《电信管理论坛（TMF）：2024探索自智网络的最佳实践研究报告：完全自智网络发展路径中的运营商定位（英文版）（51页）.pdf》由会员分享，可在线阅读，更多相关《电信管理论坛（TMF）：2024探索自智网络的最佳实践研究报告：完全自智网络发展路径中的运营商定位（英文版）（51页）.pdf（51页珍藏版）》请在三个皮匠报告上搜索。

1、Autonomousnetworks:in search of best practiceDecember 2024|www.inform.tmforum.orgSponsored by:Authors:Richard Webb,Senior AnalystDawn Bushaus,Contributing AnalystEditor:Ian Kemp,Managing Editor Autonomous network operationsBENCHMARK REPORT2contents25section 4:success stories AN best practice in acti

2、on07section 1:the business case for autonomous networks19section 3:creating a blueprint for Level 4 autonomous networks04the big picture1232section 2:survey results where are CSPs on the AN journey?section 5:award-winning Catalysts03about our sponsors38section 7:key fjndings and recommendations3540s

3、ection 6:collaboration on standards is key to AN interoperabilityadditional features&resourcesabout our sponsors3inform.tmforum.orgBENCHMARKAt Nokia,we create technology that helps the world act together.We put the worlds people,machines and devices in sync to create a more sustainable,productive an

4、d accessible future.We are a B2B technology innovation leader pioneering networks that sense,think and act.ZTE Corporation is a leading global provider of telecommunications solutions and services,delivering innovative technologies that power networks,enterprises,and consumers worldwide.Established

5、in 1985,ZTE operates in over 160 countries,specializing in comprehensive end-to-end solutions across fjxed,mobile,IT,AI and enterprise networks.With expertise in 5G,Autonomous Networks(AN),cloud computing,AI,and IoT,ZTE empowers customers to achieve network intelligence,operational effjciency,and su

6、stainability.Renowned for its commitment to research and development,ZTE drives cutting-edge advancements that enable seamless connectivity and digital transformation,building a smarter,more connected future for its global partners and communities.NokiaZTETo learn more about the benefjts of autonomo

7、us networks and how our sponsors are helping to facilitate their development,watch these videos:the big pictureThis report draws on some of our research this year to explore how AI and,more explicitly,automation can address these challenges.It looks at what CSPs are doing in terms of network automat

8、ion both within TM Forums member community and more widely,and attempts to identify some of the best practices of operators striving to go beyond simple automation to implement autonomous networks(AN).As part of this new research we conducted a survey which drew responses from 111 individuals at 82

9、companies(see section 2).Among the primary drivers for those respondents to implement AN are improving customer experience/satisfaction;simplifying operations and maintenance and improving personnel effjciency;and increasing resource effjciency and utilization of network and IT.Among large operators

10、 those with more than 100 million customers 100%said supporting agile service delivery and accelerating time to market/monetization was the most important driver.But there are many other reasons.Autonomous networks are not simply a technical evolution;they represent a transformative change in how ne

11、tworks operate,adapt and serve CSPs and their customers.Whereas automation relies on predefjned rules,network autonomy involves intelligent systems making independent decisions without input from humans.In brief,much of the value for CSPs lies in the ability of AN to address three critical areas:Ope

12、rational effjciency.AN optimizes resource management through autonomous processes,managing network complexity and enabling CSPs to signifjcantly reduce manual interventions and operational costs.Customer experience transformation.Customers demand reliable,seamless connectivity and personalized servi

13、ces.AN,based on an intent-driven architecture,enables CSPs to meet these expectations by proactively managing network quality and delivering service level guarantees.Revenue and innovation potential.AN unlocks new business models,such as network-as-a-service(NaaS),enabling CSPs to monetize their net

14、works beyond traditional connectivity.In the past couple of years we have published reports that have set out clearly the challenges for communications service providers(CSPs)in the face of increasing network complexity,limited return on investment in 5G and growing expectations for improved custome

15、r experience.Our annual telco revenue growth Benchmark reports have highlighted how diffjcult it is for CSPs to establish new revenue streams and drive operational effjciency in order to compete with more agile start-ups and hyperscale cloud service providers.More recently,the emergence of AI has pr

16、oven highly disruptive,but also represents signifjcant opportunities for operators.4inform.tmforum.orgBENCHMARK5the big pictureinform.tmforum.orgMoving up the AN levelsTM Forum is at the forefront of industry-wide collaboration to drive the deployment and evolution of AN technology,processes and bus

17、iness models.At the heart of that work is the Autonomous Networks Project.Ultimately,CSPs need to be able to scale orchestration of zero-touch,zero-wait and zero-trouble services end to end across difgerent network domains(for example,radio access,fjxed access,core,IP,optical transport and data cent

18、er/cloud networks).This requires automation of the entire service lifecycle,from ordering to fulfjllment,activation,orchestration,management,assurance,optimization and billing.In conjunction with its community of CSP and solution provider members,the Forum has developed a six-level taxonomy for CSPs

19、 to measure their progress in implementing autonomous networks.Each level has a set of characteristics describing the evolutionary stage of the CSPs journey from manual to fully autonomous operations (see graphic).According to TM Forums defjnition,Level 4 autonomous networks represent a major shift

20、from traditional human-defjned automation processes to true autonomous decision-making.This is a signifjcant leap from Level 3,where machines assist humans in decision-making but still rely on human oversight.Today,most operators are between Levels 2 and 3.But until recently its been diffjcult to ga

21、uge CSPs progress deploying AN because there was not a standardized way to evaluate and compare implementations.That changed in 2024 with further standardization of TM Forums methodology and tools for evaluating AN levels,as well see in sections 2 and 3 of this report.Autonomous network levelsFully

22、autonomous network:The system has closed-loop automation capabilities across multiple services,multiple domains(including partners domains)and the entire lifecycle via cognitive self-adaptation.Highly autonomous network:In a more complicated cross-domain environment,the system enables decision-makin

23、g based on predictive analysis or active closed-loop management of service-driven and customer experience-driven networks via AI modeling and continuous learning.Conditional autonomous network:The system senses real-time environmental changes and in certain network domains will optimize and adjust i

24、tself to the external environment to enable,closed-loop management via dynamically programmable policies.Partial autonomous network:The system enables closed-loop operations and maintenance for specifi c units under certain external environments via statically confi gured rules.Assisted operations a

25、nd maintenance:The system executes a specifi c,repetitive subtask based on pre-confi guration,which can be recorded online and traced,in order ciency.Manual operations and maintenance:The system delivers assisted monitoring capabilities,but all dynamic tasks must be executed manually.0143256the big

26、pictureinform.tmforum.orgIndustry collaboration on ANMore than 60 CSPs worldwide have signed TM Forums AN Manifesto,pledging to work towards Level 4 autonomy,and dozens are participating in the Forums AN level evaluation.Even more have participated in AN Catalyst proof of concepts.More widely,across

27、 the telecoms industry,more than 120 AN standards and research projects have been initiated,with multiple standards development organizations and open-source groups working to defjne and deploy ANs(more on this in section 6).Within TM Forums Autonomous Networks Project many CSPs have outlined their

28、AN visions and strategies,and 24 have carried out internal AN level assessments.A handful of CSPs have publicly stated they are adopting AN Level 4“high-value scenarios”,aiming to reach Level 4 for operational fmows in some domains from 2025 to 2027.At Level 4,the network can self-manage,self-optimi

29、ze and handle complex tasks.To achieve this level of autonomy,CSPs need to translate their customers business goals,or“intents”,into network requirements that can be fulfjlled without customers needing to understand the underlying infrastructure.Dave Milham,TM Forums Chief Architect,describes this t

30、ransition:“Moving to AN Level 4s intent-based interfaces is a move from prescriptive supervision automation to declarative,delegated autonomy,in which AI does the reasoning.”Transitioning from automation to autonomy is a signifjcant business and technology challenge.In November 2024,TM Forum publish

31、ed a blueprint framework for achieving AN Level 4,which emphasizes AI being embedded into all domain design,operation and processes,such as closed-loop orchestration,to enable intent-based services.Section 3 provides an overview of that blueprint.“It is vital that AN leverages intent,so that instead

32、 of selling technology CSPs are selling connectivity that can be tailored to business outcomes,”said TM Forum CTO George Glass in his keynote address at Innovate Asia24.The importance of intent in AI-driven networks and services is discussed further in the next section.This inter-dependency between

33、AI and AN is at the heart of an on-going industry dialog around AI for networks,versus networks for AI.AI is critical for the evolution of AN,but AN is also critical to support future AI-driven demands on connectivity in other words,ensuring that networks are AI-ready.CSPs must adopt AI for networks

34、,shifting from static,rules-based network automation to AI-enabled,dynamic orchestration across multiple network domains from 5G RAN to IP transport and optical core.We look at the role of AI in autonomous networks in the next section,as well as the business case for AN.Read this report to understan

35、d:The business goals and visions driving AN adoption Which areas CSPs are prioritizing for AN implementation The relationship between AI and intent-based automation The blueprint for achieving AN Level 4 How AN best practice is being defjned in real-world network deployments.To learn more about how

36、CSPs are transitioning to Level 4 autonomous networks read TM Forums Journey Guide:7section 1:the business case for autonomous networksinform.tmforum.orgIn our survey for this report,we asked about the challenges CSPs face when implementing AN.The top three challenges are complex integration challen

37、ges across domains,lack of budget,and a lack of a clear path to Level 4 AN.Anton Reynaldo Bonifacio,EVP,Chief Information Security Offjcer and Chief AI Offjcer,Globe Telecom,speaking at Innovate Asia24,emphasized the importance of establishing an AI-fjrst culture to tackle some of the inherent chall

38、enges and enhance growth and security in autonomous networks.In Globes case,this is based on a democratized,bottom-up approach that is more tactical than strategic.For most CSPs,adoption of AI comes with a dual focus:using AI to optimize their networks(as part of their transformation towards AN);and

39、 ensuring that their networks can support the dynamic requirements of emerging and potentially revenue-generating AI-powered applications.BENCHMARKAN challengesSerious challengeSomewhat of a challengeNot a challengeComplex integration challenges across domainsLack of budgetThe end-to-end architectur

40、e is not defined,and there is no clear path to Level 4Key technologies required(such as AI)are not matureWe do not have the right skillsLevel evaluation is difficult/we lack the methodology or tools to make a credible evaluationLack of clear business objectives and value indicatorsAN standards are n

41、ot mature enough for implementationLack of use cases that can be applied at scaleLack of top-level strategy and visionOur data strategy does not adequately support AN implementation53%43%34%29%26%24%23%23%22%19%18%43%48%52%60%58%58%51%53%60%57%37%4%9%14%10%16%18%26%25%18%25%45%TM Forum,20248section

42、1:the business case for autonomous networks“We equip employees with AI capabilities to use where they feel it is appropriate rather than dictate use cases to focus on,”said Bonifacio.“Not chasing ROI but considering AI in terms of employee,team,business or customer impact.It is important to have a p

43、layground approach to allow for some openness to explore,but within safety net parameters,with no risk to customers.”The ability to scale use cases for AN implementation is critical to the business case.CSPs have organized their IT operations around linear software processes managed by separate team

44、s(fulfjllment,service activation and assurance,for example).These silos exist in every domain,but CSPs need to be able to scale orchestration of zero-touch,zero-wait and zero-trouble services,end-to-end across network domains.This requires automation of the entire service lifecycle,from ordering to

45、fulfjllment,activation,orchestration,management,assurance,optimization and billing.“We have an end-to-end service operation management layer on the top of the fjve difgerent technical domains,and we are specifying for each year the technical maturity level requirements,”said Dr.Lingli Deng,Director

46、and Researcher at China Mobile,in an interview for TM Forum Inform.She said China Mobile is using its end-to-end service operation layer as the catalyst to drive overall development of technical AN capabilities.Despite the implementation challenges,ANs are crucial for CSPs to realize the full busine

47、ss potential of their infrastructure investments.In our survey,we asked respondents about their AN implementation drivers.Perhaps unsurprisingly,the top-rated AN driver relates to enhancing customer experience.Autonomous networks can help CSPs understand subscriber behavior and take proactive measur

48、es to improve customer satisfaction.For instance,AI-powered analysis enables CSPs to assess network performance,such as latency at the customer level,predict failures and improve network reliability.In a keynote address at Innovate Asia24,Shanti Juanita Johari,CCO,Consumer Strategy and Business,Tele

49、kom Malaysia,spoke about driving a service culture through AI and AN.“Weve made a customer promise to restore fjxed broadband faults within 24 hours or your money back quite a bold service difgerentiator.Owning that promise is important:to drive a customer-centric culture within Telekom Malaysia and

50、 get employees thinking about the customer impact of automation activities.”inform.tmforum.orgAN driversVery importantSomewhat importantNot importantImproving customer experience/satisfactionSimplifying operations and maintenance and improving personnel efficiencyReducing the costs of operations and

51、 maintenanceIncreasing resource efficiency and utilization(network and IT)Service optimization and ensuring quality of service/quality of experienceSupporting agile services delivery and accelerating time to market/monetizationFault management using automated closed-loop operationsEnabling exposure

52、of network-as-a-service(NaaS)capabilitiesEnabling/enhancing interaction with partnersGreen energy saving to meet sustainability goals91%1%85%84%1%1%83%82%78%70%53%50%42%8%14%15%17%17%20%28%2%2%2%41%38%46%6%12%12%TM Forum,20249section 1:the business case for autonomous networksHe added that doing so

53、has changed the internal perspective,from“a fear of losing money,to positive conversation about what we stand to gain.Engineering teams are motivated by tougher KPIs,and we are now implementing this type of promise in other areas of the business.”The next two leading drivers for our survey responden

54、ts are interconnected:operational simplicity and cost effjciency.ANs enable CSPs to optimize operations and maintenance,reduce operational expenditure,and improve profjtability all key benefjts while revenue from traditional telecoms services remains static.With autonomous networking,CSPs do not hav

55、e to use resources to conduct routine tasks such as provisioning and fault detection,lowering costs and time to rectify problems.Intent-based networks Intent is the driving force of AN,communicating requirements,constraints and preferences to an autonomous system,enabling CSPs to set the goals and p

56、arameters for AI models without prescribing specifjc actions.This allows the network to adapt more fmuidly,and enables CSPs to turn their customers business goals,or intents,into instructions for the network that can be fulfjlled without customers needing to understand the underlying infrastructure

57、abstracting its complexity and communicating with customers using a common language or ontology that can also be machine readable.The graphic shows the interaction of intents across a CSPs operational layers.Intent-based management is baked into the AN journey at least as early as Level 3 of the six

58、-level model described earlier(conditional autonomous networks)and becomes essential at levels 4 and 5 highly and fully autonomous networks,respectively.To be viable,intents must be grounded in data:observed,measured and controlled intents with which to drive the autonomous system,and verifjable und

59、er dynamically changeable conditions.And to manage these inform.tmforum.orgIntent-based,closed-loop managementTM Forum,202410section 1:the business case for autonomous networksprocesses the data must be accurate and easily accessible.To automate the full lifecycle of services,intents are combined wi

60、th closed-loop management.Closing the loop means collecting/analyzing data to fjgure out how networks can be optimized and implementing those changes in an automated way.Intent can be expressed at various operational layers(business intent,service intent and resource intent,for example).In TM Forums

61、 AN Reference Architecture(IG1251),intents guide the closed loop control process,which is a continuous process of monitoring,analyzing and adjusting performance to improve the quality of experience for end users and automate the full lifecycle of services.“Intent is our dream,our aspiration,but it d

62、epends on the maturity of the infrastructure itself,”said Indra Mardiatna,CTO of Indonesian operator Telkomsel,in an interview with TM Forum Inform.“Maybe some use cases we can get to that level by 2025,but in some very complex cases we require more power and more data,and we need time.”Using intent

63、 for slicing Network slicing will be one of the fjrst use cases for intent-based automation to enable the creation of multiple virtual(or logical)networks within a shared physical infrastructure.As network slices proliferate,CSPs will need to manage and orchestrate network resources.AI can help mana

64、ge this complexity by predicting traffjc fmows and requirements,network issues and outages,and sending instructions to difgerent network functions about solutions.“Network slicing brings management,operations and maintenance complexity,”says Yessie Yosetya,Director and Chief Enterprise&Corporate Afg

65、airs Offjcer at XL Axiata.“Multi-dimensional management of hardware,resources,slice defjnition and applications are required.Only ANs can scale up to the demands of provisioning and maintaining the slice in real time.”CSPs will need intent-driven orchestration to support dynamic slicing.Whilst initi

66、al deployments of slices in the network can be done without full automation,to repeat this at scale,and handle difgerent customer requirements,it will need to be intent-driven.The role of GenAI in ANsGenerative AI(GenAI)will play an increasingly important role in AN,working together with other AI an

67、d machine learning technologies.Machine learning can be used to identify patterns in data,predict results based on historical information and make decisions,while GenAI is well suited to tasks requiring high creativity and innovation,such as intelligent Q&A for routine operations and maintenance tas

68、ks and complex network confjguration.Our survey asked respondents whether GenAI is being introduced into network operations.In total,64%of respondents said they had either already implemented it or were planning to do so.But that still leaves one third who think that GenAI is either not mature enoug

69、h for use in network operations yet or should not be used in mission-critical network functions(see pie chart above).Combined,“traditional”AI and GenAI will be able to identify constantly changing network requirements and generate optimal network confjgurations.They can also help CSPs improve networ

70、k performance and reliability by generating code and analyzing data to detect network problems.For example,an LLM can generate network topology diagrams or build a digital twin based on text descriptions and learn patterns from historical data to identify anomalies and the root cause of problems.Chi

71、na Telecom,for example,is using GenAI to build a digital twin of its network to enable root cause analysis.We asked CSPs how they are already using GenAI in the network.The majority(74%)said they are using it for network maintenance,such as automated troubleshooting and risk prediction.A high propor

72、tion are also using it for network operations,such as intelligent customer service,and network optimization,such as risk identifjcation and automated verifjcation of optimization(see graphic on the next page).Well look at some difgerences in how small and large operators view GenAI in the next secti

73、on.inform.tmforum.orgCSPs introduction of GenAI into network operations17%8%27%47%TM Forum,2024We have introduced GenAI innetwork operationsWe plan to introduce GenAIGenerative AI is not mature enoughfor use in network operations yetThe trustworthiness of GenAI is toouncertain for use in mission-cri

74、ticalnetwork functions11section 1:the business case for autonomous networksIndeed,GenAI has multiple other potential applications for network operations:Decision-making.By interpreting complex,unstructured data such as fjeld engineer reports,GenAI can refjne AI models for better resource allocation

75、and fault diagnosis.Natural language interaction.GenAI can provide intuitive interfaces for CSPs,allowing interaction with systems using natural language queries such as“How can I improve service quality in this area?”Service innovation.GenAI can analyze market trends and suggest new service ofgerin

76、gs or network confjgurations tailored to emerging customer needs.Security and fraud detection.GenAI can be used to analyze terabytes of network data to detect anomalies,enhance security and improve fraud detection.Software coding.GenAI models can be trained on large code repositories to give complet

77、ion prompts as developers write code,or to generate new code.Real-time language translation.It can also translate languages in real time,such as documentation,HR questions,summarizing meetings and so on.In an article for TM Forum Inform,Usman Javaid,Chief Products and Marketing Offjcer,Orange Busine

78、ss and Bruno Zerbib,the companys Group CTIO,argue that networks need to change to accommodate a new era of GenAI and quantum computing.“Network service providers must lay the foundations,standards,and roadmap for a network for AI that is distributed,scalable,secure,and energy effjcient,”they say.“Th

79、is is why TM Forum members are collaborating on an industry-specifjc data reference architecture,encircling both emerging AI-enabled business models and supporting networks.”Data is keyIn TM Forums Modern Data Architecture Project members are collaborating on a telecoms-specifjc data reference archi

80、tecture,incorporating emerging AI-enabled business models.The project comes from the need to modernize data architectures to support the rapid evolution of AI and help defjne cutting-edge AI-enabled telecoms operations.The ultimate aim is to realize zero-touch effjciencies for a fully AI-enabled dat

81、a architecture.Lack of an adequate data strategy was rated as a serious challenge or somewhat of a challenge for AN implementation by 76%of respondents to our survey.Indeed,establishing a robust data culture within the CSP is critical.“A strong data foundation is fundamental to intelligent operation

82、s.It underpins our whole analytics and automation strategy,”says Olivier Simon,SVP Smart Network&Data,Orange Innovation Group.“For example,network performance is accessible to marketing employees as an automated data model,and customer complaint data can be referenced by network engineers.We refer t

83、o this as a data democracy of multi-domain data sharing.”For Orange,this has already increased usage of AI in the network,and the operator is also now increasingly using GenAI for multiple tasks including:Trouble ticket summarization Technical documentation retrieval for fjeld engineers Natural lang

84、uage query on service issues Retrieving text logs for network clusters Correlating with previously successful network fjxes.In the next two sections we provide more analysis around the survey carried out for this report,giving CSPs perspectives on implementing AN.inform.tmforum.org74%TM Forum,2024In

85、telligenttraining servicesDeploymentOptimizationMaintenanceHow are CSPs using GenAI in the network?46%Energysaving30%Planning51%Operations(e.g.,intelligentcustomer service)62%65%12section 2:survey results where are CSPs on the AN journey?inform.tmforum.orgThere are some caveats to consider in analyz

86、ing this data.Most importantly,respondents are self-reporting their progress,and we do not know how each of their companies evaluates its AN level.For example,we believe it is highly unlikely that 5%of respondents companies have achieved Level 4 AN at least in terms of overall network autonomy.We in

87、tended for the question to mean the average level achieved in the telcos network operations overall,but its possible that when answering some respondents thought about a single domain or process where they have achieved Level 4.The ambiguity highlights the pressing need for standardization of how op

88、erators assess their progress in achieving network autonomy.Indeed,until very recently true benchmarking of CSPs AN levels was not even possible.In 2021,TM Forums AN Project created the Autonomous Network Levels Evaluation Methodology(IG1252)to help operators self-assess their progress based on the

89、levels defjned in the guide.But even though the methodology has been updated over time,detailed scoring for specifjc network domains and scenarios was never standardized.A group of operators and suppliers participating in the AN Project changed that in 2024 with a series of workshops to further defj

90、ne AN level evaluation methodology and develop a more comprehensive self-service assessment tool.The team began with fault management in radio access and core networks because an internal survey of project members identifjed those as key scenarios.The latest results from TM Forums assessments show t

91、hat the average AN level score for fault management was 2.5 in the core network and 2.8 in the RAN.See section 3 for anonymized charts comparing CSPs results and for more about the AN Projects efgorts to improve standardization of AN level evaluation.CSPs are making progress implementing autonomous

92、networks,but the journey is taking longer than expected.When we began surveying operators about AN at the beginning of 2020,no respondents reported that their companies had reached Level 3,and only 4%said they had reached Level 2.However,at the time about 40%saw a clear path to reaching Level 4 with

93、in four years.Now that weve reached that 2024 milestone,only 19%of respondents to the survey for this report said their companies have gone beyond Level 2 AN,and just 5%claimed they have reached Level 4(see graphic below).BENCHMARKCSPs AN levels now and in the futureTM Forum,2024L0L1L2L3L420262030No

94、w5%14%30%39%13%23%45%25%5%81%12%6%2%1%While this is TM Forums fjrst autonomous networks Benchmark report,we have been surveying CSPs about AN since 2020.Each time,we have asked some of the same questions in the same way in order to gauge progress and changes in sentiment over time.However,while some

95、 survey respondents are the same from year to year,many are difgerent.The demographic data below shows the scope of our 2024 survey.More than 60%of respondents work for small operators those serving 25 million subscribers or fewer.We defjne midsized CSPs as serving between 25 million and 100 million

96、 customers and large as serving more than 100 million.Because of the high number of respondents from small companies,we have analyzed some questions in this section by size of company,given that small operators often have difgerent needs and priorities than large ones.This year we also asked questio

97、ns specifjcally about GenAI because of its fast-growing adoption among telcos and because AI in all its forms is necessary to achieve Level 4 AN the level at which operations shift from traditional human-defjned automation processes to true autonomous decision-making.Who are the AN Benchmark respond

98、ents?13section 2:survey results where are CSPs on the AN journey?inform.tmforum.orgTM Forum,2024Note:Some graphics throughout the report do not add to 100%because of rounding.111 from respondents82companiesProductmanagementFinanceNetworkIT/OperationsConsumer lineof business24%29%5%6%Middle East and

99、AfricaAsia/Asia PacificNorth America14%GlobalSouth America/Caribbean23%EuropeWhere are respondents from?Respondents rolesSize of CSPFewer than 5 million5 million to 25 million25 million to 50 million50 million to 100 million100 million to 150 millionMore than 150 million3%1%56%32%3%Enterprise line o

100、f business5%28%4%12%14%12%30%30%32%8%16%12%2%14section 2:survey results where are CSPs on the AN journey?inform.tmforum.orgDomains and processesOur survey results reinforce the AN Project teams decisions about which domains and processes to focus on fjrst.We found that the highest-priority domain is

101、 the mobile network followed by the IP network,with the highest-priorty processes being those that focus on improving customer experience,such as fault management,service assurance and service delivery (see graphics).Focusing on the mobile network fjrst makes sense because of its complexity(see side

102、bar on page 15)and the potential for savings on energy consumption.The radio access network(RAN)accounts for about 80%of mobile network energy consumption,for example.In addition,guaranteeing customer experience in mobile networks is challenging today.Network domains CSPs are addressing firstHigh pr

103、iorityMedium priorityLow priorityMobile RANMobile coreIP backboneIP accessData centerOptical transportOptical access61%56%52%51%50%42%36%28%29%40%39%40%38%40%12%15%8%10%10%20%25%TM Forum,2024Processes CSPs are addressing firstHigh priorityMedium priorityLow priorityFault management/anomaly detection

104、 and resolutionService assuranceService deliveryTroubleshootingQuality optimizationComplaint handlingNetwork changeEnergy efficiency optimizationPlanning and designNetwork deployment83%75%70%59%58%51%43%37%37%33%14%20%24%36%36%41%50%48%46%48%3%6%6%6%6%9%7%15%17%19%TM Forum,202415section 2:survey res

105、ults where are CSPs on the AN journey?inform.tmforum.orgRANs are prone to outages because of complex infrastructure and environmental factors such as weather and physical obstructions.Automation can address these issues to improve customer experience and satisfaction.As a network director for a UK-b

106、ased telco told Capgemini Research Institute:“RAN takes precedence due to frequent incidents.With numerous data centers and thousands of base stations,theres a signifjcant opportunity for workload reduction.In critical areas like data and voice in the core,operators maintain a conservative approach

107、due to higher risks The focus on RAN is driven by lower risk and greater automation opportunities,resulting in signifjcant savings.”The IP network is a high priority for telcos,too,in part because it is easier to automate than other domains.The IP network“by design has been implemented like an auton

108、omous network with the routing protocols”,explained Telenors Terje Jensen,SVP,Global Business Security Offjcer,during a recent webinar about AN.“Thats probably one of the most advanced network domains,which is fundamental for the whole internet.”When it comes to processes,telcos are focusing fjrst o

109、n those that can improve customer experience,which in turn can lead to lower opex.“We are focusing on fault management preventing,identifying,isolating,fjxing and checking that the fault is fjxed and the service is restored,”said Alexis Koalla,Director,Operations Strategy and Transformation,Orange,i

110、n the same AN webinar.“So,we are monitoring the KPI of the effjciency the time we take to fjx or to identify or to build the confjguration and indirectly this will save money for operations.”Overcoming challengesTwo important conclusions we can draw from the answers to our survey question about AN l

111、evels are that telcos of all sizes overwhelmingly want to achieve Level 4 autonomy,and theyre optimistic they will get there within this decade.A full 81%of respondents predicted that their companies will achieve Level 4 by 2030.Weve discussed telcos potential optimism bias in our Digital Transforma

112、tion Trackers.That phenomenon,which is the scientifjcally described propensity to overestimate the likelihood of positive events while underestimating the probability of negative ones,is almost certainly at work here,as survey respondents have overestimated their companys ability to achieve Level 4

113、year after year.The empirical evidence from TM Forums assessment pilot points to the optimism bias as well (see page 21).The reasons telcos have failed to meet their Level 4 goals to date are numerous we highlighted the main challenges CSPs are facing in the chart in section 1 on Wireless networks a

114、re complex for many reasons that make them diffjcult and costly to manage.The behavior of radio signals can be afgected by physical objects and environmental conditions,leading to issues like interference and fading.Management of the limited radio frequency spectrum adds complexity because it requir

115、es effjcient allocation to avoid interference.And users are constantly on the move,requiring seamless handovers.In an efgort to reduce such complexity,China Mobile Fujian is working with its supplier ZTE on a closed-loop,intelligent network optimization solution based on digital twins a virtual mode

116、l that replicates network infrastructure and systems,allowing operators to monitor,simulate and optimize the network in real time.The solution builds high-precision twin models to conduct quantitative pre-evaluations on KPIs like network coverage,capacity and interference.The results so far have bee

117、n impressive.Compared with traditional methods,China Mobile Fujians digital-twin approach has reduced labor costs by about 25%.The cost of vehicles and equipment is down 29%,and the number of dispatching fjeld work orders has been reduced by 34%.In addition,the network optimization cycle has been sh

118、ortened by 70%.China Mobile Fujian has applied its digital twin-based AN solution in a variety of settings.For example,it has been used to improve and enhance the network performance of major traffjc intersections in Fujian.After optimizing the networks of 25 intersections covering 30,000 customers,

119、users network speed has increased from an average of 12Mbps to 18Mbps,and the number of complaints per 10,000 users has dropped from fjve to two.The company has also provided experience guarantees for 15 large-scale sporting events or concerts.China Mobile uses AN and digital twins to reduce wireles

120、s network complexity16section 2:survey results where are CSPs on the AN journey?inform.tmforum.orgpage 7.The graphic above gives a more granular view of that data by looking at the most serious challenges by size of company.The complex integration challenge posed by cross-domain automation is the bi

121、ggest obstacle for every size of operator.Two key factors make it challenging:the vast number of legacy systems that most operators have throughout all network domains,and a legacy cultural mindset.TM Forums AN Project advocates network abstraction via network-as-a-service(NaaS)and the use of busine

122、ss,service and resource intents to facilitate end-to-end,cross-domain management.But this type of transformation is a huge shift for most telcos,and it takes time.In Capgeminis report,which drew heavily on TM Forums AN defjnitions and methodologies,the company asked about cross-domain skills.The sur

123、vey of 335 telco executive respondents,which was conducted late in 2023,found this to be the second-largest skills gap when it comes to deploying AN,behind AI and machine learning skills.Having multi-disciplinary or cross-domain skills means that employees must have a combination of expertise,such a

124、s network engineers with skills in data analytics or data analytics experts with network domain knowledge.“This is a tough ask,as these skills have historically been developed in specialized silos,”Capgemini says,noting that in the initial phases of the AN journey operators are likely to require mor

125、e data,AI and automation skills.Later on,monitoring,problem-solving and decision-making skills become more important,it says.Among the other challenges,budget is unsurprisingly a bigger concern for small and midsized telcos.They simply cannot justify hefty spending on new Serious AN challenges by si

126、ze of companyLarge(more than 100M subscribers)Medium(25M-100M subscribers)Small(fewer than 25M subscribers)Complex integration challenges across domainsLack of budgetThe end-to-end architecture is not defined,and there is no clear path to Level 4Key technologies required(such as AI)are not matureWe

127、do not have the right skillsLevel evaluation is difficult/we lack the methodology or tools to make a credible evaluationLack of clear business objectives and value indicatorsOur data strategy does not adequately support AN implementationAN standards are not mature enough for implementationLack of us

128、e cases that can be applied at scale57%42%27%23%29%26%22%20%19%14%44%48%30%22%37%15%19%26%26%22%40%27%27%33%13%20%33%20%13%27%Lack of top-level strategy and vision19%15%13%TM Forum,202417section 2:survey results where are CSPs on the AN journey?inform.tmforum.orgtechnology like AI or to acquire the

129、necessary skills for AN.Among large operators,its worth noting that nearly a third of respondents said they are concerned about being able to identify clear business objectives and value indicators for AN.This is somewhat surprising,given the potential for AN to cut operational costs and boost custo

130、mer experience.TM Forum has identifjed several high-value scenarios in its work on the AN Level 4 Industry Blueprint.Well look at some of them in the next section.Lessons learnedThis year we asked survey respondents an open-ended question:What is the most important lesson youve learned so far in imp

131、lementing autonomous networks?Many operators ofgered similar advice such as the need to focus on developing a data strategy and acquire skills.Ensuring buy-in from leadership and alignment of strategy and resources across the business were other suggestions ofgered by multiple respondents.The infogr

132、aphic on the next page shows many of the responses.The piece of advice that we think sums it up best is:“Start small,domain by domain,with a clear,long-term vision in mind.”We would add:Be realistic and be patient because you cant reach Level 4 overnight.In the next section we look at how organizati

133、ons are working to build consensus and defjne standards for autonomous networks.Nearly a third of respondents from large operators said they are concerned about being able to identify clear business objectives and value indicators for AN.18section 2:survey results where are CSPs on the AN journey?in

134、form.tmforum.orgLESSONS CSPS HAVE LEARNED SO FAR ABOUT IMPLEMENTING ANBuy-in from leadership is necessaryClean data is very,very important You need a data strategyYou need cross-domain alignment on vision and priorities Start small,domain by domain,with a clear long term vision in mindTraining and t

135、he right use cases are extremely importantIt takes alot of timeLevel assessments are neededThere are so many different ways to achieve a desired result,so careful planning is requiredUpskilling is critical for the futureSignificant investment is requiredCosts,costs,costs and skills,skills skillsCult

136、ural,governance and skills aspects must be addressed along with technologyIt takes a full alignment of strategy and resources between both IT and network organizationsInventorymanagementis keyUsers dont want their work to be automated,so progress is difficultTry to find big value cases instead of sm

137、all use casesDeveloping a consistent methodology for AN assessment and gap analysis with a solid link to high-value scenarios is a mandatory stepWe need partners to help us implementANBalancing automation with robust oversight is critical6G will require AI/automation not for technical reasons,but fo

138、r management19section 3:creating a blueprint for Level 4 autonomous networksinform.tmforum.orgWeve already discussed CSPs goals to deliver zero-wait,zero-touch and zero-trouble experiences for consumers and businesses through intelligent infrastructure and AI-enabled operations,which is TM Forums vi

139、sion for AN.Weve also explained what it means for a CSPs operations to be highly autonomous and the role for intent.To help CSPs implement AN,the AN Project has also developed a framework and target architecture,which consists of guiding architectural principles,a business architecture,and the techn

140、ical and formal reference architectures needed to realize AN.Using all these tools,CSPs that are already leveraging TM Forums Open Digital Architecture(ODA)can integrate autonomous capabilities into their existing frameworks,transforming ODA into composable,self-managing systems organized as autonom

141、ous domains.The AN Framework is an out-of-the-box implementation methodology to help CSPs plan and deploy AN in an effjcient and systematic way.It consists of four key elements:Key efgectiveness indicators(KEIs)to measure the benefjts of implementing AN use cases AN level evaluations to measure prog

142、ress Reference and technical architectures to guide implementation An AN roadmap to plan capability deployment.Measuring valueKey efgectiveness indicators,or KEIs,measure the business value and operations efgectiveness of autonomous networks.The AN KEI system included in the AN Framework is based on

143、 three sources:The balanced scorecard theory(a widely used KEI creation tool)CSPs practical experience with KEIs TM Forums indicator framework as described in TM Forum Metrics Framework(GB935),TM Forum Metric Defjnitions(GB988)and other standards documents.This value-driven system focuses on both ef

144、gectiveness and effjciency.TM Forum members have also created the AN KEI Framework(explained in Autonomous Networks Efgectiveness Indicators IG1256),which allows CSPs to add,delete or modify indicators to develop their own KEI system.In the Level 4 Blueprint,TM Forum has expanded the KEIs TM Forum i

145、s one of several standards development organizations and open-source groups that are defjning and evolving autonomous network concepts,standards,implementation methods and use cases to build consensus across the telecoms industry.Here we look at some key parts of the vision defjned by TM Forums AN P

146、roject,with particular focus on the groups creation of a blueprint for Level 4 high-value AN use cases.This section of the report is primarily taken from TM Forums latest AN guide,with the addition of some further analysis of our survey results.BENCHMARK20section 3:creating a blueprint for Level 4 a

147、utonomous networksinform.tmforum.orgoutlined in the AN Framework to include a three-layer AN Indicator Framework that also uses key business indicators(KBIs)and key capability indicators(KCIs)to measure the value of use cases.KBIs measure the business value of AN use cases and can be used to demonst

148、rate this value to executives.KCIs are used to set capability development objectives and guide all departments and operating companies in developing them.In our survey,we asked respondents how they measure efgectiveness from a customer experience perspective and from an operational effjciency perspe

149、ctive.The results are shown in the graphics below.TM Forum,2024Top customer experience improvement indicatorsComplaint handlingdurationQuality complaint levelsAutomatic service deliverysuccess rateService deliverytimeService delivery duration(time to market)Service fault recoverytimelinessService de

150、livery satisfactionService fault recovery durationNetwork quality satisfaction84%84%80%73%70%68%67%66%66%!67%70%83%67%76%59%TM Forum,2024Top operational efficiency improvement indicatorsPredictive fault work orders rateNetwork utilization/resource utilizationReduced operations andmaintenance costsSu

151、ccess rate of automatic serviceprovisioning 74%Fault handlingautomation rateRatio of manual workFault diagnosisaccuracy rateOperation duration/timeliness62%21section 3:creating a blueprint for Level 4 autonomous networksinform.tmforum.orgAN level evaluation sample question?TM Forum,2024Comparing app

152、les to applesThe AN Frameworks methodology for assessing AN levels,which we discussed briefmy in section 2,is defjned in the Autonomous Network Levels Evaluation Methodology(IG1252).It describes the evaluation approach and operation fmows,task evaluation criteria,scoring method and so on.An evaluati

153、on tool(GB1059)enables CSPs to measure the current level of autonomy of their operations.The fjrst step is identifying“evaluation objects”,which are determined by selecting an operational use case based on services domain,network technology domain and operational fmows.To complete the assessment,ope

154、rators use a questionnaire that evaluates cognitive activities in the network like intent handling,awareness,analysis,decision and execution(or IAADE).Each of these activities includes a set of tasks to be measured.In order to enhance its evaluation questionnaire and further standardize how AN level

155、s are assessed,the AN Project team used the professional grading standards provided by standards development organizations such as 3GPP and ETSI,with added weighting and detailed scoring criteria.The image below shows a sample question based on 3GPPs published standards in the RAN and core.TM Forum

156、conducted two phases of an assessment pilot in 2024 using the enhanced evaluation tool,with two dozen CSPs participating.AIS,Antel Uruguay,China Telecom,Elisa,Globe Telecom,IOH Indonesia,MTN Group,Orange,stc,Telecom Argentina,Telefnica and Telkomsel performed self-evaluations along with a dozen more

157、 companies that asked to remain anonymous.MTN,Orange and Telefnica evaluated the scenarios in multiple countries.Anonymized results for fault management in the RAN and core network for both phases of the pilot are shown in the graphics on the next page.Overall,the AN levels achieved remain relativel

158、y low considering the commitment of operators to reach Level 4 within the next year or two,with most scores clustered between Level 2 and Level 3.Core network stability was also assessed,achieving a higher average of Level 3.5.Overall,the AN levels achieved remain relatively low considering the comm

159、itment of operators to reach Level 4 within the next year or two,with most scores clustered between Level 2 and Level 3.22section 3:creating a blueprint for Level 4 autonomous networksinform.tmforum.orgSRILANSRILAN KAKAPHILIPPINESPHILIPPINESAsiaEuropeLATAMMEA0.001.002.003.004.00TM Forum,2024Asia 1As

160、ia 4Asia 5Asia 6Europe 1Europe 2Europe 4Europe 5Europe 6Europe 7Europe 8Europe 9Europe 10Europe 11LATAM 2LATAM 3MEA 1MEA 2MEA 3MEA 4MEA 5MEA 8MEA 7MEA 9 AN level(phase 1)AN level(phase 2)AN level evaluations for fault management in the RANSRILANSRILAN KAKAPHILIPPINESPHILIPPINESAsiaEuropeLATAMMEA0.00

161、1.002.003.004.00TM Forum,2024Asia 2Asia 3Asia 5Asia 6Europe 1Europe 2Europe 4Europe 5Europe 6Europe 7Europe 8Europe 10LATAM 2LATAM 3MEA 1MEA 2MEA 3MEA 4MEA 5MEA 6MEA 10LATAM 1 AN level(phase 1)AN level(phase 2)AN level evaluations for fault management in the core network23section 3:creating a bluepr

162、int for Level 4 autonomous networksinform.tmforum.orgLevel 4 needs AITM Forum sees Level 4 autonomous capabilities as critical to helping CSPs reach their business goals,which include:Faster time to market(one week or less for new services)and realization of NaaS Creation of digital employees to inc

163、rease the number of devices maintained per person and reduce the onsite maintenance workload by thousands of person-years Reduction of energy consumption and carbon-dioxide emissions Improvement in customer experience through minute-level service delivery and fault rectifjcation,and better SLA compl

164、iance.The blueprints reference architecture calls for full-stack AI operating across all layers of the AN Framework,with collaboration across all the layers supporting automated,closed-loop network operations.In analyzing our survey for this report,we looked at attitudes toward the use of GenAI by s

165、ize of company and found some interesting divisions.Perhaps understandably,many more large operators have introduced the technology into network operations than small ones.But more eye-catching was the revelation that none of the large CSPs questioned the trustworthiness of GenAI for use in mission-

166、critical network functions(see graphic).One probable explanation for smaller operators not embracing GenAI is the fact that often there is little or no hyperscaler presence in developing countries where many small telcos operate.The big hyperscalers like AWS,Google Cloud and Microsoft Azure ofger sc

167、alable infrastructure and advanced AI tools that lower the barriers to adoption.Without these resources,small operators may not be able to fjnd the skills they need or make the hefty investments required.In addition,small CSPs often are not able to integrate data for use by LLMs.But even in the face

168、 of these challenges,a handful of respondents from small CSPs say their companies are already using GenAI,and close to half plan to.High-value scenariosIdentifying Level 4 high-value scenarios enables CSPs to defjne the practice scope of autonomous networks,which in turn will enhance their return on

169、 investment and help them achieve business and operational improvements.Scenarios are use cases that combine an operation fmow and a network domain for example,fault management on the RAN and Level 4 scenarios are those that are likely to have maximum impact.Introduction of GenAI by size of company5

170、3%37%9%27%56%45%20%26%28%0%3%12%We plan to introduce GenAIWe have introduced GenAIin network operationsGenerative AI is notmature enough for use innetwork operations yetThe trustworthiness of GenAI is too uncertain foruse in mission critical network functionsTM Forum,2024LargeMediumSmall24section 3:

171、creating a blueprint for Level 4 autonomous networksinform.tmforum.orgThe AN Level 4 Industry Blueprint recommends that CSPs take two steps to identify high-value scenarios:Clarify the scope of key services and networks by aligning them with end-to-end network operation processes identifjed in the A

172、N Framework Thoroughly analyze the cost,the human resources required,the potential impact on cybersecurity operations and the technical feasibility to identify the most promising scenarios.The graphic opposite shows the high-value scenarios AN Project members have identifjed as priorities for the ne

173、xt two years.For example,service assurance and fault management across most domains are viewed as key,while network planning and deployment are not yet high priority.In September 2024,TM Forum launched a special type of Catalyst proof of concept called the Autonomous Network Innovation Pioneer Proje

174、ct to study AN Level 4 high-value scenarios.That teams fjrst implementations were demonstrated at Innovate Asia24.The goal for each scenario is fjrst to identify gaps between the current level of automation in CSPs networks and Level 4 autonomy.Then,the teams design solutions,implement and test them

175、,and fjnally verify the achievement of AN Level 4.In the next section we look at the progress operators have made in moving towards AN Level 4,detailing some of the success stories and best practice implementations.AN Level 4 high-value scenarios 2025-2027TM Forum,2024Identifying Level 4 high-value

176、scenarios enables CSPs to defjne the practice scope of ANs,which in turn will enhance their return on investment.25section 4:success stories AN best practice in actioninform.tmforum.orgAs shown in our survey,CSPs have predominantly focused their AN activity so far on improving network performance,wo

177、rking towards TM Forums model for zero-X(zero-wait,zero-touch,zero-trouble)and self-X,(self-serving,self-fulfjlling,self-assuring)AN capabilities.Automated orchestration,potentially combined with technologies such as digital twin simulation,can help operators increase operations and maintenance(O&M)

178、effjciency by reducing dependency on manual intervention and supporting core network agility without sacrifjcing stability.As explained in previous sections,CSPs participating in the AN Project are using TM Forums assessment methodology to gauge the AN level of organizations and processes.stc in Sau

179、di Arabia,for example,has worked with consultants to apply TM Forums six-level scale in order to evaluate several network operations processes as part as an overall digital transformation initiative.The company calls the model“critical to the success of our digital transformation”.Orange has created

180、 a taskforce to drive AN acceleration,combining top-level project ownership with domain-level expertise RAN,core,IP,cloud and so on to support evolution across each network within the group,share experiences and solutions,and create a cross-domain learning culture.Its initial task was a realistic as

181、sessment of its network operations maturity,using TM Forums Autonomous Network Levels Evaluation Methodology to establish a baseline for each of its 25 networks.TM Forums AN Framework,including the Autonomous Network Levels Evaluation Methodology,enables both network and service operational transfor

182、mation and provides a platform for innovation.But autonomous networking capabilities must be aligned with value outcomes and,as weve seen in our survey results,many service providers are concerned about being able to demonstrate the business value of implementing AN.But early adopters are making pro

183、gress.In this section we share examples of AN best practice in action,and the benefjts CSPs are seeing from implementation.BENCHMARK26section 4:success stories AN best practice in actionAccording to Asser el-Nahas,AN Practice Lead,Orange,the operator is at Level 2 on average,and higher in some areas

184、.Level 2 is defjned as“partial autonomous network”,where closed-loop operations and maintenance is enabled in some environments;Level 3 introduces alarm fjltering,as well as new root cause analysis and monitoring capabilities.Speaking at Innovate Asia 2024,Indra Mardiatna,CTO of Telkomsel,said that

185、joining TM Forums AN Manifesto through which companies agree to follow the ODA-aligned Autonomous Networks Architecture to accelerate the adoption of network automation helped the CSP to shift from AN Level 1.8 to Level 3 for some operations in 2023.Telkomsel is now investing in more analytics to pr

186、epare for AN Level 4,with the possibility of some scenarios achieving this in 2025.But,Mardiatna added,through this process Telkomsel uncovered about 6,000 discrepancies in its network which needed to be addressed to continue AN progress.Zain,which has also signed the Manifesto,is encouraged by the

187、service and operational benefjts that AN is delivering,notably around reducing network and operations complexity.In this video interview,recorded at DTW24-Ignite in Copenhagen,Zains Mohammed Jaber Al-Tarawneh,Head of IP and Transmission,and Raad Abu Dayeh,Transmission and IP Network Manager,discuss

188、the companys AN strategy with Dean Ramsay,Principal Analyst at TM Forum.They explain how increased intelligence and automation is enabling Zain to respond to changing connectivity demands and enhance customer experience.“The main advantage of autonomous networks for us as a mobile operator is reduci

189、ng network complexity and optimizing operational complexity,automating some tasks currently done manually,”says Jaber Al-Tarawneh.“The other important part is to enhance user experience against network quality of experience indicators,particularly for new 5G use cases and applications,which add new

190、challenges to our network.”He adds that in 2023 the company accelerated its autonomous network practices“and put many use cases into commercial use”.Watch the video:Focus on fault managementThe AN Project recently conducted a survey to determine which domains and processes(scenarios)to target fjrst

191、for level evaluation.Fault management that is,fault monitoring combined with detection and resolution was chosen by the majority of participants.This is not surprising.Efgective fault management ensures higher network reliability and customer satisfaction by quickly identifying and resolving issues

192、and even preventing them.It can also cut operational expenses by reducing the need for manual intervention and speeding up repair times.In a case study contributed to TM Forums latest AN Journey Guide,MTN explained its approach to overcoming fault management challenges in a multi-vendor environment.

193、With the companys consumer and enterprise businesses growing rapidly,MTN needed to provide end-to-end O&M and increase reliability of its transport network across Africa,where it has operations in more than 16 markets.Across operating companies,MTN is aggressively deploying automation and software-d

194、efjned networking(SDN)in its core network as it seeks to improve customer experience in alignment with its Ambition 2025 strategy.At the same time,MTNs operating companies had experienced more than 4,000 faults and 3,000 fjber service cuts over the course of a year,typically taking a day to demarcat

195、e faults and another three days to rectify them.To address its goals and improve fault management,MTN adopted a framework for its transport network domain which seeks to ensure that network design and planning adhere to the key performance parameters of the transport network:availability,reliability

196、 and latency.MTNs operating companies are deploying domain controllers in the IP,optical and microwave networks,together with critical,value-driven use cases.The operator opted for a hierarchical controller to enable abstraction,network slicing and end-to-end inform.tmforum.orgThe main advantage of

197、AN for us as a mobile operator is reducing network complexity.”Mohammed Jaber Al-Tarawneh,Telkomsel27section 4:success stories AN best practice in actionorchestration,with multivendor and multidomain visibility of its IP,optical and microwave networks.This meant working with three difgerent vendors

198、to implement an autonomous network that delivers self-monitoring,self-healing and self-optimization capabilities.For example,in the past,faults were handled by IP and optical network teams independently in difgerent systems.Now,multi-layer faults are diagnosed,demarcated and located automatically in

199、 one system.MTN South Africa has deployed the solution with promising results,including a reduction in fault demarcation time from one day to just fjve minutes.inform.tmforum.orgAs weve reported in our research throughout 2024 CSPs,and in particular large operators,are embracing GenAI across their b

200、usinesses.While many of the earliest uses have been in the customer contact center,increasingly operators are applying the technology in network operations and maintenance(O&M).China Mobiles use of GenAI to increase automation in O&M,for example,won a TM Forum Excellence Award in 2024.China Mobile h

201、as a total of 190 million subscribers in its Guangdong,Beijing and Fujian provinces,where the company has been working to develop a GenAI solution to help with work order optimization,fjeld testing and performance optimization of 4G and 5G networks.By leveraging GenAI and LLMs,the operator is achiev

202、ing an accurate understanding of users intentions,which are then translated into intelligent orchestration of network operations.As a result,China Mobile is saving about$7 million a year and realizing an 80%reduction in knowledge acquisition time,as well as a 72%increase in data analysis effjciency,

203、all of which is bringing the company closer to its goal of achieving a highly autonomous network by 2025.China Mobile was looking to address three main challenges with the GenAI solution it has developed with its vendor partner Nokia:Cut the time it takes to query knowledge bases when trying to reso

204、lve faults.A large amount of knowledge about telecoms network operations exists in various manuals,case libraries and reports,making it diffjcult to run queries.Indeed,it was taking China Mobiles engineers an average of fjve to ten minutes to fjnd information,with some complex cases taking more than

205、 an hour.Reduce the cost of daily data analysis.In O&M there are many temporary and personalized data analysis requirements,which are labor-intensive and must be completed by highly paid experts.In the Guangdong province alone,there are more than 3,000 tables in China Mobiles big data platform.With

206、more than 400 engineers conducting data analysis tasks every day,the cost of data analysis was about$6.8 million a year.Improve DevOps development.Managing diverse and dynamic business requirements in network operations demands signifjcant time from software engineers,often spanning several days or

207、even weeks.This results in low effjciency and high costs,with around 200 software engineers working in the three China Mobile provinces.China Mobiles GenAI solution addresses these issues by providing a natural language capability for intelligent Q&A,automation of data analysis and DevOps code gener

208、ation.It is being rolled out in three phases.The fjrst,called Intelligent Assistant,was completed in 2023 and helped the operator achieve Level 3 AN.The results detailed in this case study were achieved in this phase.The second phase,called Expert,is happening now and will be complete in 2025.During

209、 this phase the team is building on its LLM to automatically create network confjguration instructions and optimization directives in complex operations scenarios and implement solution recommendations.China Mobile hopes to reach Level 4 AN as a result.The fjnal Master phase in 2025 and beyond will

210、support more autonomous end-to-end operational processes with the ultimate aim of achieving Level 5.Read the full case study:China Mobile uses GenAI to help move up the levels28section 4:success stories AN best practice in actionValue creationIn a conversation at Innovate Asia,Christian Maitre,CEO o

211、f Caspera-Lab,and Olta Vangjeli,Program Director for Cloud Native IT&Networks at TM Forum,discuss the Forums AN standards work and how value creation is,as Maitre puts it,“inside the AN philosophy”.He adds:“Doing things like zero-touch,fmexibility,scalability,will create new types of services for B2

212、B,B2C,B2X combined with operations for many domains:smart cities,smart fjnance,smart energy AN at the core is really the engine of this.”In the video they also consider the need for clean data,and developing the skills required as operators move along the AN journey and introduce higher levels of AI

213、 into networks.Maitre,who previously led smart city strategy at Orange,was the leader of an important multi-phase Catalyst project called AN hyperloops,which has fed learnings into the AN Project (see page 32).Measuring the value of automation and using it to drive new value,is a key focus for China

214、 Mobile,Dr.Lingli Deng,Researcher Director at the CSP,told TM Forum Inform in 2023.“In addition to effjciencies from the internal operation perspective,we are also trying to set business value metrics,so that we can actually expose the automation and autonomy capabilities to our customers,or use tho

215、se capabilities to fjnd new customers,”she said.Sustainability,in the form of energy savings,is another important part of the AN value proposition,and an area yielding positive,measurable results.For instance,the three major Chinese CSPs are already reporting signifjcant benefjts in this regard(see

216、table on p.29 for more on their progress and plans):China Mobile savings of 4 billion kilowatt hours(KWh)of electricity in 2023 China Telecom annual savings of 1 billion KWh of electricity China Unicom monthly savings of 40 million KWh of electricity.In a research note for TM Forum,Ehsan Dadrasnia,V

217、P,Global Telco Industry:Cloud Network Transformation&Partnership Strategy,Capgemini,echoed the value of AN to sustainability targets.“Those CSPs that are moving fast on autonomous networks are also realizing cost and sustainability benefjts.In just the past two years,telcos have on average achieved

218、a 20%improvement in operational effjciency and an 18%reduction in opex through autonomous networks,”he said.“Those who transition to a higher-level autonomous network can expect energy reduction of somewhere between 7.5%-15%in their networks carbon emissions.As generative AI grows in prominence,were

219、 going to see these results improve much quicker.”As we have seen throughout this report,many CSPs are well underway on their AN journey,with a handful of operators already adopting TM Forums AN Level 4 high-value scenarios,specifying their planned evolution paths from 2025 to 2027.The table on the

220、next pages shows the domains and processes some of these CSPs intend to target as well as highlighting measurable points of progress.In the next section we showcase some award-winning Catalysts focused on autonomous networks and AI.inform.tmforum.orgIn just the past two years,telcos have on average

221、achieved a 20%improvement in operational effjciency.through autonomous networks.”Ehsan Dadrasnia,Capgemini29section 4:success stories AN best practice in actioninform.tmforum.orgCSPs AN strategies and progressCSPTM Forums ODA,AN framework and AN level taxonomy are the foundation of AISs Cognitive Te

222、chco strategyAN strategyLevel 4 high-value focusProgress and plans Core network fault management Wireless network optimization:customer experience driven Optical and transport service assuranceIs participating in TM Forums AN level assessment pilot;achieved L3 in the two targeted processes(RAN and c

223、ore fault management);results include:CX optimized by 10%Traffic loss reduced by 90%in extreme scenarios Network efficiency improved by+20%Is extending AN practices to network optimization,configuration,planning,change and other processes.China Mobile was the first CSP to set a goal to reach L4 on a

224、 large scale in its production network;technology function is centralized,and the company is using end-to-end service operation layer across domains to drive overall development of technical AN capabilities Service delivery Network change Energy-efficiency optimization Service-quality optimization T

225、roubleshooting Complaint handlingAchieved overall AN L3.2 in 2023;results include:Significant improvement in customer satisfaction 80%reduction in the number of major network faults 20%increase in average number of sites maintained per person Savings of 3,200 person-years of human labor and 4 billio

226、n Kwh of electricity Contribution to 70 AN technical standards and 15 whitepapers Has launched a telecoms LLM with 13.9 billion parameters,trialing applications like data self-service and AI chatbots.China Telecom is guided by 3-ultra Vision,considers autonomous cloud-network operations as an import

227、ant stepping stone for the transformation toward network+cloud 5G core network:complaint handling IP network:network changes(intelligent online simulation)Innovative practice of network LLM in network maintenance scenarioIs participating in AN level assessment pilot.Achieved overall AN L3+in 2023;re

228、sults include:Return on equity up from 6.41%in 2022 to 6.93%in 2023 Savings of 1 billion kilowatt-hours of electricity annuallyIn 2024 aims to achieve L4 in 40%of high-value scenarios and will promote large-scale application of its LLM in the network,creating thousands of digital employees.China Uni

229、com aims to implement autonomous systems across all stages of planning,development,maintenance,optimization and operations to enable digital operations,intelligent scenarios and agile services.Optical private lines service assurance Intelligent planning and construction Intelligent energy savings in

230、 base stations Closed-loop control of optical cable risksIs performing ANL evaluations for 5 networks across 8 provinces focusing on high-value scenarios using GenAI;results include:1.5 million customer complaint diagnoses per month with an 80%one-time problem resolution rate Intelligent energy-savi

231、ng systems cover 98%of base stations,saving over 40 million kilowatt-hours monthly Open platform aggregates 2,800 capabilities,utilized over 38 billion times per monthWill focus on additional high-value L4 AN scenarios in the next two years.30section 4:success stories AN best practice in actioninfor

232、m.tmforum.orgCSPs AN strategies and progressCSPChunghwa Telecom has been working on closed-loop automation for specific network scenarios since 2019.AN strategyLevel 4 high-value focusProgress and plansSince 2023,the company has expanded AN operations management to enhance customer experience and re

233、duce operational costs;results include:Conducted ANL assessments for fixed and mobile broadband services,achieving L3 autonomy Developed an automated fixed broadband service rate upgrade mechanism,benefiting 1.16 million usersPlans to apply AN to mobile broadband and IP transport networks,aiming to

234、achieve L4 by 2025.Fault management in RAN and IP transport domains Energy-efficiency optimization in the RAN MTN is implementing AN as part of its Ambition 2025 Strategy,which includes a technology framework called PACE(Platform,Agile,Connectivity and Experience).All MTN operating companies aligned

235、 with the companys AN strategy and target architecture,which is based on TM Forums AN framework.Cross-domain network fault management IP and optical network autonomous O&MParticipated in the AN level assessment pilot and will continue to align to the assessment for RAN,core,IP and change management.

236、In 2023,MTN released its AN Blueprint 2.0,establishing ANL standards and developing a comprehensive use case library.In 2024,the company plans to design Version 3 of the blueprint,identifying high-value scenarios to finalize assessment criteria,effectiveness targets,processes and architecture for AN

237、 L4.Singtel initiated its AN program in 2023,fostering group-level collaboration to benchmark,share best practices and apply them across the group.Designated Singtel Group operator AIS as an AN coach to standardize the four key elements of the AN Framework and support operating companies in L3 and L

238、3+use-case deployment.In 2024,will create its AN L4 Blueprint to transition from automation to autonomy.Fault management and customer-complaint management RAN energy saving and network optimizationHKT aims to bring the right digital solutions to society to accelerate growth and create a sustainable

239、future for all.Wireless network optimization focusing on energy usage according to data traffic using AI technology Optical and transport service to minimize energy consumption and improve energy efficiency through AI technologyGlobes autonomous networks strategy focuses on AN Level 4 attainment.Cor

240、e network fault managementIs participating in TM Forums AN level assessment pilot.Globe is working on automation in the core network to ensure uninterrupted service and realize some scenarios of AN Level 4.It is continuing building a network that can self-heal,predict and prevent outages,and adapt t

241、o changing conditions in real time based on digital twins technology.Is participating in TM Forums AN level assessment pilot.Focus in 2023 was on reducing 20%of HKTs total energy consumption(compared with 2018 baseline),and HKT plans to further cut energy consumption this year and in the future usin

242、g innovative methods over its AN.31section 4:success stories AN best practice in actioninform.tmforum.orgCSPs AN strategies and progressCSPTelefnica launched a program called Autonomous Network Journey(ANJ)in 2021 to increase autonomy and transform its architecture,management of data,and organizatio

243、nal structure to support AN;ANJ is being applied in all operating companies in Europe and Latin America,but deployment progress varies among OpCos.AN strategyLevel 4 high-value focusProgress and plansTelkomsel adopted AN as part of its digital innovations,aiming to reach L4 by 2025.Ensuring high ser

244、vice quality during peak events.Transport network(fronthaul optical link):fault predictive maintenance Wireless network:fault management(cross-domain)Wireless network:optimizationIs participating in AN Level assessment pilot.In 2023,Telkomsels cross-domain experts,guided by the Networks Quality Digi

245、talization Department,made significant progress towards L3.Notably,the self-assurance use case for massive events was recognized at TM Forum DTW2023-Ignite and deployed across 37 sites in Indonesia.Is participating in the AN level assessment pilot.Anticipates several key improvements in achieving AN

246、 Level 4 including:AI-powered operations assurance,where closed-loop automation will prevent service disruptions.Real-time optimization and decision-making,where automated systems will manage network configurations,including dynamic features like network slicing and NaaS capabilities.Efficient plann

247、ing and deployment,where real-time inventory management,scenario-based planning(e.g.,what-if analysis)and AI-driven optimization will streamline deployment processes.TM Forum,2024Is focusing on getting to AN Level 4 for multiple processes(O&M,optimization/configuration,testing,deployment and plannin

248、g)in multiple domains(radio access,transport,IP,core,IT cloud,and fixed access networks).The company is placing high emphasis on O&M in all domains and on RAN optimization,IP deployment and transport planning.Telecom Argentina drives its strategy of developing ANs and contributing to TM Forums best

249、practices through its Autonomous Network Transformation Office(ANTO).Wi-Fi optimization Core network:network optimizationIs participating in AN level assessment pilot.Has committed to achieving its goals through a two-stage approach,with the first focusing primarily on developing technical use cases

250、 and the second aiming for more balance between technical and business domains.stc has a goal to achieve ANL 4 in 2025 in some scenariosIs participating in TM Forums AN level assessment pilot.2024 focus was on automatic network change and AIOps capabilities including:Digital twin-based cloud-network

251、 visualization Intelligent signaling storm simulation and emergency assistance Pipeline-based network change and deployment automationResults included:Fault management-CNF and cloud health status unified visualization,moving from 5+portals to 1 and cutting fault-demarcation time from 2 hours to 15 m

252、inutes Network changes and deployment:ultra-security(automatic execution by unified pipeline reducing 80%of manual breakpoints);high-efficiency(deploying 200+test cases in minutes);and full scenarios for both telco-cloud and core networkIn 2025,stc plans to introduce foundational models for alarm an

253、alysis,an intelligent copilot and a complaint handling expert.Core network:fault management,network changes,design and deployment32section 5:award-winning Catalystsinform.tmforum.orgAn award-winning TM Forum Catalyst called Autonomous networks hyperloops has revolutionized the concept of how autonom

254、ous networks should be designed and function and in the process is helping telcos understand how to deliver a signifjcantly more dynamic customer experience that rivals hyperscale cloud providers.The hyperloops project,which began in 2019 and is now complete after fjve phases,focused on defjning the

255、 global architecture needed to support a wide range of vertical use cases and the role for new technologies like metaverse,digital twins and AI.In various phases,the proof of concept has been championed by CSPs including Chunghwa Telecom,NTT,Orange,TIM and Verizon.It won a 2024 Catalyst award in the

256、 Outstanding Catalyst Beyond Telco category,and learnings from the project have been incorporated into the AN Projects work.Indeed,the familiar“Intent-based,closed-loop management”graphic used in many TM Forum publications and our own research reports(see page 9)originated in the hyperloops Catalyst

257、.The earliest phases of the project looked at how to enable smart vertical use cases such as distance learning/education and smart farming,stadiums and manufacturing.In later phases the team showed how to use intent,closed loops,5G network slicing,AI and digital twins,to deliver pop-up mobile networ

258、ks in the event of a natural disaster.Most recently it demonstrated how telcos can provide a“virtual command center”,or VCC,as a service.“This idea of a virtual command and control center is multi-industry,”explains TM Forums In this section we look at two award-winning proofs of concept focusing on

259、 autonomous networks and AI.They are followed by a list of other Catalysts that won awards in 2024,with further information and links to help you fjnd out more.BENCHMARK33section 5:verticals delivering growth opportunities for someChief Architect Dave Milham.“What I like about it is that its not abo

260、ut ordering telco services.Its about saying:What is it that the end user actually wants?not the customer,the end user and then setting up an environment closer to what the end user really wants.”In the VCC setup,an end user in an enterprises operations center is presented with a dashboard that provi

261、des a visual representation of all the assets the company has(including those owned and leased).This provides what Milham calls“situational awareness”.“In principle,you could have an icon that starts fmashing amber and says you havent got enough capacity on this link;its reached over 60%capacity.And

262、 some AI behind it can say,It looks like youre going to run out and have severe capacity problems here,”he explains.Then,the user is given options of either increasing the bandwidth on that link or rerouting existing traffjc to free up capacity.“But did an order or a trouble ticket appear anywhere i

263、n any of this?”Milham asks.“No.The tasks that are being presented are in the language of the person whos running an operations center.”Enabling this kind of dynamic ordering,provisioning and management of communications services requires completely rethinking how networks are architected,and that is

264、 one of the most important takeaways from the project,says Caspera-Labs Christian Maitre,leader of the hyperloops Catalyst.As the project progressed,it became clear that the way CSPs and their suppliers were approaching AN would not scale,says Maitre.The network must become fully virtualized,softwar

265、e-defjned and AI-driven with a“3D”architecture.“We call it the onion.You have the core and difgerent layers around it in three dimensions,”he explains.“We had to create a new vision for the architecture.Its not fmat or hierarchical,but 3D.”Perhaps the most important lesson learned from the hyperloop

266、s Catalyst is that“AN is still to be defjned,”says Maitre.“There is a lot to still be imagined.”Read the teams whitepaper and watch the video to learn more:GenAI powers computing network solutionMany TM Forum Catalysts have been looking specifjcally at how GenAI can be applied in autonomous networks

267、.As GenAI and LLMs become integral to many telecoms processes and services,the demand for computing power will escalate exponentially.This makes effjcient allocation of resources critical.To address this challenge,China Unicom,China Mobile,China Telecom and Telecom Malaysia have been collaborating i

268、n a TM Forum Catalyst proof of concept to streamline the use of computing power and network resources.This requires customers original intentions to be aligned with the capabilities and services of the computing network.The GenAI empowers computing force network Catalyst won an award in 2024 in the

269、Outstanding Catalyst Innovative&Futuristic category for its creation of an AI-native computing network operation system to help CSPs schedule computing and network resources across a cloud-edge network.The solution uses advanced LLMs,multi-agent collaboration and automated code generation to underst

270、and,ask questions about and supplement the customers business scenarios and needs.The idea is to autonomously create a customized computing and networking integration proposal that precisely meets the customers requirements.Then,the customer can refjne it using an intuitive chatbot,which replaces th

271、e conventional product catalog.The computing network solution integrates diverse resources from hyperscale cloud platforms,telco clouds and third-party services to deliver zero-touch customer experience and zero-wait and self-provisioning capabilities.Using this approach,CSPs can develop and support

272、 a unique ecosystem and revenue-sharing business model.As a result of the Catalyst,a new workstream called Autonomous Computing Network was formed in the AN Project,says Zie Zhang,Principal Researcher,Digital Transformation Strategy Consultant,at China Unicom.The team plans to conduct Catalysts annu

273、ally and is working on a guide to explain its learnings so far.A second phase of the computing force Catalyst,which will be inform.tmforum.orgAN is still to be defjned.There is a lot to still be imagined.”Christian Maitre,Caspera-Lab34section 5:verticals delivering growth opportunities for somedemon

274、strated at DTW25-Ignite in June,is looking at two scenarios:How to create a low-energy,high-effjciency computing infrastructure using AI How to use new technologies such as cross-domain orchestration and heterogeneous resource scheduling to give customers a one-stop intelligent solution that uses pu

275、blic cloud,network services,platform services and AI.The group also plans to contribute an API to TM Forums Operate APIs,says Zhang.Watch the video to fjnd out more:In the next section we look at standards work being carried out to drive telco interoperability,and fjnd out which organizations our su

276、rvey respondents are working with.inform.tmforum.orgHere are some other 2024 Catalyst award winners that have focused on the use of AI and GenAI in autonomous networks for you to explore:Accelerated service-centric autonomous network operations provides a pragmatic framework for operation transforma

277、tion which integrates processes,platforms,people(organizations)and technologyAI for customer perception network optimization anticipates customer pain points through AI-led digital twins and explores how CSPs can move to a perception-centric approach to wireless network optimizationAutonomous networ

278、ks hyperloops Phase V virtual command center as a service combines AI with autonomous networks and digital twins in a single interface to provide a comprehensive real-time view of operations,assets and infrastructure(see also p.32)GenAI empowers computing force network uses GenAI to help CSPs schedu

279、le computing and network resources across a cloud-edge network(see also p.33)GenAI for AN looks at how to use GenAI to automate network maintenance GenAI-powered toolkit for network&service management Phase II aims to create a knowledge graph that maps relationships between network elements,topology

280、,faults and inventory data to enable predictive fault correlation,proactive decision-making and closed-loop automationHarnessing GenAIs business value for operations transformation sets out to transform O&M by boosting effjciency and network quality using AIRealtime intelligent customer experiences

281、shows how AI and automation,combined with telco data,can enable personalized experiences and boost digital revenue to deliver telco growthSustainability telco by green network evolution aims for net-zero carbon emissions in difgerent geographical conditions using a digital twin and AI platform.Award

282、-winning Catalysts leverage AI and GenAI35section 6:collaboration on standards is key to AN interoperabilityinform.tmforum.orgAchieving cross-domain interoperability in autonomous networks is a challenge that requires collaboration among standards development organizations(SDOs),open-source groups a

283、nd other organizations.In our survey we asked respondents which groups theyre participating in,allowing them to pick as many as necessary.The table on the next page shows the results for a non-exhaustive list of organizations doing work that could impact ANs.BENCHMARKIt is not surprising that half o

284、r more are working in 3GPP,GSMA and TM Forum projects.Most companies surveyed are TM Forum members,and mobile networks are an early target for AN for reasons outlined throughout this report primarily a desire to reduce energy consumption and improve fault management in those domains,especially the R

285、AN.Weve discussed standards collaboration in previous reports and articles published on Inform.It can be a thorny issue because politics are sometimes involved,with people understandably wanting to advance and protect their work within specifjc groups.For telcos and their suppliers,joining any one o

286、rganization usually involves spending money on membership and the commitment of an employees (or multiple employees)time.Sometimes a single stafg member has responsibility for a companys participation in more than one group.We should simplifyBut CSPs increasingly are calling for more collaboration a

287、nd alignment across groups as they seek to automate ordering,provisioning,orchestration,assurance and billing of services end to end across multiple domains,which oftentimes include their partners networks.In a recent interview with Inform,Laurent Leboucher,Group CTO at Orange,said standards bodies

288、and other organizations need to recognize and co-ordinate overlapping work.“I think we should simplify at some point.We dont need everybody to do the same,”he says,pointing to duplication of efgort between TM Forum and ETSI on zero-touch and network automation as an example.“Within Orange when we wo

289、rk in forums or standards we try to keep a very consistent position in the difgerent bodies.But its hard because there are too many bodies,”Leboucher says.To unlock the full potential of AI-driven networks,collaboration specifjcally on standardization of network APIs and defjning common information

290、and data models is needed.A signifjcant amount of collaboration is,indeed,happening on network APIs.For more than a year,TM Forum members have been developing 36section 6:collaboration on standards is key to AN interoperabilityinform.tmforum.orgTM Forum,2024SDOs/open-source groups working on telco i

291、nteroperabilityOrganizationFocusRespondents participatingTMConsortium of 7 SDOs from around the world that is creating protocols for next-generation mobile networks;responsible for developing 5G non-standalone(NSA)and standalone(SA)specs.49%Linux Foundation project formed by the merger of two other

292、open-source projects:the Cloud iNfrastructure Telco Taskforce and OPNFV;goal is to deliver a common model,standardized reference infrastructure specs,and frameworks for VNFs/CNFs.1%Develops next-generation IP network specs focusing on architecture,management and digital CPE to help CSPs deploy and m

293、anage hybrid networks from a single IP-centric platform.19%Joint efgort between GSMA and the Linux Foundation to provide a single,centralized repository for end-user APIs,including those developed by TM Forum;focuses on how to expose mobile network capabilities via APIs for external consumption by d

294、evelopers outside telecoms.20%The China Communications Standards Association(CCSA)was established in 2002 by the Chinese Ministry of Information Industry.It is involved in international standards development as a partner of 3GPP and participant in the ITU.TM Forum and CCSA are collaborating on AN st

295、andards.3%ETSIs standardization activities focus on Europe.Examples include Zero-touch Network&Service Management(ETSI-ZSM),Experiential Networked Intelligence(ETSI-ENI)and the ETSI ISG NFV working group,which develops requirements and architecture for virtualization of network functions.21%GSMAs Op

296、en Gateway initiative aims to drive adoption of APIs that expose network capabilities without the need for developers to understand network technology or terminology;work includes mapping between CAMARA APIs and technical APIs from bodies like 3GPP,IETF,ETSI NFV and TM Forum.49%Develops internet sta

297、ndards;CSPs use IETFs NETCONF protocol to install and manage the confjguration of network devices,while the YANG data modelling language describes the confjguration changes.23%Standardizes Carrier Ethernet services so that enterprise LANs can extend across WANs;added extensions to the Open APIs to c

298、reate its LSO APIs and now ofgers an API blending tool.21%Joint Linux Foundation and Google Cloud project focusing on cloud-and intent-based automation;goal is to simplify deployment and management of multi-vendor cloud infrastructure and network functions at scale.2%Association of mobile network op

299、erators focusing on network management,disaggregation,sustainability,and 5G and 6G use cases.3%Founded by AT&T,China Mobile,Deutsche Telekom,NTT and Orange to develop fully interoperable mobile networks by focusing on interworking between difgerent components of the RAN;is developing standards for o

300、pen RAN service management and orchestration that interwork with TM Forum Open APIs.17%Linux Foundation Europe project that aims to create a new,open-source,production-grade telco cloud stack;aims to develop a common cloud software framework and reference implementation.3%Engineering-focused initiat

301、ive led by Meta to bring CSPs,infrastructure providers,systems integrators and other tech companies together to collaborate on development of new technologies.18%Develops the Open Digital Architecture which provides programmable access to capabilities in OSS/BSS;more than 70 Open APIs are available

302、and widely implemented.85%37section 6:collaboration on standards is key to AN interoperabilityinform.tmforum.orgOpen APIs to complement network APIs from The CAMARA project,with a goal of providing a standardized way for mobile operators to make the interfaces commercially available to aggregators a

303、nd developers.CAMARA is an open-source project within Linux Foundation Networking that works in close collaboration with the GSMA Operator Platform Group(OPG).OPG brings together operators,developers and platform providers to transform multi-access edge computing(MEC)services.Its Open Gateway initia

304、tive aims to drive universal adoption of APIs designed to expose network capabilities without the need for end-user developers to understand telecoms network technology or terminology.While CAMARA is developing the APIs to expose network capabilities such as quality on demand(QoD)and device location

305、 to developers,APIs for IT operations are needed to automate service delivery,track usage and provide billing capabilities,so that third parties can buy the right to use CAMARA APIs.TM Forum calls these Operate APIs.TM Forum Open APIs also support the internal processes for delivering network servic

306、es via CAMARA APIs for example,the NaaS Open API Component Suite abstracts technology-specifjc and vendor-specifjc network management APIs to a single API into the service layer above,so that confjguring and managing the network can be done using a generic NaaS API rather than lots of difgerent APIs

307、 performing the same function for difgerent technology domains below.These interfaces are evolving to intent-based operation,where the outcome or characteristics of the service are requested via the API(high-quality video,for example)rather than the specifjc instructions for the technical resources

308、delivering the service.Orange has been active in network API and Operate API development.In September,the company,along with 11 other telcos,announced a global joint venture with Ericsson to sell network APIs.Standardizing the right APIs requires“working backwards from concrete use cases”,Leboucher

309、says.“That means that we cannot just do it between telcos in a closed room.It has to be together with AI players like Google,like Meta,like OpenAI,Mistral and so on they have to be in the room,”he explains.“Maybe the fjrst defjnition will be a kind of quality observation or quality-on-demand API;may

310、be it will be difgerent from the one that we thought about at the beginning,butits really an industry move,”he says.In the fjnal section we recap the key fjndings from our research and give some recommendations for implementing AN best practice.38section 7:key fjndings and recommendationsinform.tmfo

311、rum.orgThroughout this report we have focused on the opportunities,challenges and best practices for CSPs as they progress their autonomous network strategies.This section recaps key fjndings from our research,including from our survey based on 111 respondents from 82 companies,and makes suggestions

312、 for AN planning for the future,based on our observations of industry best practice.BENCHMARKKey fjndings AN momentum is growing rapidly.More than 60 CSPs have signed TM Forums AN Manifesto,pledging to work towards Level 4 autonomy,and to date 24 CSPs have carried out internal AN level assessments.S

313、ome,including Chinas three major operators,plus MTN,Orange and Telecom Argentina,have publicly stated they are adopting AN Level 4“high-value scenarios”aiming to reach that stage for operational fmows in some domains from 2025 to 2027.Across the industry,more than 120 AN standards and research proje

314、cts have been initiated.Autonomous networks use AI,machine learning and advanced analytics to automate operations,including the ability to detect and diagnose network performance issues,reroute network traffjc in case of network failure,and predict/prevent potential problems.This is a signifjcant fo

315、cus of current AN activity,rated as a high priority by 83%of our survey respondents.Customer experience is a critical context for measuring AN value.According to our survey 84%of respondents rate network quality satisfaction as the most important CX metric for AN capabilities together with service f

316、ault recovery duration(84%),followed closely by service delivery satisfaction(80%).About two-thirds of survey respondents have implemented or plan to deploy GenAI within their network operations.Nearly 75%expect GenAI to be utilized to support network maintenance use cases,such as automated troubles

317、hooting and risk prediction,while 65%are looking to use GenAI for network operations applications such as intelligent customer service.Just over 60%plan to use it for network optimization purposes such as risk identifjcation and automated verifjcation of optimization.AI and autonomous networking can

318、 drive operational effjciency,revenue growth and CX enhancement while improving compliance,sustainability and risk management.But despite the value of AI and AN becoming better understood,not all CSPs are set up to take advantage.Legacy systems,operational complexity and organizational structures ca

319、n all prevent their full value being realized.Integration complexity was rated as a serious challenge by 53%of our survey respondents(and somewhat of a challenge by another 43%),followed by lack of budget(43%).A third of respondents said the fact that an end-to-end architecture is not clearly defjne

320、d is a serious challenge(a further 52%said it is somewhat of a challgenge),and that there is no clear pathway to AN Level 4.39section 7:key fjndings and recommendations Transitioning from automation to true autonomy is a signifjcant business and technology challenge.TM Forum oversees teams working c

321、ollaboratively to create and evolve autonomous network and AIOps frameworks,maturity models,APIs and tools to help CSPs scale solutions.In November 2024,the Forum published its blueprint framework for achieving AN Level 4,to support CSPs with this challenge.To unlock the full potential of AI-driven

322、networks,there is a need for standardized network APIs,common data models and aligning automation efgorts across difgerent standards development organizations(SDOs).This must happen in both AI-for-network and network-for-AI scenarios.CSPs need to collaborate with each other and with AI players such

323、as hyperscalers to understand how to unleash network capabilities through APIs.Recommendations A clear strategic vision is vital.On the business side,leadership,strategy and an AI-centric culture are all important.On the technology side,end-to-end model management for AI is critical,and a target arc

324、hitecture for autonomous networks is necessary.As we have seen,many CSPs are using TM Forums AN standards to defjne their own target network architectures as they aim for AN Level 4 and shift from static,rules-based network automation to AI-enabled,dynamic orchestration across multiple network domai

325、ns.Build a long-term AN business case.While the business case is typically founded on cost-savings,and performance effjciency use cases which address immediate concerns around network performance and customer retention,it should also evolve to include AN enablement of intent-based service opportunit

326、ies such as NaaS and network slicing.It is vital that AN leverages intent,so that connectivity can be tailored to business outcomes.Benchmarking AN progress is required.Many CSPs use their own standard KPIs for measuring progress for example,how well automation efgorts are improving metrics like the

327、 duration of a service disruption.Other CSPs are measuring automation success as part of an overall digital transformation project,usually using metrics such as percentage of trouble reports that are automatically diagnosed.But it is necessary for the telecoms industry to agree on how to measure suc

328、cessful AN deployment so that operators can benchmark their progress against their peers and support end-to-end use cases.In addition,success fuels continued investment,so being able to demonstrate positive impact on customer experience is a fundamental aspect of the AN business case.Use AI for netw

329、orks to create networks for AI.AI is critical for the evolution of AN,but AN is also critical to support future AI-driven demands on connectivity,to ensure that networks are AI-ready.TM Forums AN Level 4 Blueprint emphasizes the fundamental reliance of ANs on AI being embedded into all domain design

330、,operation and processes,such as closed-loop orchestration,to enable intent-based services.It provides detailed guidance on level assessment mechanisms and criteria for achieving AN Level 4.High-value scenarios help CSPs focus on AN impact.Moving to AN Level 4 is compelling as it helps CSPs unlock n

331、ew,high-value scenarios that combine an operation fmow and a network domain(for example,fault management on the RAN)and are most likely to have maximum impact on the operations and/or business of a CSP.Use case selection for AN Level 4 high-value scenarios is an iterative process.Accurately designin

332、g the Level 4 AN target state for high-value scenarios enables CSPs and suppliers to achieve technical breakthroughs,drive application innovation and accelerate adoption of best practices.Establishing a data culture is critical.A strong data foundation,based on data quality and multi-domain data sha

333、ring,is fundamental to AN success.Lack of a data strategy was rated as one of the biggest challenges for AN implementation in our survey(considered serious or somewhat of a challenge by three-quarters of respondents).As part of the Modern Data Architecture project,TM Forum members are collaborating on an industry-specifjc data reference architecture,incorporating emerging AI-enabled business model