WIXLIB

Autonomous Driving Network Special Issues Issue 1iMaster NCE Enables Intelligent O&M and Advances ADNiMaster NCE: Intelligent Management and Control Unit for FBB07Cloud AI Telecom AIOpsiMaster NAIEKnowledgeO&M designData Model Developmentasset & processTrainingdataManager ControllerAnalyzeriMaster MAEAIalgorithmsManager ControllerAnalyzeriMaster NCENetwork AI Real-time closed loopMBBFBBSimplified networkServiceorchestrationIntelligent O&MiMaster AUTINAll-IntelligenceThe telecom industry has been exploringdigitalization, automation, and intelligence.Initially, customers hoped to use SDN, NFV,and cloud technologies to improve serviceand network agility as well as reduce O&Mcomplexity and costs. Progress has beenmade in SDN with regard to net workautomation in scenarios such as datacenter (DC ), WAN opt imizat ion, andSD-WAN; whereas NFV is equipped withimproved resource utilization in vIMS,vEPC, and vCPE scenarios. In addition,DevOps processes and continuousintegration and delivery tools improveproduct development efficiency. In general,network automation still lags behind interms of technological development — itcannot address the complexity inmanaging and operating old and newnetworks, or performance issues regardinglarge-scale application. A widespreadchallenge in the industry involvesdiscovering an appropriate method toimprove network O&M efficiency.The autonomous driving network (ADN)was developed under this context. It takessimplified networks and simplified O&M asits core philosophy, and aims to drive thetelecom industry from automation tointelligence by employing variousintelligent technologies and leveragingNE AI Real-time awarenessFigure 1: Architecture of the ADN solutionintegration advantages. Products are planned,designed, and developed according to this philosophy.ADN adopts a three-layer open architecture, whichintroduces AI into edges, networks, and clouds. Theproduct components include the cloud network AI unitiMaster NAIE, cross-domain intelligent O&M unitiMaster AUTIN, network intelligent management andcontrol units iMaster MAE and iMaster NCE, as well asnetwork infrastructure with built-in AI. The ADNsolution includes fixed broadband (FBB) and mobilebroadband (MBB) sub-solutions that enable carriers toincrease quality, revenue, and efficiency while loweringcosts.iMaster NCE is an intelligent management andcontrol unit for FBB. As the first system in theindustr y to integrate Manager, Controller,Analyzer, and AI, iMaster NCE effectively connectsphysical networks with business intents andimplements centralized management, control, andanalysis of global networks. It enables resourcecloudification, full-lifecycle automation, and dataanalytics-driven intelligent closed loops accordingto business and service intents, as well as providesopen network APIs for rapid integration with ITsystems. Mainly utilized in carrier networks, DCs,enterprise campuses, and enterprise private lines,iMaster NCE creates simple, smart, open, andsecure networks while also accelerating servicetransformation and innovation for carriers andenterprises.ITPlanningOSSApplicationiMaster NCE OfferingNCE-IPOptimizationConstruction MaintenanceNCE-TAnalyzerOpen NBIAPIorchestrationSouthboundintegrationManager ControllerNCE-SuperCOVER ARTICLENCE-FANTransportnetworkProduct Design Philosophy:Single-Domain AutonomyAs telecom services become intelligent and agile,telecom networks increasingly integrate withoperation systems. Based on servicecharacterist ics, net work technologies, andmaintenance patterns, carriers divide basicnetworks and operation systems into differentcombinations, with each one being referred to asan autonomous domain. Autonomous domainscan independently complete entire processes(covering data collection, analysis, control, andoptimization) and provide agilenetwork-as-a-service (NaaS) for the outsideworld. Conventional network operation systemsinvolve complex processes and collaborationamong dozens or even hundreds of systems,which leads to complex and time-consumingservice provisioning and fault diagnosis. Operationsystems can be simplified by integrating therelated net work management , control, andanalysis systems so that they can evolve fromautomation to autonomy in OVER ARTICLEDatanetworkFigure 2: Architecture of iMaster NCEManager, Controller, and Analyzer in iMaster NCEcan be flexibly deployed in different scenarios,and single-domain NCE can evolve tomulti-domain NCE.Manager Controller Analyzer:Scenario-Centered Closed-Loop AutonomyiMaster NCE combines conventional elementmanagement systems (EMSs), SDN controllers,and performance and traffic analysis systems. Toensure that data can be shared among Manager,Controller, and Analyzer, it uniformly modelsnetwork, service, and user experience data. Inaddition, iMaster NCE provides a unified servicemodel and real-time network status awareness,enabling automated service provisioning andclosed-loop service optimization.Fo r e x a m p l e , i M a s t e r N C E i s d e d i c a t e d t oimplementing automation and closed-loopa u t o n o m y t h ro u g h o u t t h e l i fe c y c l e o f D Cnetworks. During the planning, design,08

COVER ARTICLEinstallation, and deployment phases, typical networks are designed automatically, one-click deploymentis supported, and basic network construction or capacity expansion can be completed within 1 hour.During the service provisioning and change phases, no network configuration errors are made andservices can be automatically rolled out within a few minutes. During the monitoring and O&M phase,typical faults can be automatically detected, diagnosed, and rectified within minutes, ensuring thatservices run stably.AI: Enhanced Closed-Loop Autonomy in Complex ScenariosTelecom network services are complex, uncertain,and subject to changes. Specifically, diversifiedservices pose various requirements on networks,and network traffic fluctuates while service pathschange during different time periods. If a networkfault occurs, troubleshooting is driven by customercomplaints or network alarms. Experts needoperations support systems (OSSs), EMSs, or toolsduring manual analysis, decision-making, andfault closure. In complex scenarios, this methodcannot be used to quickly rectify faults or ensureservice experience.As big data and AI technologies mature, thetelecom industry is starting to introduce AItheories and engineering algorithms into complexand uncertain scenarios, including fast faultdiagnosis and precise traffic prediction, to enhanceclosed-loop autonomy capabilities. To digitalizeexpert experience, iMaster NCE injects a broadrange of management rules, troubleshootingmethods, and additional knowledge into ac e n t ra l i z e d k n o w l e d g e b a s e t h a t c a n b eunderstood and used by computers. It alsoexplores the industry's mainstream AI algorithmsand matches the optimal ones with specificscenarios to provide proactive analysis andprediction.COVER ARTICLEAutonomous Driving Network Special Issues Issue 1iMaster NCE Enables Intelligent O&M and Advances ADNOpen Programmability:Simplified O&M andOpen EcosystemTelecom network O&M is complex.One reason for this is that telecomnetworks usually consist of devicesfrom multiple vendors, creatingdifficulties in E2E service provisioningand maintenance. Another reason isthat the networking and technicalsolutions for different networks arenot uniform, leading to differentO&M processes and approaches aswell as low reusability.By offering an abundant set of APIsor scenario-specific O&M apps,iMaster NCE simplifies network O&M,opens network service capabilities,and enables third-party ecosystems.An API proxy service enables iMasterNCE to offer a unified API access andauthentication mechanism as well asnorthbound and southbound networkprogrammability. Additionally, iMasterNCE supports orchestration of atomicAPIs (as scenario-specific APIs) andO&M automation (based onworkflows), and its model-baseddevelopment (MBD) framework canadapt to third-party devices to deliverfast , programmable third-part ynetwork integration capabilities.4 Engines in the Core of theProduct ArchitectureThe architecture of iMaster NCE can be summarized as onecloud platform, one digital twin network map, four engines,and one API gateway. Based on a unified cloud platform inmicroservice-oriented architecture, iMaster NCE is equippedwith cloud architecture capabilities such as auto-scaling andload balancing. The digital twin network map is responsiblefor network data governance and digital mapping of physicalnetworks to support upper-layer system applications. Fourengines (automation, intent, analytics, and intelligenceengines) collaborate to implement network automation andintelligence, and iMaster NCE's capabilities at each layer canbe opened through APIs to quickly integrate with peripheralsystems.AI training centerDesign StudioThird-party appOpen API nceIntent engineIntelligence engineAutomation engineAnalytics engineWorkflowPipelineSND/SSPCloud gure 3: Product architecture with four engines as the core0910

COVER ARTICLEAutonomous Driving Network Special Issues Issue 1iMaster NCE Enables Intelligent O&M and Advances ADNAutomation Engine:Automated Network ServicesThe automation engine integrates two importantcomponents: Manager and Controller. It providesfunctions such as automatic NE management,network service management, network pathcomputation, and network optimization. Whenusers deploy services on a network, this engineautomatically computes and delivers service paths.In addition, Controller can centrally schedule globalresources to meet SLA requirements and improveresource utilization. In MBD mode, the automationengine quickly integrates and manages third-partyNEs as well as expands network serviceautomation capabilities. It opens NE- andnet work-layer capabilities to support quickintegration with third-party systems and networks.Intent Engine:Deep Automation in Complex ScenariosIntents refer to users' expectations for networkand service status. In this regard, the intent engineenables networks and services to run based onusers' intents. The intent engine of iMaster NCEcan convert simple user language to complexnetwork models and language, automaticallygenerate call plans for services in each domain,and break down complex tasks. For example, toprovision private line services across multiplen e t w o r k d o m a i n s t h a t u s e d i ff e r e n t p i p etechnologies (such as L3VPN or PWE3), users onlyneed to specify the physical locations of thesource and sink sites as well as bandwidthre q u i re m e n t s . T h e i n t e n t e n g i n e c a n t h e nautomatically select the optimal network path,optimal network resources, designated pipetechnology, and appropriate QoS profile tocomplete service provisioning efficiently.The intent engine also supports automatic closureof intents. This involves users predefining theexpected status (such as SLAs) of scenario-specifict a s k s , a s w e l l a s c o r re c t i v e m e a s u re s a n dimplementation policies when the status does notmeet expectations. If the intent engine discoversthat the actual status and expected status areinconsistent, it automatically takes the predefinedmeasure (for example, it recomputes paths oradjusts bandwidth) to maintain networks orservices in the expected status, thereby achievingautonomous network O&M.5 Key CapabilitiesCOVER ARTICLEWeb UIIn terms of product design, development, andverification, iMaster NCE provides a series of keyc a p a b i l i t i e s , s u c h a s s y s t e m r e l i a b i l i t y,multi- dimensional data visualization, faultdiagnosis and recovery in minutes, closed-loopautomation, and open programmability, enablingintelligent O&M and approaching ADN on FBBnetworks.Analytics Engine and Intelligence Engine:Proactive Analysis and ForecastMulti-Dimensional Reliability Technologies,The analytics engine mainly provides the datacollection, storage, and analysis functions. Byintroducing the Telemetry protocol, the analyticsengine collects network and service KPIs inseconds, detects network and service status in realtime, processes and stores network data, extractsdata characteristics, and classifies raw data. Theresults are used as input for the intelligenceengine, which in turn uses machine learning andAI technologies and algorithms to conductinference on the data and provide analysis results.The intelligence engine also verifies and evaluatesthe advantages and disadvantages of AI models,and continuously optimizes the models throughlocal training to increase analysis accuracy.In the future, network data will be large in scale,multi-dimensional, and of a high frequency;therefore, it will be difficult for humans toanalyze, summarize, and evaluate data. Toovercome this challenge, the analytics andintelligence engines of iMaster NCE quickly extractkey information and provide guidance andsuggestions, enabling users to proactively forecastand prevent faults.iMaster NCE uses the local redundancy and remotedisaster recovery (DR) mechanisms to ensure highsystem reliability. The local system supportshierarchical redundancy protection, including 1 1protection for hardware, master/slave protectionfor databases, cluster protection for the softwareplatform, active/standby or cluster protection forapps, and active/standby protection for clientaccess. When a single object is faulty, the localsystem automatically switches to its standbycounterpart within 1 minute to quickly recoverservices.Client accesslayer1Floating IP soleRouteagentAPPMicroService1Keeping the System e1APPMicroService2DBProxyDBProxyVMVMData layer e 4: Local redundancy mechanismDR protection is available between primary and secondary sites, and it relies on database replication. TheDR service is located at the application layer (independent of hardware or virtualization platforms), and itachieves fast fault detection and automatic switchover between the primary and secondary sites. If theprimary site fails, the secondary site can take over services in less than 15 minutes.Third siteArbitrationArbitration linkPrimary siteDR HFSArbitration linkHeartbeatSynchronizeSynchronizeSynchronizeNCE DRDR serviceSecondary Figure 5: Remote DR mechanism1112

COVER ARTICLEAutonomous Driving Network Special Issues Issue 1iMaster NCE Enables Intelligent O&M and Advances ADNDigital Twin Network Map,a main monitor for network driving. This mapVisualizing Multi-dimensional Data for ADNcombines multi-dimension information fromTopology is frequently used as an O&M entrance.Conventional EMSs use topology as a visualizationmethod, which combin es with basic alarminformation and status to help monitoringpersonnel understand networks. iMaster NCEdifferent applications and environments into onemain topology based on scenarios, therebyFault detection3,000,000 （every day）Mass alarmsoffering unified management and O&M of allexperience to network administrators.AlarmBoard �fault）builds a digital twin network map that aims to beKPIIncremental Rendering by DataOverlayLabel addedConfigurationKnowledgeAlgorithmRoot cause（minimum unit for troubleshooting)Fault visualizationFault tree(expert experience)Faultpropagation diagram(AI knowledge graph)Accurate site visit & repair(hardware, line,and other faults)Auto rectification & closure(software, performance,and other faults)Figure 7: Framework of the "Fault 1-3-5" serviceIncremental Rendering by DataOverlayAttachment addedLabel addedDataOverlay3(Name,Node)Label addedNE6 addedNE210.10.10.2DataOverlay2(Service Path)DataOverlay1(Alarm)NE2NE1TopologyFault rectificationOne fault, one ticket300 （every day）Line faultAIinferenceresources as well as one-stop operat ionFault diagnosisCOVER ARTICLENE310.10.10.4BGP RouterID:#1244566 NE110.10.10.1BGP RouterID:#123645NE5NE4Topology SkeletonWith DataOverlaysNE6 addedNE5NE3NE4TopologySkeletonTime10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00Figure 6: Digital twin network mapBy cont inuously collect ing and storing NEinventory data, service configuration data,performance data, alarms, and logs, Manager andController digitally map the physical network tothe digital twin network map. Based on models,the digital twin module automatically generates anetwork topology and intelligently restores thephy sical net work topology, enabling O&Mpersonnel to quickly recognize network andd e v i c e c h a ra c t e r i s t i c s a n d i m p r o v e O & Mefficiency. At every moment, multi-dimensionalinformation is saved as snapshots to recordcurrent and historical network status. Historicalstatus can be replayed on demand, and snapshotsat different time points can be compared tointuitively display multi-dimensional networktrends.AI-Powered "Fault 1-3-5" Service, Achieving L3 ADN on DC NetworksNetwork autonomy is the ultimate goal of ADN. To achieve this, humans need to gradually withdrawfrom participating in network O&M. iMaster NCE provides a "Fault 1-3-5" service, which refers to faultdetection in 1 minute, diagnosis in 3 minutes, and rectification in 5 minutes. It is the foundation forreducing manual O&M operations and gradually achieving ADN.13Incorporating Huawei's exert experience, whichfault, one ticket." On Henan Mobile's 5G transporthas been accumulated over the last 30 years, thenetwork, the intelligent alarm service diagnoses"Fault 1-3-5" service further develops new faultapproximately 200,000 alarms as 400 faults daily,analysis rules through continuous fault modelwith over 90% of faults being matched to tickets.training. With the help of AI and knowledgeiMaster NCE also applies "Fault 1-3-5" to DCinference, it can quickly identify the root causes ofnetworks. Based on technologies such as real-timefaults, judge network exceptions, forecast networkcollection with Telemetry, network health check,risks, and recover services manually orknowledge graph optimization, and intent-basedautomatically based on intents.intelligent decision, iMaster NCE can proactivelyFault 1-3-5 is the foundation of intelligent alarmdetect 80% of DC network faults within 1 minute,management in iMaster NCE, and it significantlyidentify the root causes of 75 types of faultsreduces the number of alarms through grouping,within 3 minutes, and rectify typical faults withinclassification, and noise reduction, as well as5 minutes due to continuous learning, training,makes intelligent recommendation possibleand optimization. In this way, DC networks arethrough accurate root cause analysis (RCA). Intaking the lead in achieving L3 ADN, that is,collaboration with customer OSSs, it ensures "oneconditional autonomous driving.14

COVER ARTICLEIntent-Driven Closed-Loop Automationevaluation. Users can intervene and select actionsIntent-based network autonomy requires constantintent assurance, that is, continuous serviceprovisioning. Intent-driven closed-loop automationimplements autonomous management of theentire service lifecycle.or let the system take actions automatically. If noECA policy has been defined, users can select theclosed-loop policy provided by the system toautomatically re-allocate network resources andrestore services.For example, to balance traffic on a DC network,users first define service intents for balancingIntent enginetraffic on leaf nodes. Based on the intents,Intentcatalog ECA n timeMakes decisionsbased on kes actionClosed-loopanalysisIntentdecisionmonitoring tasks are automatically created anddelivered to the intelligent analysis module, whichcollects and analyzes traffic and other KPIs on eachleaf node in real time. If the traffic on a leaf nodeexceeds the preset threshold, an event will bereported to the decision module, which thenMonitorseventsIntent statusmonitoringdecides (based on initial service intents) that someVMs need to be migrated from that node toalternative idle leaf nodes. After the decisionpasses simulation verification, iMaster NCEautomat ically delivers ser vice and net workNetworkoperationsAutomation engineNetworkmonitoringResourcestatusAnalytics engineconfigurations to keep traffic well balanced amongleaf nodes.COVER ARTICLEAutonomous Driving Network Special Issues Issue 1iMaster NCE Enables Intelligent O&M and Advances ADNDesign Studio, Building a Secure andEasy-to-Use Network Openness PlatformCurrently, AOC hashelped China UnicomGuangdong develop afunction to automaticallyactivate multi-vendor IPRAN devices in 5Gtran

NCE-WAN IT Access network Transport network Data network OSS Application Planning Construction Maintenance Optimization Open NBI API orchestration Southbound integration Analyzer Manager Controller Manager Controller Analyzer integration advantages. Products are planned, designed, and developed according to this philosophy.