Quantum entanglement-assisted adaptive firewall intelligence for next-generation cybersecurity

Signature- and rule-based firewalls score packets one at a time against a fixed threshold, which leaves them weak against multi-stage attacks and drifting traffic. We present the final design and a complete re-evaluation of QEAFIS, a quantum-inspired architecture that places three modules around a simulated 7-qubit quantum neural network (QNN). Entanglement-Guided Inter-Packet Correlation (EGIPC) scores a session with the entropy of a dephased register qubit. Behavioral Fingerprint Memory (BFM) matches traffic against confirmed threats. Dynamic Threshold Adaptation (DTA) moves the decision threshold by stochastic-approximation descent on an explicit cost. Every result was regenerated on NSL-KDD and UNSW-NB15 under de-duplicated, leakage-free partitions with up to ten seeds. The evaluation covers the end-to-end pipeline, a ten-configuration ablation, classical and neural sequence baselines, withheld attack categories, poisoning and drift tests, and per-stage latency. BFM with an admission guard and DTA on rank-calibrated scores gave practically meaningful gains: DTA lowered the post-shift error cost by 0.104 and 0.047, and on NSL-KDD the guard held the BFM false-positive rate at 1.0% with 20% of memory poisoned, against 68.7% without it. The quantum-inspired parts did not. We show that the EGIPC score is a classical sequential statistic. It matched a transition-weighted CUSUM, trailed LSTM, GRU, Transformer and TCN models by 0.23–0.28 AUROC on evasive NSL-KDD attacks, and saturated under distribution shift. The QNN detected no better than a classical network at about thirty times its per-event cost (about 1 ms), and withheld categories were detected unreliably. Earlier shadow-mode results are retained only as provisional.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-17
DOI
https://doi.org/10.1038/s41598-026-78276-w
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

Quantum entanglement-assisted adaptive firewall intelligence for next-generation cybersecurity

Abdullah Albalawi, P. Senthilkumar, V. Asha
Scientific Reports
Network Security and Intrusion Detection
article

Quantum entanglement-assisted adaptive firewall intelligence for next-generation cybersecurity

Abdullah Albalawi, P. Senthilkumar, V. Asha
article en

Abstract

Signature- and rule-based firewalls score packets one at a time against a fixed threshold, which leaves them weak against multi-stage attacks and drifting traffic. We present the final design and a complete re-evaluation of QEAFIS, a quantum-inspired architecture that places three modules around a simulated 7-qubit quantum neural network (QNN). Entanglement-Guided Inter-Packet Correlation (EGIPC) scores a session with the entropy of a dephased register qubit. Behavioral Fingerprint Memory (BFM) matches traffic against confirmed threats. Dynamic Threshold Adaptation (DTA) moves the decision threshold by stochastic-approximation descent on an explicit cost. Every result was regenerated on NSL-KDD and UNSW-NB15 under de-duplicated, leakage-free partitions with up to ten seeds. The evaluation covers the end-to-end pipeline, a ten-configuration ablation, classical and neural sequence baselines, withheld attack categories, poisoning and drift tests, and per-stage latency. BFM with an admission guard and DTA on rank-calibrated scores gave practically meaningful gains: DTA lowered the post-shift error cost by 0.104 and 0.047, and on NSL-KDD the guard held the BFM false-positive rate at 1.0% with 20% of memory poisoned, against 68.7% without it. The quantum-inspired parts did not. We show that the EGIPC score is a classical sequential statistic. It matched a transition-weighted CUSUM, trailed LSTM, GRU, Transformer and TCN models by 0.23–0.28 AUROC on evasive NSL-KDD attacks, and saturated under distribution shift. The QNN detected no better than a classical network at about thirty times its per-event cost (about 1 ms), and withheld categories were detected unreliably. Earlier shadow-mode results are retained only as provisional.

Scientific Reports
SRM Institute of Science and Technology (IN), Shaqra University (SA)
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