direction-of-arrival (DoA) estimation followed by a power-independent cyclostationarity detector. For each direction, the received signal consists of either a single authentic satellite signal or multiple counterfeit (spoofer) signals. The proposed detector decides between these two hypotheses by estimating the number of impinging signals, and we construct an asymptotically optimal detector using a suitable separating function. In addition, we propose a spatial filter to obtain a spooferfree signal, thereby enhancing the robustness of GNSS receivers against lowpower and multiSAs. Simulation results demonstrate the effectiveness of the proposed sparsearray approach in separating and localizing multiple individual spoofers—even in multipath environments—each transmitting multiple counterfeit signals.

This paper addresses moving target detection in the presence of an Interfering Target (IT) using Passive Bistatic Radar (PBR) systems affected by hardware mismatches. We focus on In-phase/Quadrature-phase Imbalances (IQIs) common in Direct-Conversion Receivers (DCRs), where IT echoes mask the Main Target (MT). When unmitigated, both effects significantly degrade detection performance. Unlike existing approaches, we assume unknown noise power and unknown target Doppler Shifts (DSs), further increasing the detection challenge. We develop a new signal model for this scenario, formulate a Binary Composite Hypothesis Testing (BCHT) problem, and derive a closed-form asymptotically optimal Generalized Likelihood Ratio Test (GLRT) detector using the Neyman-Pearson criterion. To improve computational efficiency, we exploit Fast Fourier Transform (FFT) symmetry properties and employ batch data processing, substantially reducing computational complexity and execution time. We also analyze the impact of amplitude and phase mismatches on the PBR signal model. Theoretical analysis and simulations demonstrate improved detection probability and enhanced robustness to IQI compared with state-of-the-art methods.

The integration of sensing and communication functions is a fundamental paradigm for enabling robust and resource-efficient unmanned aerial vehicle (UAV)-assisted wireless networks. This paper addresses the optimization of integrated sensing and communication (ISAC) systems in UAV-aided wireless networks featuring wireless power transfer (WPT). We propose a novel architecture wherein multiple UAV-based radars concurrently serve multiple clusters of energy-limited communication users while performing sensing tasks. Initially, radars sense the environment, enabling users to harvest and store energy from radar transmissions. Subsequently, this stored energy facilitates uplink communication from nodes to UAVs. Our multi-objective design problem optimizes UAV trajectories, radar transmit waveforms, radar receive filters, time scheduling, and uplink powers to enhance both radar and communication system performance. Incorporating user location uncertainty, we formulate a robust non-convex optimization problem. To address this, we employ an alternating optimization approach, complemented by fractional programming, S-procedure, and majorization-minimization (MM) techniques. Numerical examples illustrate the efficacy of our method across diverse scenarios.

Wetlands are vital for maintaining ecosystems and supporting biodiversity, but they face increasing threats from climate change and human activities. Accurate mapping of wetlands is essential to detect detrimental changes and guide effective conservation efforts. However, many wetland mapping (WM) methods using convolutional neural networks rely on kernels with fixed sizes and shapes, limiting their ability to capture the multiscale features of wetlands. To enhance their ability, we propose Multiscale Deformable DenseNet (MDD) by integrating deformable convolutions into the DenseNet architecture and employing a dual feature extractor. The deformable convolutions adapt kernel shapes and sampling to capture spatial patterns across scales, while the dual feature extractor uses varied kernel sizes for diverse receptive fields. These innovations significantly improve the classification accuracy for the complex WM task, where classes are often highly similar. Experimental results demonstrate that MDD achieves the highest overall accuracy (OA) in three hyperspectral data sets, with an OA of 97.24%, 98.23%, and 94.59%, compared to the best competing OAs of 96.33%, 97.16% and 92.70%, respectively. These results highlight MDD’s superiority in WM¹.

Accurate classification of hyperspectral images is particularly challenging in wetlands regions, where land cover classes are highly similar. While integrating local and global features can enhance feature representation and improve classification accuracy, this approach remains largely unexplored in wetland mapping (WM) applications. Additionally, existing methods often suffer from information loss during feature extraction and integration processes. To address these challenges, this paper introduces ConViT, which combines convolutional neural networks and vision transformers to simultaneously leverage local and global features for WM. ConViT consists of four components: (1) local feature extraction using DenseNet, (2) a novel local embedding for tokenization that preserves spatial relationships, (3) a convolutional-transformer block (CTB), and (4) a softmax-based classifier. By integrating DenseNet with Squeeze-and-Excitation layers, ConViT enhances feature reuse and boosts sensitivity to subtle variations in wetland classes. Our contributions include a dual-input CTB with skip connections for richer feature extraction, a unique local embedding method to maintain spatial context, and efficient global-local feature fusion. ConViT outperforms two traditional and seven deep-learning classifiers for WM across four datasets. In addition, ConViT exhibits superiority in visual inspection, statistical significance tests, and focused comparison experiments, which pinpoints its potential use for WM. The source codes of this work will be available at https://github.com/halizz821/ConViT.

The performance of widely linear beamforming (WLB) is superior to adaptive beamforming, but it is limited by the uniform linear array geometry and non-uniform noise. In this paper, to overcome these limitations together, we propose a framework for widely linear beamforming in coprime array with non-uniform noise (WLB-CANUN). We subtract the non-uniform noise component from the coprime array sample covariance matrix, and vectorize the resulted matrix to create the difference co-array (DCA). Since the DCA is not uniform, we interpolate it and recover its signal by formulating the atomic norm minimization problem with the Toeplitz and orthogonal subspace constraints.The pseudo sample covariance matrix of coprime array does not contain the non-uniform noise component, which can be directly vectorized to create the sum co-array (SCA). Due to the non-uniformity of SCA, we interpolate it and recover its signal by formulating another atomic norm minimization problem with the Hankel and orthogonal subspace constraints. The directions of non-circular signals can be estimated by the traditional subspace method, which are utilized to estimate their non-circular coefficients. A least square optimization problem using the sample and pseudo sample covariance matrices of coprime array is formulated and solved to estimate the powers of non-circular signals. The interference-plus-noise covariance matrix (INCM), pseudo INCM and augmented INCM of coprime array are reconstructed, so that the ultimate augmented weight vector can be calculated. Simulation results indicate that the proposed WLB-CANUN method overcomes the limitations of WLB in coprime array with non-uniform noise, and enhances the performance compared to the existing WLB methods.

Non-stationary signals in several applications appear sparse in the time-frequency (TF) domain. In this paper, we optimize a quadratic TF distribution (QTFD) versus its kernel by exploiting this property by maximizing power per nonzero components in the TFD domain. This approach reduces the number of large cross-terms (CTs) values while preserving autoterms. This optimization is conducted by introducing an efficient approximation of ℓ ₀ and ℓ ₁ norm without any assumption about auto-terms' locations and CTs in the TF domain. The proposed TFD exhibits superior performance in instantaneous frequency (IF) estimation and fewer CTs than the state-of-the-art TFDs