This document summarizes the proceedings of the 17th EAI International Conference on Tools for Design, Implementation and Verification of Emerging Information Technologies (TridentCom 2022).
TL;DR
This volume collects 11 peer-reviewed papers exploring the cutting edge of network security, 5G infrastructure, blockchain efficiency, and federated machine learning. It provides a cross-disciplinary look at how researchers are tackling the design and verification challenges of next-generation, decentralized information systems.
What Problem Does This Paper Address?
As technologies like 5G, blockchain, and edge computing move from theoretical frameworks to real-world deployment, they face significant hurdles in reliability, security, and scalability. This collection addresses the practical "how-to" of these emerging fields.
The contributors tackle specific gaps such as:
- Industrial Security: How to monitor and secure PROFINET industrial operations using self-learning frameworks.
- Edge Efficiency: Optimizing machine learning (specifically knowledge distillation) for resource-constrained edge devices.
- Adversarial Threats: Detecting sophisticated web bots and defending against AI-generated (GAN-based) energy theft attacks.
- Infrastructure Management: Streamlining 5G experimental platforms and managing application migration in Multi-access Edge Computing (MEC).
Methods
The papers in this volume employ a diverse range of modern computational and engineering techniques:
- Generative Adversarial Networks (GANs): Used to model real-time covert energy theft attacks to test the robustness of data-driven detectors.
- Federated Learning: Specifically, the implementation of Federated Support Vector Machines (FSVM) to enable privacy-preserving machine learning for smart city applications.
- Knowledge Distillation: A novel "NFT-assisted" framework is proposed to compress and transfer knowledge to edge devices efficiently.
- Experimental Testbeds: Extensive use of open-source and standardized tools like OpenAirInterface for 5G testing and AdvantEDGE for MEC application migration.
- Digital Logic Design: Implementation of an all-digital Data Recovery Circuit (DRC) for 100BASE-FX communication.
Datasets
While specific benchmarking datasets vary by individual paper, the volume highlights several key data sources:
- HTTP Access Logs: Utilized for training and evaluating web bot detection models based on hidden behavioral features.
- Distributed Cluster Data: Explored in the context of importing large-scale data into cloud platforms via Java Message Service (JMS).
- 3D Style Transfer Data: Reviewed in a comprehensive survey of the field.
- Simulated Network Traffic: Many papers rely on data generated through real-time simulation environments (like AdvantEDGE) to model 5G and MEC scenarios.
Key Insights and Ideas
- Security as a Moving Target: The inclusion of GAN-based attack modeling suggests a shift toward "adversarial verification"—designing systems that are tested against AI-generated threats rather than just static rules.
- Privacy-First Smart Cities: The focus on Federated SVMs highlights a growing requirement for urban infrastructure to process data locally, protecting citizen privacy while still deriving collective insights.
- Democratizing 5G Research: By utilizing OpenAirInterface, researchers are moving away from proprietary "black box" hardware, allowing for more transparent and reproducible 5G experimentation.
- Efficiency in Decentralization: The blockchain contribution focuses on data retrieval, acknowledging that while storage is decentralized, access speed remains a critical bottleneck for mainstream adoption.
Key Results
As a proceedings volume, the results are distributed across several domains:
- Network Security: Successful detection of bots using hidden HTTP features and the development of the "POET" self-learning framework for industrial protocols.
- Communication & Services: Successful implementation of high-speed digital recovery circuits and streamlined data pipelines for cloud-cluster integration.
- Machine Learning: Validation of federated learning models that maintain high accuracy in smart city contexts without compromising data privacy.
- Infrastructure: Demonstrated feasibility of dynamic edge resource management through successful MEC application migration.
Limitations and Future Work
The editors and authors acknowledge several areas for continued research:
- Scalability: Moving federated learning models from small-scale simulations to massive, heterogeneous smart city networks.
- Real-time Performance: Improving the latency of GAN-based security detectors to ensure they can stop attacks as they happen.
- Beyond 5G: Expanding experimental platforms to include 6G precursors and more complex network slicing scenarios.
- Sustainability: Addressing the long-term energy consumption of blockchain and edge computing frameworks, which was not a primary focus of this specific volume.
Summary updated