Native Network Management using Artificial Intelligence over an Adaptive B5G Network

3.1. Terminology and Abbreviations Theorem

SDAF: Security Data Analytics Function¶

SBA: Service-Base Architecture¶

SBI: Service-Based Interface¶

NWDAF: Network Data Analytics Function¶

AF: Application Function¶

AMF: Access and Mobility Management Function¶

AOI: Area of Interest¶

ML: Machine Learning¶

MTLF: Model Training Logical Function¶

PCF: Policy Control Function¶

UPF: User Plane Function¶

SMF: Session Management Function¶

NF: Network Function¶

UE: User Equipment¶

gNB: gNodeB¶

SBMA: Service-Based Management Architecture¶

SIEM: Security Information and Event Management¶

SOAR: Security Orchestration, Automation, and Response¶

AnLF: Analytics Logical Function¶

MTLF: Model Training Logical Function¶

3.2. Autonomous Network

The autonomous network concept is defined differently depending on the standardization organization, and these contents are as follows.¶

3GPP: SON (Self-Organizing Networks)¶

ETS/ITU-T/GSMA: Autonomous Network¶

ETSI: ZSM (Zero touch network and Service Management)¶

Hauwei: AND (Autonomous Driving Network)¶

Juniper: SDN (Self-Driving Network)¶

Cisco: DNA (Digital(Data) Network Architecture)¶

Ericson: ZTN (Zero Touch Network)¶

Autonomous network levels can also be divided into six different levels.¶

Level 0 - Manual Network: The system is supported by a monitoring function to manually execute dynamic tasks (SNMP, CLI)¶

Level 1 - Assisted Network: The system executes specific and repetitive subtasks that are preconfigured to increase execution efficiency (Tack-Centric)¶

Level 2 - Partial Autonomous Network: The system enables closed-loop O and M for specific devices based on AI models in specific external environments (Node-Centric)¶

Level 3 - Conditional Autonomous network: L2-based system has functions to detect real-time environment change, specific network domain, and intention device. Semi-closed loop management is possible to optimize and adjust to the external environment (Service-Centric)¶

Level 4 - Highly autonomous Network: L3-based system has capabilities of service- and customer-experience-centric in a more complex cross-domain environment. It can analyze and make decisions based on predictive or active closed-loop management of the network(User-Centric)¶

Level 5 - Fully autonomous network: The system is a fully autonomous network with multi-services, multi-domains, and full lifecycle (Value-Centric)¶

4.1. Model and framework structure for SBA structure linkage

In this study, we aim to design at the NF level to internalize an AI-based autonomous security control model in the SBA structure of 5G System Architecture. There are 11 major NFs of SBA such as AMF, SMF, UPF, NSSF, NEF, NRF, PCF, UDM, UDR, and NWDAF. Among the NFs constituting SBA, NWDAF is used and analyzed NF that utilizes intelligent technologies such as AI for network operation. The NWDAF is selected as a reference model to design AI-based security analysis functions using network data.¶

4.2. Model and framework structure for SBMA structural linkage

The SBMA structure defined by 3GPP into account is the Management Plane in the SBA structure. The prior study was designed by setting the SBA structure considering the control plane and the user plane. We will analyze MDAS or MDAF in the SBMA structure and conduct research and conceptual design in consideration of security management in the future.¶

TBD¶

4.3. Model and framework structure for AI-Enabled network structural linkage

This structure aims to the conceptual design of element technology for an AI-based autonomous security control model over the 6G networks. AI-Enabled Network is designed using SBA's AI-based NWDAF. Since NWDAF is an NF that analyzes network data for network operations, it seems necessary for Security NF of security analysis. SDAF is designed and analyzed using the NWDAF Wrapper method for AI-based security of NF. NWDAF is an analysis of NF using AI and it can consist of logical functions (AnLF and/or MTLF). It is also the conceptual design of elemental technology for security intelligence with the two functions of security internalization (NWDAF and SDAF). It can need an AI Model Training for security intelligence. The following shows the structure according to two candidate designs and the proposed model.¶

TBD¶

4.4. NWDAF (Network Data Analytics Function)

NWDAF is one of the network functions located on the control plane in the SBA structure of 5GC. It is based on 5G Core, MEC (Cloud), and user equipment (UE) in edge networks. It is also used with data collection and data analysis depending on the application function (AF) and Operations and Administration Maintenance(OAM). The purpose of NWDAF is to simplify the complexity of interfacing with 5GC and 3rd analytic solution providers. 5GC-related data in NWDAF is collected with 5G network data and the data can be analyzed by machine learning and statistical analysis. The analyzed result data is provided to other 5G core network functions to optimize each network function and to improve performance as its main function.¶

     +----+
     +NF-1+----------------+--------------------+             +----+
     +----+                + NWDAF              +-------------+NF-1+
+-------------+   +-----+  +                    +             +----+
+ Untrusted AF+---+NEF-1+--+                    +
+-------------+   +-----+  +  +--------------+  +             +-----+
+----------+               +  +analytic Model+  +-------------+NEF-1+
+Trusted AF+---------------+  +(Static,      +  +             +-----+
+----------+               +  + algorithm)-3 +  +
     +-----+               +  +--------------+  +-------------------+
     +UDR-1+---------------+                    +
     +-----+               +                    +             +-----+
     +-----+               +                    +-------------+OAM-2+
     +OAM-2+---------------+                    +             +-----+
     +-----+               +--------------------+
----------------------------------------------------------------------------
    .................................................
    .  1: Core NF             2: Network Management .
    .  3: Proprietary Function                      .
    .................................................

     NWDAF
     +--------------------+       +--------------------+
     + AnLF               +       + MTLF               +
     + (Analytics         +       + (Model Training    +
     + Logical Function   +       + Logical Function)  +
     +--------------------+       +--------------------+

4.5. Management for other Standardization

NFV (Network Functions Virtualization) Management and NFVO (NFV Orchestration): Identify network function (NF) lifecycle management procedures in virtualized environments¶

OSM (Open Source MANO): E2E Network Service Orchestrator (NFV + Slice + Cross Domain)¶

MEC (Multi access Edge Computing) Management, F5G (Fifth Generation Fixed Network)¶

ENI (Experiential Networked Intelligence):¶

(a)Cognitive Netwok Management architecture using AI and Context aware Polices¶

(b)Method to add intelligence to legacy systems¶

(c)Method between API Broker layer and Legacy System (OSS/BSS, NF, User, etc. and ENI system)¶

ZSM (Zero Touch Network and Service Management):¶

(a)Fully Autonomous Management and Operations Framework level¶

(b)Management Services of Domain, Unified Fabric and Cross Domain¶

(c)Closed Loop Control Acquisition, Analysis, Determination and Execution Method Actions¶

(D)Domain level Management, Cross Domain Level Management, Business Service Level Management¶

This document present to aim B5G-based autonomous security and intend framework based on the constant security quality guarantee to provide the super-trusted infrastructure of the new convergent network security service without cyber threats¶

5.1. Purpose of B5G Native Network Management Framework

It is necessary to verify native security element skill to analyze the detailed functions such as B5G wireless access, D2D and infrastructure virtualization. It is also needed to analyze B5G global network security-based intelligence and internalization technology, security vulnerability in flying base station and quantum security for security application system. We also propose of design of B5G native network management and requirement for B5G wireless access/D2D/infrastructure virtualization attack model, AI-based B5G autonomous security control of security native intelligence, networking security and intrusion detection in flying base station and quantum security for application of B5G security system. In addition, security native modeling and verification are also necessary in B5G native network management framework.¶

5.2. Scope of B5G Native Network Management

Above all, the first scope is to analyze of B5G wireless access/D2D/infrastructure virtualization elements and to define security requirements such as B5G wireless access, Ultra-high-density of B5G D2D and infrastructure virtualization. In next scope, there are B5G wireless access/D2D/infrastructure virtualization attack model development and threat analysis, design of AI-based B5G autonomous security control and security intelligence internalization concept. It is also necessary to design networking security, intrusion detection element technology in flying base station and design of quantum security technology for B5G security application. Verification of the contents in advance is also additionally required with the following scope.¶