AN AI-ENHANCED DPI ALGORITHM FOR SDN NETWORKS

Jumayev, Sodiq

Al-Farg'oniy avlodlari · 2026-yil

Annotatsiya

In medium-scale ICT networks, SDN centralizes control, yet TLS/HTTPS encryption, QUIC/WebRTC, and DDoS threats reduce the effectiveness of conventional DPI. This paper proposes TCQFG-DPI, a hybrid approach that offloads DPI computation to edge points, updates the model via federated learning, and enforces policies in the data plane using OpenFlow/P4. The pipeline includes flow separation and feature extraction, AI-DPI inference, local training/updates, FedAvg-based global aggregation at the controller, and rule installation for blocking, rate limiting, QoS, or rerouting. The MCLA-DPI configuration achieves 98.9% accuracy with 7–9 ms latency, improving privacy and scalability for real-time defense.

Maqola ma’lumotlari
MualliflarJumayev, Sodiq
JurnalAl-Farg'oniy avlodlari
Nashr sanasi2026-03-18
Son1
Betlar225-234
TilRus

Kalit so‘zlar

SDN; Deep Packet Inspection (DPI); Shifrlangan trafik (TLS/QUIC); edge inspection; AI-DPI; federated learning; FedAvg; OpenFlow; P4; risk scoring; policy enforcement., SDN; Deep Packet Inspection (DPI); шифрованный трафик (TLS/QUIC); edge-инспекция; AI-DPI; федеративное обучение; FedAvg; OpenFlow; P4; оценка риска (risk score); применение политик., SDN; Deep Packet Inspection (DPI); shifrlangan trafik (TLS/QUIC); edge inspeksiya; sun’iy intellektli DPI (AI-DPI); federativ o‘qitish; FedAvg; OpenFlow; P4; risk baholash (risk score); siyosatni joriy etish.

Ilmiy soha

Al-Farg'oniy avlodlari jurnalidan boshqa maqolalar

Al-Farg'oniy avlodlari — barcha maqolalar