בית הספר
להנדסת חשמל ומחשבים

Name: Moshe Buchris Degree: MS.c in Electrical Engineering Supervisor: Prof. Haim Permuter Research topic: "Convolutional Unfolding Weighted Minimum Mean Square Error for MU-MIMO Downlink Beamforming" Summary: CUNNet, convolutional deep unfolding framework for the WMMSE beamforming algorithm in multi-user MIMO downlink systems. By learning structured precoder updates that exploit the spatial properties of the channel. CUNNet improves performance under fixed computational budgets while preserving the interpretability and stability of classical WMMSE. Name: Amit Brilant Department: Electrical and Computer Engineering Advisors: Prof. Ofer Hadar, Dr. Oshrit Hopper (External-Afeka) Seminar title: Detection and Diagnosis of Ovarian Carcinoma from Complex Cysts Using Computer Vision and Ultrasound Imaging Seminar summary: This research develops an AI system to assist physicians in detecting ovarian cancer from ultrasound images. The system analyzes transvaginal ultrasound scans from postmenopausal women and classifies ovarian findings as healthy, benign cysts, or malignant cysts. By combining image analysis with clinical patient data, the system aims to improve diagnostic accuracy, reduce physician subjectivity, and enable earlier detection of ovarian malignancies. Student: Ido Yakir, M.Sc in electrical engineering, electromagnetics and waves Supervisor: Prof. Reuven Shavit Seminar title: Neural-Network-Assisted Design of Conformal Metasurface Airborne Radomes Seminar abstract: Protective radomes often degrade antenna radiation performance through insertion loss and boresight errors, particularly in conformal shapes required for aerodynamic aircrafts. Conventional design methods using commercial EM software are computationally expensive for these electrically large structures and struggle with performance variability based on antenna orientation. This study proposes a conformal radome design utilizing a hybrid structure of dielectric and metasurface layers. To enable rapid optimization, the system is modeled using an equivalent transmission line circuit with shunt admittance and cascaded ABCD matrices, significantly reducing computational overhead compared to full-wave solvers. Furthermore, a neural network is employed to address performance variability. Trained on a dataset of unit cell parameters and transmission coefficients, the network predicts optimal geometries for each point on the radome. This approach allows for local compensation of aperture field distortions and minimization of insertion loss across multiple antenna orientations.

Name: Ariel Harush Igal Bilik, Gilad Katz Master-Agent “proto-plan” System (MAPS) Distributed Learning-based Approach for Joint Navigation in Local Environments :Abstract Coordinating autonomous vehicles at unsignalized inter- sections remains challenging for multi-agent reinforcement learning (MARL) systems, which often require large action spaces, privileged information, or homogeneous agent de- signs that limit real-world deployment. We propose a hier- archical deep reinforcement learning (DRL) architecture in which a central Master agent generates a compact embed- ding, denoted as a “proto-plan,” that encodes a high-level coordination strategy. Individual Worker agents then com- bine this “proto-plan” with their local observations to deter- mine vehicle-specific actions, effectively decoupling strate- gic coordination from low-level control. This design enables heterogeneous agent types, supports flexible action spaces without combinatorial explosion, and allows independent re- training of coordination and control modules. Experiments across 72 diverse intersection configurations in the High- wayEnv simulator demonstrate that the proposed approach achieves collision-free navigation while maintaining high traffic throughput, outperforming state-of-the-art baselines. Our results show that “proto-plan”-based hierarchical learn- ing provides a scalable and adaptable framework for multi- vehicle coordination in complex traffic scenarios. Full Name: Nadav Amitai. Degree: M.Sc. in Electrical and Computer Engineering. Supervisors: Koby Todros and Igal Bilik. Seminar Title: Measure-Transformed Recursive Least Squares Abstract: This research deals with robust system identification in the presence of impulsive noise. To this end, we present a new robust variant of the recursive-least-squares (RLS) estimator, called measure-transformed (MT)-RLS. The MT-RLS is an exact recursive counterpart to the recently developed MT least-squares estimator (MT-LSE), which operates by applying a transform to the probability measure of the data. This stands in contrast to other robust RLS variants that only approximate robust batch estimators. The considered measure transform is generated by a non-negative data-weighting function, called the MT-function. We have previously shown that a properly chosen MT-function can significantly mitigate the influence of outliers arising from impulsive noise, thereby substantially enhancing the estimation accuracy of the MT-LSE. Consequently, as an exact recursive extension, the MT-RLS fully inherits the robustness of MT-LSE. The MT-RLS is illustrated in simulations, underscoring its advantage over RLS and other robust extensions in both stationary and non-stationary environments. Full name: Kobi Kenzi Name of supervisor: Dr. Dan Vilenchik Degree program: M.Sc. in Communication Systems Engineering Seminar topic: From Messy ICU Signals to Causal Insight: Learning Meaningful Patient States from Irregular Clinical Time Series Seminar abstract: This seminar presents a research pipeline for applying causal inference methods to irregular clinical time-series data, focusing on ICU datasets such as PhysioNet 2012 and MIMIC-III. The work combines time-series preprocessing, clinically meaningful latent variable construction, causal graph design, confounder selection, and treatment-effect estimation. The main goal is to explore how latent clinical states and domain-informed causal assumptions can support more interpretable analysis of patient trajectories and potential clinical effects, while avoiding strong unsupported medical causal claims. The seminar emphasizes the methodological framework, implementation challenges, validation strategies, and the role of causal inference in extracting meaningful insights from complex medical time-series data

Student full name - Guy Moshe Atias Seminar Object - Asynchronous Swarm-Intelligence Navigation for GPS-Denied IMU-Only Drone Swarms. Research supervisor - Professor Igal Bilik Abstract - This work presents a low-cost, fully asynchronous GPS-denied navigation framework for drone swarms, leveraging IMU measurements, lightweight peer-to-peer communication, swarm intelligence, and error-correction to enable accurate cooperative positioning for rapid reconnaissance and disaster- area scanning Name : Daniel Sheftel Degree : M.Sc. in Computer and Electrical Engineering Supervisor : Prof. Mor Peretz Seminar topic : Design and control of high frequency capacitive power transfer systems: Capacitive Power Transfer (CPT) systems require defined design and control mechanisms to maintain operational stability and enable spatial freedom between the coupling interface. This seminar examines the double-sided LCLC compensation topology, focusing on the mathematical modeling required to quantify steady-state behavior and dynamic responses. The presentation details the derivation of small-signal transfer models, establishing the necessity for closed-loop regulation to mitigate the effects of variable coupling parameters and hardware detuning. Furthermore, a specific control methodology is introduced, demonstrating the use of a continuous-conduction buck converter to dynamically modulate the equivalent output resistance of the LCLC network. This approach provides an active mechanism to regulate power transfer capabilities and maintain system authority across varying load conditions and spatial misalignments. Fatema Abu Elheija Master’s Student, Department of Electrical and Computer Engineering Supervisor :Gil Shalev. Seminar name : Chemical Field-Effect Transistor for the Specific and Label-free Detection of Sodium Hypochlorite Solution Abstract Rapid detection of toxic chemical species is important for environmental monitoring and chemical safety. In this work, we demonstrate label-free, real-time, and quantitative detection of sodium hypochlorite solution (SHC) using a Meta-Nano-Channel Chemical Field-Effect Transistor (MNC ChemFET) in PBS solution. The SiO₂ sensing surface is functionalized with an indole-based recognition layer that selectively reacts with hypochlorite molecules, producing changes in interfacial surface charge that modulate the transistor’s electrical response. The multi-channel MNC structure enhances electrostatic coupling and improves sensitivity and signal stability under electrolyte conditions. Calibration and control measurements confirm quantitative sensing performance, selectivity, and operational stability. Name : Dor Krief Degree : M.Sc. in Computer and Electrical Engineering Supervisor : Prof. Mor Peretz Seminar topic : Generic hardware platform for loads energy management using Smart Algorithms This research presents a unique and robust hardware platform designed for centralized energy management in multi-voltage systems, such as modern "smart soldier" platforms. The study addresses the inefficiencies of legacy power distribution-characterized by redundant energy storage, excessive weight, and double-conversion losses-by implementing a novel centralized architecture , that allows both battery charging and load supply at the same time . The platform utilizes smart algorithms to enable real-time, continuous power supply, predictive maintenance, and efficient energy shuttling between numerous loads. By integrating a distributed conversion approach and advanced data logging, the research aims to significantly extend battery lifetime, reduce system volume, and optimize power density for mission-critical applications. Student: Maor Kakun Program: M.Sc. Electrical Engineering (Signal Processing Specialization) Advisor: Professor Yitzhak Yitzhaky Research Subject: Assessing Emotional States from Facial Videos Using Remote Photoplethysmography. Research Overview: This research investigates emotion recognition using Remote Photoplethysmography (rPPG) signals extracted from facial videos. rPPG estimates the blood volume pulse by tracking subtle temporal variations in facial pixel intensities, enabling non-contact physiological measurement. Extracting high-quality rPPG signals under motion artefacts, video compression, and challenging illumination conditions remains difficult. Current state-of-the-art approaches rely on large neural network models that operate as black boxes, directly mapping video frames to rPPG signals with high computational cost and limited interpretability. This study proposes a hybrid framework that combines classical signal processing techniques, physically interpretable feature extraction, and a lightweight deep learning architecture. The proposed approach reduces computational complexity while improving model explainability and robustness across varying recording conditions. Using the resulting high-quality rPPG signals, reliable heart rate variability features are obtained, enabling effective emotion rec

Name: Tali Dror Degree: MSc in Electrical Engineering Supervisor: Prof. Haim Permuter Title: Reconstructing High-Resolution Faces from Speech Using Speaker-Aligned Diffusion Models Humans can infer aspects of a person’s facial identity from their voice alone, suggesting that speech and appearance share underlying anatomical and physiological cues. Learning this relationship computationally, however, is challenging due to the ambiguity of voice-to-face mapping, limited paired data, and the difficulty of preserving identity in realistic face generation. In this work we present a speech-to-face synthesis framework that addresses these challenges by combining identity-aware speaker representation learning with a pretrained diffusion-based face generator. By aligning speech and face embeddings in a shared identity space and carefully conditioning a diffusion model under limited supervision, our approach achieves significantly improved identity fidelity. Yuval Hadad Electrical Engineer Department Igal Bilik Topic: Adversarial Attack on Automotive Radar Point Cloud Classifiers Summary: Automotive radar systems have become critical components in autonomous driving, providing robust object detection capabilities across diverse weather conditions and lighting scenarios. This thesis investigates the vulnerability of deep learning based radar point cloud classifiers to adversarial attacks, addressing a critical gap in automotive safety research. The research is presented in two main contributions. First, we introduce the Radar PointNet architecture, specifically designed for automotive radar point cloud classification. R PointNet extends the conventional PointNet architecture by incorporating radar specific features including radial velocity, radar cross section values, and range information, enabling more effective classification of vehicles, motorcycles, and pedestrians. Second, we propose a comprehensive adversarial attack framework targeting radar based perception systems. Our initial approach introduces the Radar Car lini and Wagner attack, demonstrating that even small perturbations to radar point clouds can significantly degrade classification performance. Building on this foundation, we develop a novel group wise adversarial attack methodology that generates physically plausible perturbations by applying structured affine transformations to spatially coherent point groups. The group wise attack framework reduces optimization complexity from 256 to 100 learnable parameters for typical point clouds while paradoxically improving both attack effectiveness and imperceptibility. We introduce two grouping strategies: frame based grouping when temporal metadata is available, and PCA based grouping that approximates temporal structure using only spatial coordinates. Both methods achieve approximately 19% attack success rates while maintaining superior imperceptibility compared to unstructured baselines. Experimental evaluation on the RadarScenes dataset demonstrates R-PointNet achieves 94% classification accuracy compared to 60% for conventional PointNet. The adversarial attacks reveal critical vulnerabilities, with motorcycle targets showing highest susceptibility (54.5% attack success rate) and even well i ii classified vehicle categories experiencing 14.7% vulnerability. These findings underscore the urgent need for robust defense mechanisms in automotive radar perception systems before widespread deployment in safety critical autonomous driving applications. The research contributes both to advancing radar based deep learning architectures and to understanding their security vulnerabilities, providing essential groundwork for developing more resilient automotive perception systems Speaker: Shani Budilovsky M.Sc. student in Communication Systems Engineering, Electrical and Computer Engineering School, Ben-Gurion University of the Negev. Titel: Data Augmentation for Spoofing-robust Automatic Speaker Verification Supervisors: Dr. Yehuda Ben-Shimol and Prof. Itshak Lapidot Abstract: Automatic Speaker Verification (ASV) is a biometric authentication technology that verifies a speaker’s identity based on their voice characteristics and is widely used in security-sensitive applications such as access control and identity verification. Despite their success, ASV systems are vulnerable to spoofing attacks, including replay, synthetic, and manipulated speech. To address this challenge, Spoofing-robust ASV (SASV) systems combine speaker verification models with dedicated spoofing countermeasures to jointly improve verification accuracy and attack resistance. Data augmentation has become a key technique for improving the generalization and robustness of deep learning-based ASV and countermeasure models by increasing training data diversity and simulating real-world acoustic variability. Augmentation methods can be applied at the waveform level or at the time-frequency representation level, such as spectrogram-based techniques. However, in the SASV setting, not all augmentation strategies are equally beneficial, and some may improve speaker discrimination while degrading spoof detection performance, or vice versa. This seminar examines the effect of different data augmentation strategies on both ASV and spoofing-robust ASV systems, and discusses the importance of carefully designing augmentation pipelines that balance speaker verification accuracy with spoofing robustness in practical SASV systems. Student name: Shalev Haimovich Advisor : Matan Gal-Katziri M.Sc. in Electrical Engineering Seminar subject: A CMOS-Ready 1064 nm TFLN PIC for 3D Heterogeneous Integration. Abstract: Photonic integrated circuits (PICs) offer a promising path toward high-bandwidth and energy-efficient systems, yet active photonic devices are typically controlled, driven, and interfaced with electronic circuits. This seminar presents the design of a CMOS-ready Silicon Nitride-loaded Thin-Film Lithium Niobate PIC for electro-optic phase modulation at 1064 nm, developed toward a future 3D heterogeneous multi-chip module. Instead of treating the modulator as a standalone optical device, the design considers the PIC as both the photonic layer and part of the RF interface of the integrated system. The seminar will discuss how the optical platform, traveling-wave electrodes, RF input path, impedance and termination strategy, electromagnetic modeling, and CMOS output-stage constraints are brought together to support a practical CMOS–PIC integration scheme.