Research

My research centers on mobile security and privacy within wireless and cellular network environments. I investigate how mobile human behavior data can be harnessed to enhance security and network resilience, while proactively addressing emerging privacy threats—particularly in the context of growing AI-powered infrastructures. My approach draws on techniques from machine learning, statistics, graph theory, and time series modeling, and is structured around the following three axes:

1. Data-driven Mobile Networks Security: Leveraging behavioral data from mobile networks to detect fraud (e.g., SIMBox), identify anomalies, model normal usage patterns, and optimize network configurations for increased robustness and adaptability.

2. Privacy-preserving Data Publishing: Developing methods to safely release mobile datasets such as Call Detail Records (CDRs) by modeling user heterogeneity, synthesizing realistic and usable data traces, and minimizing the risk of re-identification or misuse.

3. Mobile Users Privacy Protection: Exploring privacy vulnerabilities arising from behavioral data and permission-less sensors, and designing novel, effective countermeasures to safeguard mobile users against adversarial tracking and information leakage.

Below are some projects I have been working on within this framework along with related code and datasets:

projects

Adaptive Detection of SIMBox Fraud in Mobile Networks

MLNS2-funded research tackling international bypass fraud through behavioral modeling and adaptive machine learning techniques.

Privacy-Preserving Generation of Realistic Mobile Network Datasets

STIC-AmSud LINT project exploring AI-based modeling and synthesis of complete mobile datasets with privacy guarantees.

Machine Learning for Offensive Computer Security (MALFOY)

ERC-funded research exploring how ML can be misused by attackers to reveal vulnerabilities and build automated threats.

code & datasets

En-WDM Mobility Model

Enhanced Working Day Mobility simulator for generating mobility-rich mobile datasets.

FraudZen Simulator

Simulator for traffic and mobility of legitimate and SIMBox users in cellular networks.

SigN: Latency-Based SIMBox Detection

Dataset and scripts for SIMBox latency anomaly detection at the cellular edge.

MSMDF Dataset

Motion sensor fingerprinting dataset collected across 42 smartphones under diverse conditions.