Mavenir: Evolution of Real-Time Machine Learning (RRML) in Core Network Security and Anti-Fraud Into 5G and New Services
1 like1,116 views
The document discusses the evolution of real-time machine learning in core network security and anti-fraud, particularly within the context of 5G technology. It highlights the various security challenges and fraud risks facing telecommunications, as well as the integration of machine learning to enhance security measures and fraud prevention. Additionally, it outlines Mavenir's security solutions and architecture designed to address these issues while ensuring compliance with regulatory standards.
Mavenir: Evolution of Real-Time Machine Learning (RRML) in Core Network Security and Anti-Fraud Into 5G and New Services
- 1. Copyright Mavenir 2019. Proprietary and Confidential. mavenir.com Evolution of Real-Time Machine Learning (RRML) in Core Network Security and Anti-Fraud Into 5G and New Services
- 2. Context Copyright Mavenir 2019. mavenir.com Supply Chain Privacy and Data protection Signaling Cloud Internet of Things Human Device Security Threat Landscape Global Fraud Survey Global loss – $29.2B Identity fraud PBX hacking IRSF By-pass Arbitrage Wholesale Challenge Expertise Technology (5G) Organization
- 3. 5G PPP Security requirements Higher security and privacy level Security Automation Security Monitoring Security Management Liability schemes Slice isolation End to end encryption Regulation Copyright Mavenir 2019. 3 Any security trade- off would be misused by an attacker in form of Fraud or damage
- 4. 5GC Business Channels and Workstreams • Asset Tracking • Security & GDPR • Robotics • Industrial Automation • AR/VR • High Density • High Demand • Elastic Scale • Healthcare Automation • Patient Record Mgt. • Security & HIPAA • Device Diagnostics • NLP & NLU • Industrial Automation • Process Optimization • Supply Chain • Security & Safety • Video Monitoring • AR/VR • Retail Promotion • Heat Map • Facial Recognition • Security & Safety • Asset Mgt. • Video Surveillance • Private LTE/5G • Security & GDPR Mall Industrial AutomationStadium Warehouse Hospital Office Park Remote Developer 4
- 5. 5G security Architecture • 5G reference architecture offers comprehensive security but still could be exploited – Implementation trade-offs – Multi-vendor environments – IT technology • Typical Fraud scenarios are not (fully) covered Copyright Mavenir 2019. 5 User Application Provider Application SN HE 3GPP AN Non-3GPP AN (I) (I) (I) (II) (IV) (V) Application Stratum Home Stratum/ Serving Stratum Transport Stratum (II) (I) (III) (I) ME USIM (I) SEPP AUSF ARPF SEAF
- 6. Why Telco needs help ? MNO • Security is a cost factor contradicting OPEX KPI • Fraud team is often detached from the network • Regulation • Marketing • International risks/supply chain Copyright Mavenir 2019. 6 Banking • Security = Anti-Fraud • Security = Main business • Security is embedded into technology • Direct law enforcement VS
- 7. Mavenir approach Copyright Mavenir 2019. 7 Expertise • Hands-on understanding of vulnerabilities • Business drivers • Creativity Technology • Machine Learning and AI • Deep understanding of protocols Service • Knowledge • Protection • Security as a Service
- 8. Machine Learning for Security solutions Copyright Mavenir 2019. 8 Business and Technical data Technology Proficiency Controlled AI Enable and Support the Telco Transformation
- 9. Key technical principles for Machine Learning in Telco Copyright Mavenir 2018. 9 Specialized/tailored algorithms Real-Time factor (response time in under 10ms) Subject area focus rather then generic data focus Seamless integration of Machine Learning and rule engine 5G and Service Mesh architecture readiness Horizontal approach (focus beyond fraud and security)
- 10. Extending the role of NWDAF into security and Anti-Fraud SMSF NWDAF Constant insight Real-time ML Reaction to anomalies Fraud prevention Attack detection Security assurance Copyright Mavenir 2019.
- 11. Copyright Mavenir 2019. 11 Mavenir Security Suite Network/Voice Module Subscription Fraud Messaging Fraud/ SpamShield Wholesale Fraud IoT Security Credit Scoring Firewall layer: SS7, ISUP, Diameter, SIP, GTP, HTTP/2 Robocall control AI and Machine Learning engines • Single Platform, Single architecture, multiple modules • Leveraging NWDAF architecture • Natively developed with Real-time Machine Learning • Fraud prediction based on ML • Rule-based configuration integrated with ML • Dedicated algorithms for major use-cases • Flexible reporting and real-time dashboards • Multiple integration options • NFV & Microservices architecture Real-time Fraud prevention and detection Optimal OPEX for mobile operators Revenue savings proven in the market Regulator compliancy Efficient subscriber protection Future proof Data Anti-Fraud
- 12. Copyright Mavenir 2019. Machine Learning / AI for Automation and Insights Insight IoT Profile 1 IoT Profile 2 IoT Profile 3 Profile Change Pattern Network Function RuleEngine Mediation Featurization Machine Learning Learning Module Reporting & Investigation Alert Management FMS Analytics Additional Data Feeds Standard market deployed architecture addressing: DDoS detection and IoT service anomaly detection, Voice & Messaging Fraud, Signaling and Service protection 12
- 13. Copyright Mavenir 2019. 13 Thank you