Enabling Real-Time, Explainable DDoS Mitigation via On-Premise Large Language Models and Flow Analysis
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Distributed Denial of Service (DDoS) attacks continue to escalate in both frequency and sophistication, often overwhelming critical network infrastructures. While deep learning methods excel at recognizing malicious patterns, their lack of transparency undermines trust and hampers effective mitigation. This paper introduces a unified, on-premise pipeline that integrates an advanced flow based attack classifier with a local large language model (LLM) to deliver explainable, real-time DDoS defense. The proposed approach detects threats at the flow level, rapidly fags suspicious traffic, and then generates human-readable analyses and device specific countermeasures ranging from firewall rules to intrusion prevention system signatures all without transmitting data of-site. Through comprehensive testing on diverse, large scale network traces, we demonstrate that this framework not only achieves near-perfect detection accuracy but also considerably reduces operational costs and privacy risks associated with external cloud services. Furthermore, evaluators confirm the clarity and correctness of the automatically generated mitigation strategies, highlighting the system’s practicality in enterprise environments. Overall, our results validate on-premise, LLM-enhanced DDoS defense as a robust, transparent, and economical solution for safeguarding modern network ecosystems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.005 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it