Abstract
The technology described in this paper relates to AI-powered anomaly detection for identifying and mitigating content delivery network (CDN) leeching activities in real-time, with particular focus on live stream protection. A multi-stage detection approach, leveraging regression followed by anomaly detection, addresses unauthorized content access and redistribution. This is achieved by ingesting real-time data from various sources—including but not limited to CDN logs, digital rights management (DRM) system telemetry, and application-level interaction data—which is then transformed into analytical features to establish baseline behavioral patterns before applying anomaly detection. Utilizing machine learning models, trained to understand user, device, and session-level behaviors, the approach performs real-time analysis of traffic and user token usage patterns, as well as internet protocol (IP) address behaviors, to detect anomalies. The anomaly detection generates severity-based alerts and supports automated responses including token revocation and IP address blocking through edge server deployment
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
N/A, "Combating CDN Leeching with AI-Powered Anomaly Detection", Technical Disclosure Commons, (May 29, 2025)
https://www.tdcommons.org/dpubs_series/8168