A fully functional SDV emulation platform combining Mininet, the Ryu SDN controller (OpenFlow 1.3), and MQTT to enable centrally controlled, low-latency vehicle-to-infrastructure communication with real-time telemetry and monitoring. The platform features a live web dashboard, a REST API for network visibility, dynamic remote reconfiguration of vehicle behavior, and a physics-based energy consumption model, making it readily extensible for research into vehicle telemetry protocols, infrastructure-based fleet monitoring, QoS prioritization, and as a foundation for vehicle-to-vehicle coordination and smart-city integration.
AI-powered autonomous drone swarms for the security and protection of critical infrastructure. The system combines fire segmentation and early-warning capabilities with leader–follower swarm coordination, enabling drones to operate collaboratively, monitor high-risk areas, and support faster response to potential threats.
We developed a simulation of a simplified BitTorrent-like network to study peer-to-peer file distribution. A central tracker coordinates exchanges by recording which peers possess specific pieces and which still need them. The file is divided into multiple parts, which peers share with their nearest neighbors. Once a leecher acquires its first piece, it also acts as a seeder, accelerating distribution. To capture network dynamics, peer mobility is modeled so that neighborhood connections change over time.
A privacy-preserving edge-cloud federated learning system for multi-label road surface classification. Each client trains a model locally using its own road image dataset and shares only the model weights with a central server. The server aggregates these weights to update the global model and redistributes it to the clients, enabling collaborative learning without sharing or centrally storing the original data.
We have developed and we are constantly enhancing serious games for the prevention and management of obesity among young populations. Designed to improve children’s nutrition knowledge, dietary behavior, and overall food skills by incorporating goal setting, social support, and positive reinforcement.
RL-based anomaly detection for video and image frames captured in real-time to identify faults in camera-based perception systems.
A real-time video processing system, designed for object detection and tracking in both recorded video files and live streaming feeds using the Yolo deep learning models.
© 2026 Intelligent Computer Systems and Applications (ICSA). All rights reserved.
