A Graph Neural Network Approach for Joint Security and Performance Management

The dilemma in today's cloud data centers is how to maximize resources and optimize network performance in the face of this rapidly expanding attack surface of denial of service, lateral movement and anomalous resource exploitation.Typical optimization techniques, such as heuristic schedulers and queue-based load balancers, or individual machine learning predictors, are generally addressing these objectives individually, and are not delivering optimal solutions or timely threat response.To cope with these challenges, the paper proposes a novel intelligent framework SecOpt-GNN that incorporates the cloud network performance optimization and anomaly detection for security.The performance optimization head makes task placement and routing decisions and the security head is real-time anomaly and intrusion scoring.A Graph Attention Network (GAT) based encoder is learned to provide two collaborating heads with joint representations: one head generates task placements and routing decisions, the other head scores real-time anomalies and intrusions.The framework is trained and tested with a real-world Google cluster trace data.Experimental results show that SecOpt-GNN obtains 96.4% accuracy for anomaly detection, averaged as the mean of the per-load percentage reductions across the six load levels, SecOpt-GNN reduces scheduling latency by 37.8% relative to round-robin (equivalently, a 45.1% reduction in mean latency, from 42.2 ms to 23.2 ms; per-load reductions range 16.7%-52.2%),and better results than the standard GCN, GraphSAGE, LSTM, ContrastNet, and Random Forest baselines on all scoring metrics.The results indicate that the current methods that optimize for the security or performance objectives of each network separately are sub-optimal when considering a holistic view of security and performance objectives via joint graph-based modelling.

Authors

Institutions

Publication Details

Journal
International journal of intelligent engineering and systems
Published
2026-11-17
DOI
https://doi.org/10.22266/ijies2026.1231.50
Primary Topic
Network Security and Intrusion Detection
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A Graph Neural Network Approach for Joint Security and Performance Management

Yousra Abdul Alsahib S. Aldeen, Abeer E. Abed
International journal of intelligent engineering and systems
Network Security and Intrusion Detection
article

A Graph Neural Network Approach for Joint Security and Performance Management

Yousra Abdul Alsahib S. Aldeen, Abeer E. Abed
article en

Abstract

The dilemma in today's cloud data centers is how to maximize resources and optimize network performance in the face of this rapidly expanding attack surface of denial of service, lateral movement and anomalous resource exploitation.Typical optimization techniques, such as heuristic schedulers and queue-based load balancers, or individual machine learning predictors, are generally addressing these objectives individually, and are not delivering optimal solutions or timely threat response.To cope with these challenges, the paper proposes a novel intelligent framework SecOpt-GNN that incorporates the cloud network performance optimization and anomaly detection for security.The performance optimization head makes task placement and routing decisions and the security head is real-time anomaly and intrusion scoring.A Graph Attention Network (GAT) based encoder is learned to provide two collaborating heads with joint representations: one head generates task placements and routing decisions, the other head scores real-time anomalies and intrusions.The framework is trained and tested with a real-world Google cluster trace data.Experimental results show that SecOpt-GNN obtains 96.4% accuracy for anomaly detection, averaged as the mean of the per-load percentage reductions across the six load levels, SecOpt-GNN reduces scheduling latency by 37.8% relative to round-robin (equivalently, a 45.1% reduction in mean latency, from 42.2 ms to 23.2 ms; per-load reductions range 16.7%-52.2%),and better results than the standard GCN, GraphSAGE, LSTM, ContrastNet, and Random Forest baselines on all scoring metrics.The results indicate that the current methods that optimize for the security or performance objectives of each network separately are sub-optimal when considering a holistic view of security and performance objectives via joint graph-based modelling.

International journal of intelligent engineering and systemsVol. 19(12)
University of Baghdad (IQ)
Network Security and Intrusion Detection
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon