The Age of Telecom Network Automation

Home , Automation

The Age of Telecom Network Automation Automating Engineering and Network Operation RAN Processes Telecom Engineering Centre of Excellence (TEE) Introduction 3 Operating networks in 5G: 5 A brave new world Leap ahead through Automation 6 Key components of RAN automation 7 Is automation only about technology? 11

Which RAN use cases are we seeing 15 on the market? How are we succeeding 18 with our customers? Conclusions 20 An ecosystem of automation 21 specialist partners Who we are 22 Glossary 23 References 24 Contacts 25 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Introduction

We are in the dawn of a new era in what Ultimately, the goal of communication We are in the dawn of regards telecommunications networks' services providers (CSP) is to become a new reality in Telecom reality. This reality extends to multiple digital services providers (DSP), domains: network services provisioning, exposing their network into on-demand networks, not another network services architecture, and consumable services: flexible, fast to incremental step network engineering and operations, while provision and manage, with tailored being driven by disparate requirements quality of service and service level or evolution. ranging from service flexibility and agreements that need to cope with the increased service quality, to the need to promise of the three main pillars of 5G lower Operational Expenditure (OPEX) and (enhanced mobile broadband, ultra Capital Expenditure (CAPEX). reliable low latency and massive machine type communications). From a foundational perspective, in In such a world, automation is not an a world moving to 5G, this is not an option but mandatory, as it will need to incremental step. The new standalone dynamically manage and orchestrate architecture (SA) and distributed vRAN all the above services at such a volume (Virtual Radio Access Network) that and complexity whilst at the same time 5G will bring is being defined as cloud coordinating a multitude of data and native, and not merely Network Function technical domains, that it would never Virtualization (NFV) oriented, meaning be feasible based only on manual human a huge disruption compared to today’s operations. reality. Furthermore, the new paradigm will In fact, as of today, the network roles are be about managing end-to-end (E2E) either provided by monolithic solutions, customer service and not about managing or by software components running on networks, meaning that orchestration and top of virtual machines, but rare are the automation need to be service focused vendors that have embraced the cloud and not merely network focused. This native Virtualized Network Functions requires quite a disruptive approach to (VNF) challenge. aspects such as organizational structure, data governance, company culture, and technical skills, as organizations move out of the traditional NetOps model.

3 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

E2E service orchestration is still far from yet delivering fully on their promise, In this new era, mass adoption, but which building blocks with the advent of 5G a cloud native of automation in 5G are already appearing approach using automation is automation is not an in the market? unavoidable. option but mandatory. And which are the lessons learned This paper provides a view on the above Defining the automation we are already seeing as CSP start to topics, with a focus on Radio Access building blocks, really automate their engineering and Network, as well as presenting Deloitte’s operations activities? Telecom Engineering Centre of Excellence learning from the early Even more relevant: How to define the EMEA (Tee CoE) proposed methodology adopters and having journey to automation as we move to 5G? to address the challenges that arise from this new Telecom Era. Following a solid approach is While both software defined network this document, our team will publish paramount. (SDN) and NFV domains have been subsequent points of view for different maturing for some time and are not automation domains.

Pedro Tavares Hugo Pinto Partner Senior Manager, TEE Head of TEE Network Automation Knowledge Lake Head

4 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Operating networks in 5G: A brave new world

5G is the first mobile telecommunications 5G Standalone architecture (SA) is The above topics, coupled with the fact generation that was thought from its natively thought to be as flexible and that 4G, 3G and 2G still subsist and will still inception to be based on virtualization, automated as possible, not being another need to coexist with 5G for a significant implementing concepts like network overlaid radio technology with some core amount of time together with other slicing, relying on open features such as network adaptations. 5G is therefore factors as illustrated in Picture 1, lead to API (Application Programming Interface), the mobile generation that will support a simple conclusion: the current model and a service based architecture to deliver the migration of a business model from of operation of telecommunications the promise of tailored connectivity for a CSP to DSP, where the latter intends to networks in a world that changes to 5G vast universe of use cases. New concepts expose their assets via API to provision is doomed. Automation will therefore be like network slicing and its provisioning and monetize innovative services with a a vital step for the new operating model of requirements, as well as the exponential flexible and reduced time-to-market, that DSP, and the application of automation in increase of both connected devices and are completely new and far more similar operation and network management will data traffic on the network, will make to the ones of over-the-top players and be a major step of the transformation of engineering and operations tasks, such hyperscalers than traditional telecom CSP to DSP. as network optimization, provisioning, connectivity services of the old days. supervision and maintenance, impossible to be carried out without automation. Operators need to move to Digital Service Provider Model, managing end-to-end services on the fly

Operators' multiple access networks Operators are implementing (2G/3G/4G/5G), force the need to tackle Network Virtualization, moving network operational complexity to a far more dynamic environment

Operators need to manage a growing Operators have relentless need to diversity of network services and OPEX and minimize CAPEX massive Internet of Things (IoT)

Picture 1 – Reasons for the change of the telecommunications network operating model in a world that moves to 5G

Automation CSPs’ approach to network management by legacy infrastructure and proprietary automation and overall business hardware systems, severely restricting their digitalization depends on several factors, ability to extract value from the deployment the most important of which include of new software applications, and ultimately Orchestration strategic priorities, competitive pressures, slowing down innovation. However, by region of operation, and the state of existing automating operations, implementing infrastructure. both E2E service orchestration and a domain Mediation/Abstraction/ Some of the biggest bottlenecks in a Orchestration layer, and using Artificial CSP’s digital transformation journey are Intelligence/Machine Learning Common Infrastructure the existing systems and architectural engines, CSPs enable their digital structures. Most CSPs are overwhelmed transformation journey towards DSP. Picture 2 – Future High-Level Network Architecture 5 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Leap ahead through automation

To tackle the challenges laid out Telco investments in AI will reach around US$11,2B by 20251 previously, automation will act as the foundation pillar to manage the E2E user experience. Those who do not want to be left trailing are heavily investing in leaping SON is estimated to represent a US$5,5B market until the end of 20222 towards the benefits of automation and its underlying technologies such as service orchestration, automation solutions such as (in the radio access) self Orchestration and NFV market is projected to surpass US$70B by 20243 organizing networks (SON) and Machine Learning (ML), as well as network virtualization, as can be easily inferred Sources: Tractica, Research and Markets, and Market Study Report from market study reports regarding the sector.

Early investors in automation are Improved Tier 3-5 CSP clients in Automation and Orchestration have already envisioning secure long- ROI experienced roughly 400% ROI over the past 5 years4 term returns and acknowledging the ability that automation applied to Sources: Cisco networking is demonstrating. This can be seen in relevant indicators such as cost reduction, increased group Improved 70% to 90% of repetitive network maintenance work can be Consistency automated via machine learning, improving consistency and efficiency5 efficiency, consistent service delivery, swift time-to-market and customer Sources: Deloitte, Tupl experience.

One of the front-running banners Team Team effort reduction of around37% in RAN engineering efficiency management by deploying SON algorithms5 carried by the automation wave is efficiency increase across the Sources: Deloitte, Cellwize whole chain that comprises the implementation and provisioning of modern networks. Automation gradually Time to 80% less engineering effort and1-3 hours to get a site ready for air5 voids numerous costly and time- Market consuming legacy practices regarding Sources: Cellwize manual equipment configuration, troubleshooting, and maintenance, greatly reducing OPEX. Reducing the risk Customer 100% consistent, 100x faster response to customer5 of human error also improves network Experience performance, taking to a self-learning, Sources: Tupl self-improving automated network.

6 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Frequent upgrades inherent to an automated configuration and cloud based auto-scaling network automation implementation avoiding network will require a shift from networking features to flexibly adapt in real-time to any the hassle of complex configurations, specific engineering staff into new skills typical unexpected disruptions. to each RAN vendor deployed in the network. of IT staff, as automation will take care of Network optimisation and configuration networking oversight leaving the application, Furthermore, automation enables stress-free is done at a much higher level as the software-based layer to human input. network maintenance as most repetitive network layer becomes abstracted, Moreover, the seamless implementation and time-consuming tasks are taken care of orchestrated and open via provisioning of these upgrades in the system will be by specialized engines blending Rule Base, gateways, parameter policies engines executed quickly and in a distributed manner, Closed Loop and ML operations. In the future and API that automatically translate as opposite to complex upgrade projects of RAN (Radio Access Network) ecosystem, abstracted decisions into networking today, coupled with a new-found network these engines run over an abstraction/ orders that the underlying network resilience that makes use of its dynamic orchestration layer that enables seamless elements OSS understand and execute. Key components of RAN automation

RAN automation steps until today have mostly been focused on Distributed-SON The automation of engineering and operations tasks, aiming and Centralized-SON, reaping benefits at increased consistency, efficiency and accuracy, ultimately that can be boosted by far more disruptive leading to cost reduction and improved customer experience via RAN automation approaches which can be network and specially service performance. implemented based on open platforms as we will see. While temporarily coexisting with legacy technologies, we see several The automation of E2E service provisioning, therefore leading new components and key building blocks to improved network agility, faster and simpler (self-)provisioning, for a new telecom operating model reality, shorter service deployment times and optimized resource usage, delivering the DSP reality promised by 5G. ultimately leading to business agility and improved customer This reality includes two major realms: experience via service management performance. Automation of engineering and operations tasks

For the automation of engineering and these open platforms expose data from (configuration management and operations tasks on 5G, we see as key the 5G network (and obviously, legacy performance management data) in an components Open RAN platforms that 4G/3G/2G) in an abstracted way, enabling abstracted way. Looking southbound, enable the northbound deployment of the algorithms to decide on optimization these platforms include a provisioning automation use cases. These can enable actions without needing to care about gateway that implements the provisioning algorithmic based applications, such as the specificities of each hardware vendor, for all RAN vendors also abstracting SON, or ML based applications which meaning that a network Orchestration the specific vendor details as much as can work together to deliver complex layer is paramount. possible, hiding the implementation engineering use cases not possible specificities. Conclusion is thata network before in a far more efficient way. The Looking northbound, these platforms provisioning gateway or module is a key fundamental concept is the fact that therefore expose the network Information central piece of open automation.

7 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Typically another key component is This field is the domain of expert Management, but should also be open to configuration management automation companies that have the expertise and integration of Customer Data, traces and tools that manage all the configuration market reach that enables them to be on geo-located data, ticketing systems and management templates and policies (rule the cutting edge of OSS provisioning for other data sources or systems that allow based, which may be dynamic), therefore traditional architectures and ORAN (Open the full digitalization of use cases. On top enforcing the CSP parameter policies RAN) or ONAP (Open Network Application of that, these platforms should adopt big and strategies, abstracting these from Platform). data frameworks that are scalable, robust the need to know specific parameters and easy to integrate with any kind of for every RAN equipment vendor, and ML platforms dedicated to telecoms data source and actuation engines that removing the hassle of having to change are another crucial component in RAN can use API to trigger network actions all templates and policies in case of a RAN automation, being typically northbound via southbound entities like OSS or RAN software upgrade. Full abstraction and to the Open RAN platforms, as the former abstraction/orchestration platforms. automation potential can only happen consumes data from the latter, and also if the platforms include a parameter use the latter to enforce and provision the For simple applications, ML algorithms policies management module. decisions taken by ML algorithms. based on ML programming may suffice, but to reach maximum automation Such platforms show traits of low code maturity, ML-enabling platforms with 56% of Mobile and should be designed to be used by the right mix of rules or algorithms Network Operators engineers with basic knowledge of based automation, northbound to globally have little data science and analytics, as opposed abstraction, provisioning and policy to generic ML frameworks which layers are mandatory. or no automation are designed for data scientists. ML in their networks6. platforms tailored for telecoms engineering It is really important to understand that in Sources: GSMA and operations use network data, including such an ecosystem, the role of engineers Performance, Configuration, and Fault suffers significant changes as:

Engineer focus on repetitive Engineer focus on mastering Engineer focus on manual Before cases diagnosis the parameter specificity of troubleshooting and work each vendor orders creation

Engineers focus of Engineers digitalize their Engineers spend more After automating well known tasks knowledge to ML or rule time on automating new and diagnosing complex based systems use cases and borderline cases

Picture 3 – Evolution of engineering work with the new automation paradigm

8 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

The automation of E2E service provisioning

By 2025, Becoming a DSP implies evolving from the In such an operating model, new CSP traditional approach of the manual services are mapped to technical of MNOs management and provisioning of E2E requirements which are then 80% service solutions. Instead, DSPs reach translated to E2E network functions expect to have flexible service chaining architectures service chains, which in their turn automated where instead of individually managing deliver a tailored fit connectivity each service with a E2E approach, slice to each individual customer/ 40% of independent, self-managing domains subscriber. their network are used. Additionally, these domains must be prepared to be orchestrated by a In the journey evolution from CSP to DSP, 6 operations . coordination entity or an external system. many network operators today have

Sources: GSMA Empowering services disaggregation orchestrated network and IT domains, acknowledges the implementation of new but few have the capability to orchestrate technological services, ability to operate E2E services given the fact that physical within a variety of vendors, and facilitate network functions still coexist with VNFs the integration of possible technological (non cloud native, but running over virtual partners. Rather than operating within machines) and with cloud native VNF. On an old-fashioned highly detailed view top of this mixed ecosystem, 5G non- of the network, organizations enhance standalone architectures are the norm for orchestration with a dynamic attitude. the next couple of years, meaning that the full promise of network slicing and tailor fit E2E service provisioning therefore relies on connectivity is yet in its early days. E2E service orchestrators that assure that the service chain is properly configured E2E. In a universe of 5G where the network Full automation will also resources are virtualized and the concept of slicing is introduced, the goal is for not happen without an E2E a fast service provisioning, preferably service orchestration. subscribed and managed by the customers themselves, delivering tailor-made quality of service on the fly.

Needless to say that domain orchestrators need to cascade below E2E service orchestrators, therefore assuring the coherence of all domains, such as RAN, IP backhaul, IP and DWDM transport, the core VNFs, potentially multi-access edge computing resources, and all the underlying virtual resources, managed via Management and Network Orchestration (MANO).

9 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

As a quick summary, Machine Learning and APP driven operations, automating tasks and enabling the move from reactive to proactive/predictive event the components where detection and solving key decisions are going to be needed for an Open Abstraction/Orchestration platforms and SON, enabling the implementation & provisioning of automation use cases across automated ecosystem all RAN types and vendors, including vRAN and traditional RAN in a 5G world are the following: E2E Orchestration, automatically provisioning the network slices across all involved domains and orchestrating the necessary quality of experience

The subsequent pictures illustrates in a very simple schematic the multiple Digital Channel domains to be considered in an

E2E mobile network and highlights BSS particularly the RAN Domain automation components, as well as the E2E key automation components. E2E Automation Apps based on ML, RPA, Rules… Key concepts are per domain orchestration and automation, as well E2E Service Orchestration as E2E orchestration and automation to provide E2E service chaining, assurance and automation. SDN Domain RAN Domain Other NFV Domains

Automation APPs APPs SON AI/ML ... Automation APPs Security Layer Integration Layer Analytics & Big Data SDN Orchestrator RAN Orchestration Domain Orchestration

OSS OSS MANO OSS MANO

SDN Controllers vCU, vDU, vBBU VNF

Picture 4 – Overall Network Architecture SDN Routers/ with highlight RAN and E2E automation (V)RAN Resources NFVI Whiteboxes key components

10 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Is automation only about technology?

Although it may seem The automation and virtualization journey key lessons learned that, if not addressed, has a clear impact on various dimensions will impair the overall success of the counter-intuitive, of the operating model and we have automation journey and subsequently the automation is definitely not collected valuable lessons from interacting larger journey of CSP to DSP. with customers and partners while mainly about technology. delivering projects. Namely, we collected 6

Focus on use cases

As any manager experienced in change is the way to deliver business value fast, management knows, organizational gain supporters across the organization up Focus on change requires incremental successes to executive suit, and test concepts quickly. use cases that can show the delivered value so that the organization can adhere and gain In fact, stating problems (in the current Manage Leverage momentum. There is no other way document scope, technical or service the change the Cloud in automation implementation but problems), putting a value to their Automation to implement a use case focused resolution (either expressed as a monetary key lessons approach, preferably in agile mode. or service quality impact value ), making learned a quick business case and delivering Blend Rule Think Other approaches such as starting by a minimum viable product (MVP) that Base and Open defining a one-size-fits-all technology stack addresses such problems is a way to ML or launch huge waterfall initiatives will deliver value early and test the use case Disrupt imply such large timeframes that benefits quickly. In case of success, the value the silos will not be delivered until a time where is realized, further improvements are the organization no longer believes in the potentially achievable, and the organization vision of the program anyway. moves ahead to new use cases. Picture 5 – Key lessons learned around telco automation Opposite to these approaches, focusing on As we will see, use cases need to be making the data available and manageable prioritized and change management is for exploitation, based on incremental use necessary, but without a use case mind- cases and delivered via an agile approach set, the organization will not start its by an empowered and skilled organization, journey in the right way.

11 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Disrupt the silos

One of the key lessons that we see A third vector is the fact that the ultimate In fact, without this structural change, customers learning the hard way is that it goal is now to manage digital services automation will only increase the is paramount to disrupt the data and that run on top of the networks, and not complexity of network operations instead operational silos associated with the just to manage the networks themselves. of reducing them. Among the biggest traditional CSP operational model. This implies a much richer set of data challenges is finding qualified people to to be analysed by network engineers, work on new and existing structures, due In fact, telecom networks engineering which ultimately needs to be embedded to the coexistence of the legacy network and operations are converging with IT, into a service management culture. Silo and the new automated processes. Many given the fact that networks are becoming disruption is therefore needed because CSPs are actively hiring new employees virtualized and hosted in IT infrastructure. in order to have an holistic approach to with skills to work with virtualization and automation and use case implementation, automation. Adding to this domain convergence, the engineers will need access to data that boundaries of engineering and operations used to be owned and managed by teams are becoming blurred, given the fact that that existed in an operating model that no telecom equipment software versions will longer makes sense. Use case must drive be managed as workloads in a DevOps data sharing, usage and the automation environment, as opposite to the very long platforms. On top, customer impacting deployment cycles of the past. use case centricity must drive the engineers mind-set.

Manage the change

All the previous topics lead to an management blueprints, orchestration, DevOps delivery approaches, new skills immediate conclusion: change organization changes, workforce skills need to be introduced. This means that management is paramount. changes among others. in addition to hiring new employees, there is an aspect that we identified as We are not alone in this conclusion It is obvious that new automated use important, which is the right training and as even technology delivery focused cases impact the old processes and identification of talents in the current staff companies like Ericsson state that "60% tasks delivered by the engineering and that can be used to catalyse the change. of the effort is change management operations teams, which lead to process - new processes, ways of working, management, toolchain changes and most Change management is ultimately about cross-functional collaboration, skills and likely to new IT integrations. value realization and culture change, being competences7". the other components to be managed, However, given that automation requires means to reach the ultimate ends of Change in this context is holistic as tools based on ML, SON, mediation, supporting the journey of moving from automation management encompasses orchestration and others, using agile CSP to DSP. processes, tools, IT systems, data software development methods and

12 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Think Open

More Effective • Best Practices Having an open-minded way of • Metrics networking is what is propelling the main spearheading leaders in More Affordable the sector. It is not an architecture, Quicker Reward • Collaborative R&D but rather a perspective. Developing an • Common language • Standardization open ecosystem where different vendor • Blueprints & Toolkits generated data is both abstracted into accessible mediums as well as not time locked inside organizational clusters, empowers the whole operation by Cost removing communication hurdles and streamlining access to useful information for everyone, by everyone, besides without collaboration with collaboration allowing the organization to embrace self empowerment via agile use cases Sources: adapted from ‘Open Digital Architecture’, tmforum8 development. Thinking open is therefore a technical extension of silo disruption. Picture 6 - How collaboration helps transformation in the communications industry

The smartest players know that efficiency orders from higher level abstraction The common language of the modern comes from shared knowledge, promoting enabled interfaces. digital world – APIs – is the main foundation open platforms where useful data and behind the current surge of big data-based API can be accessed to generate value for Furthermore, this open-thinking services, a standardized way companies everyone. This “think open” philosophy perspective should be implemented use to share useful data in order to is commanding the abstraction and locally by piercing broad communication maximize data usage and monetization. orchestration revolution. Providing channels and disrupting internal silos in Don’t get comfortable in looking at what ways for different vendors to intertwine organizations, employing an open data is possible today in a post-API world, their solutions to the management and philosophy. Enforcing a systematic sync adopt a future-proof way of thinking and automation layers, leaving provisioning to between the BSS (Business Support plan to make all your current and future automated processes that enable these System) and OSS components is key to tools API accessible – empower others by different equipments to be accessed via achieve an hive-mind operation, where empowering you. APIs are the key engine provisioning gateways that implement group efficiency is the end-goal. behind abstraction, providing a common network orchestration and automation input/output language for all vendors.

Blend Rule Base and ML

So the saying goes: “When you have a Knowing when to employ simpler rule- automation skills internally and accessing hammer, everything looks like a nail.” No based techniques or more complex data scientists is becoming increasingly one denies the ground-breaking power of closed loop differential algorithms such harder. ML can be left to applications ML applications in many use-cases, reaping as SON to specific use-cases where where it can really make a difference, while the benefits brought by a self-learning high efficiency is already attainable, as organizations can leverage what they digital tool which is going to completely opposed to employing potentially even already have with rule-based techniques. revolutionize some sectors as well as more complex to model ML-based ones, This dynamic blend enforces efficiency generate completely new ones. However, is what can differentiate the ROI (Return – it uses the best of multiple worlds. not all use-cases are suited for/need ML to on Investment). Additionally, most create automation benefits. CSPs already have access to rule-based

13 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Don’t use a one-size-fits-all approach. CSPs moving to DSP need to look at Cognitive Technologies their available arsenal and gauge which initiatives and use cases really benefit from each ground-breaking technology.

Machine Neural Robotic Process Natural Language Learning Networks Automation Processing

IDC forecasts show spending on this segment to reach USD 77.6 billion in 2022

“Personalize and contextualize the human-technology interaction, allowing businesses to provide tailored language and image-based information and services, with minimal or no human involvement9.” Sources: Deloitte Tech Trends 2020

Picture 7 – Overview of ML Technologies

Leverage the Cloud

Avoid the storm and choose the cloud. communication, CSPs not venturing edge-focused quality of experience (QoE) Changing an organization’s network into cloud-supported architectures philosophy. architecture into a cloud-based one are going to be left behind in a legacy- is a way to leverage the benefits of ridden, complex, and inefficient Again there is no one-size-fits-all for any decentralized computing for modern ecosystem in the long term. This makes case. Telcos wanting to course-correct into network delivery and oversight. It makes no sense, even because very relevant cloud-automation should analyse their it so that organizations can deallocate the moves from the hyperscalers already use-cases and see first, which can benefit time and investments previously put in enable their full stack on-premises, from a cloud adaptation, and second, costly legacy processing power and shift it therefore bringing the scale benefits into which cloud provider best suits those to service development and optimization distributed datacentre architectures, applications, as well as where each network – directly steering those resources into blending the advantages of public and function and platform will reside, from core value-generating operations instead of private cloud. to edge. value consuming ones, whilst at the same time localizing the delivery of their services No CSP-to-DSP transformation is closer to the end-user as needed. achievable without the cloud. It is a mandatory requirement for a connected By 2025, one third of In a path to a world ruled by data collection world where data based automation of MNOs will have automated and analysis on an unfathomable level, cloud native VNF is second-nature and 5G and where a new breed of networking represents networking as whole and where 80% of their network is on the way to being virtualized, open, software reigns above a self-adapting, operations6. self-organizing and abstracted, enabling decentralized, hardware infrastructure. flexible and easy to provision slices of 5G The focus is on the end-user and an Sources: GSMA

14 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Which RAN use cases are we seeing on the market?

Among our experiences with clients and market research, we have assembled a set of powerful use cases, which have shown great results where implemented. The automation use cases that we have explored mainly can be separated into three families or large domains, namely: Network Planning and Engineering, Network & Service Operations and Customer or Service care:

Picture 8 – RAN high priority use cases

Network Planning and Engineering

VoLTE Automated Optimization interactions regardless of operating layer automation techniques enabling the Contrary to data, IP voice services have or vendor to ensure compatibility and power to engineer or operate networks. significant less tolerance to drops or delays coherence between the different services. As a consequence, engineering human in communication packets when compared Automation of the aforementioned tasks, resources can be redirected to more to streaming applications or internet traffic. allows the maintenance of consistency important and meaningful tasks and Therefore, CSPs providing IP voice services between OSS, design workbooks and the organizations can benefit from time savings over LTE (VoLTE) need to differentiate and live sites across design to implementation and decreasing complexity. automate parameter optimization in order process, ending on optimization. to maintain or gain competitive advantage CAPEX Avoidance Via Capacity in the market. Automating dynamic link 5G and 4G Zero Touch Network Rollout Management quality management and automating Network densification will be a crucial Implementing measures to ensure the mobility management will greatly improve activity to address continuous increase on optimization of the utilization of available the call setup and drop rates, packet loss capacity demands on mobile networks. By resources is of utmost importance in a and voice quality, having a direct impact on fully automating the mass rollout of new CAPEX constrained reality. Automation consumer quality of experience (QoE). sites, not only customer experience but allows the understanding of traffic also engineering efficiency will increase. trends within the network, overused Configuration and Policy Automation Being able to create an abstraction layer and underused neighbour sites and Controlling and managing networks across between network providers and rollout recommend actions to attend the a landscape of vendors and proprietary management and operations will empower dynamics of continuous yet dynamic technologies endures a daunting task for faster adaptability to specific requirements traffic growth. Machine learning enables organizations. Managing the engineering such as massive capacity, infinitesimal the detection of both the traffic trends, as knowledge base expressed in parameter latency, supreme reliability, personalized well as the correlation of such trends with templates, and fighting parameter customer experience , etc. sites usage and forecasting of the impact inconsistencies is another key challenge. of traffic management, allowing to an Networks are in need of solutions It is imperial for organizations to adopt improved asset usage. that manage policy and configuration zero touch technology, by employing

15 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Use Case Typical Data Main Outputs Key Benefits

VoLTE VoLTE and Mobile Configuration • Enhance E2E call Automated Broadband Configuration regarding handovers, execution Optimization (CM) and Performance codes, antennae tilts, • Improve drop call rate Management (PM) data uplink parameters, • Better QoE neighbouring cells

Configuration Configuration Seamless • Parameters consistency and Policy Management Data, configuration of • Policies control and Automation potentially from multiple RAN enforcement different vendors platforms • Personalization of practices and technologies • Know how sharing

5G and 4G Zero CM and PM data, Pre-optimized • Increased efficiency Touch Network as well as physical configuration scripts • Rollout capacity increase for automated network Rollout sites data • Time savings integration

CAPEX Avoidance Performance Optimized • Increase CAPEX efficiency via Capacity management data, offloading to • Reduce network costs Management physical data information accessible assets • Improved quality of service and configuration in case of network overload management data

Table 1: Network Engineering

Network & Service Operations

NOC Automation (Tier-1) CSPs can harness the transformation on the perceived user experience. Each day Network Operation Centres of data into insights. By being able to Additionally, cost reduction will also be (NOC) are challenged with enormous uninterruptedly gather information from a product of the automation of these amounts of alarms and warnings. With service, configuration and performance tasks, as these are technically challenging the increasing network complexity and data it is possible to detect, diagnose and activities and often cause long localized customer demand for uninterrupted forecast events and trends to act instantly service shutdowns. services, expectations are sky-high. The on involved network elements that are pace at which organizations are able not malfunctioning. These diagnostics Predictive Network Maintenance to respond will determine their future can be real-time triggers of optimization Improper reactive approach maintenance success. By leveraging ML to correlate actions implemented by the RAN practices of the physical layer of the alarms and selecting meaningful insights, orchestrators and can enable increasingly network results in serious networking automating activities of Tier-1 is possible. fast and adaptable networks along with issues that can be hard to track down The automatic root cause analysis minimization of errors originated by in complex networks. Automating the and action recommendation result in human behaviour. maintenance process and moving from a troubleshooting accuracy, proactive reactive to predictive practice, is a relevant responses and significant time savings Self-Healing step in the right direction towards ensuring due to faster, systematic and consistent Having a network capable of regenerating a consistent operation of the network by processes at this level of the NOC. On itself after a failure in a world of minimizing the risk of sudden equipment top of these benefits, Tier-2 tickets are highly dense networks is a mandatory malfunction and potentially reducing reduced and team efficiency is dramatically requirement. Automating the process expensive truck roll costs by minimizing increased. of executing handoff and compensation their urgency or even avoiding them, techniques to seamlessly offload while maximizing customer experience via Continuous Service Improvement workloads to healthy surrounding outage avoidance. Networks today produce an overwhelming nodes before initiating troubleshooting amount of data. However only few procedures can have massive impact

16 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Use Case Typical Data Main Outputs Key Benefits

NOC Fault management (FM) Automated root cause • Troubleshoot accuracy Automation and network elements analysis and tickets • Time to resolve (Tier-1) status data closure and • Team efficiency automated action • Escalations reduction recommendations

Continuous CM, FM and PM data, Automated diagnostics • Faster issue resolution Service service data, geolocation, to be used as triggers for • Higher efficiency Improvement physical sites data and engineering, operations • Human error minimization others or orchestration actions • Maximized QoE

Self-Healing PM, FM and CM data as Actions that handoff • Autonomous corrections well as network elements the traffic to non • Constant fix of network status data affected nodes impairments • Improved QoE

Predictive Fault management Pre-emptive • Operations consistency Network and network elements Maintenance • Minimize malfunctioning Maintenance status data Instructions risk

Table 2: Network & Services Operation

Customer Care

Automatic Service Tickets Resolution in handling repetitive tasks as well as of management processes supporting Handling customer complaint tickets is improving response times to costumers, effective configuration, assurance, and both an inevitable and resource-heavy while moving the organization to a service control of services and their supporting undertaking for any CSP. Employing an centric approach. resources. An orchestrated environment ML-based Automatic Service Tickets allows the definition of end-to-end SLA Resolution system is paramount to greatly Network Service Assurance enforcement / assurance / resolution improve complaints resolution efficiency Connectivity Services relies on the policies associated with the Product by having specialized algorithms learn and orchestration of distributed capabilities Offering (which contains QoS/QoE KPIs detect patterns across the information across networks domains and even to be enforced). Going further the policy gathered from complaints over time and operators to enable end-to-end based rules can be further adjusted automatically act upon them, minimizing management. Orchestration via policy- automatically by applying ML-based human interaction and time wasted based management capabilities provide capabilities. by engineering and operations teams rules-based coordination and automation

Use Case Typical Data Main Outputs Key Benefits

Automatic Customer complaint Automatic • Minimize human interaction Service Tickets tickets, PM, FM, CM troubleshooting • Improve response times Resolution and customer recommendation • Time savings service data • QoE

Network PM, FM, CM Service orchestration • Improved service provisioning Service and customer rules enforced end-to- • Faster time to market Assurance service data • Consistent network experience -end in multiple OSS • High Efficiency

Table 3: Customer and Service Care

17 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

How are we succeeding with our customers?

At the Deloitte Telecoms Engineering Automation & Partnerships – Using Studies confirm that Centre of Expertise, we believe striving an healthy synergy between our own in the automation journey requires three engineering and project management the time-to-market factor very different sets of skills. First, both skills and our open philosophy to work for organizations using engineering and business perspectives closely through partnerships with need to work together to avoid the best of breed companies specialising agile approaches improves “build and they will come” trap. With in key components for automation on average by 18 to 20 both lenses defining the strategy, the implementation we use the right solutions implementation requires automation skills to the required use cases, avoiding a one- percent. Agile helps but also an ecosystem of partnerships size-fits-all approach. organizations of all sizes that allow getting the best of breed in each specific situation, without allowing Transformation – It is obvious, although to respond faster to customizations and adaptations to hinder many times neglected, that if we market demands and future changes in the business. And as we implement automation, the old processes, have explored in a previous chapter, this metrics, tools, functions and even deliver customers what journey is much more than technology, so people skills are impacted, meaning that they really want faster a transformational focus needs to be in process transformation and measuring 10 place. the benefits is a key factor for success than ever before . Our three pillars of automation are based and for providing the visibility that will Sources: adapted from Atlassian upon a functional narrative of the whole catalyse further change via automation. transformative process from a non- Transformation management is key. automated starting point, into a more mature automation states via Business Engineering & Business Transformation Process (BTP):

Engineering & Business – We use specialized network engineering resources and the business know-how. Having a deep understanding of both engineering and of how telecom business is run enables us to optimize our solutions from Automation & a use case based perspective. Partnerships

Transformation

Picture 9 – The Big Three in Automation

18 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

With the three pillars in place, a succeed Define and Build: The strategic phase. a functioning solution. From there, an or fail fast methodology needs to be used. An holistic assessment is done to iterative refinement process is carried Therefore, our agile-based approach understand client’s needs and draw a out in order to shape the solution to cater for Network automation is structured in roadmap for the journey. Use cases and directly to the client’s target use case three typical phases: Define and Build, respective business cases are defined and and business benefit. The business case Agile Delivery and Agile Benefits given different priorities to aid in laying validation process is carried out to see if Realization, where we apply our out the most suitable rollout plan. If/ First Office Application (FOA) is possible. engineering and business capabilities, when needed, RFPs are drawn up to find automation skills and partnerships and the best partnerships for the journey. Agile Benefits Realization: Smooth our transformation best practices. We Alternatively we can deploy our partners delivery. With client’s approval, the use this arsenal of diverse tools to be to deliver joint value. functioning automation model is prepared for various environments and scaled up for real-world deployment, client necessities – it makes our solution Agile Delivery: Delivery and execution its performance analysed by tracking versatile without compromising on its through an agile approach. We focus on the benefits manifested by the robustness for a large spectrum of implementing the planned solution in a implementation, and optimized by applications. controlled scenario to guarantee harm- refining it for the use cases. On top, free testing and securing a final, functional transformation management is carried out result. A MVP is established, ensuring v ia BTP.

Agile Delivery

Run & optimize

Identify Design & Deploy Achieve Scale Up Manage Goals acquire MVP FOA the change solution

• Automation building blocks deployment • MVP design and implementation • Agile Solution and Project Management • Business case validation and GO/NO validation reports

Picture 10 – Methodology and standard deliverables

19 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Conclusions

The fifth generation of mobile consolidating them into a single the network: from network engineering telecommunication systems will point of data integration, including planning to network & service operations. provide a disruptive leap forward in the performance management and Prioritizing investments according to the telecom industry, but unlike previous configuration management data potential benefits, helping to unlock the generations, it won’t be mostly about (or any other network or service highest possible value from the journey more bandwidth. CAPEX, OPEX and performance related data sources). where three foundational competencies revenue constrained CSPs will have to are key: business and engineering revolutionize themselves and evolve • Northbound to the RAN orchestration perspectives, transformational capabilities to DSPs, exposing their assets via API lies the automation applications and technical automation skills. to provision and monetize innovative layer using machine learning and APP services with a flexibility and reduced driven operations. These can include time to market that are totally new and SON engines, rule based algorithms, It is important to reiterate: far more similar to over-the-top players the ML engines, a blend of the two, and hyperscalers than before. And all and so on. there is no attainable way this will happen while investments in of migrating from CSP to densifying the network will increase to • Even more northbound is the end- cope with the raging quality of service to-end service orchestration, DSP without modularly and bandwidth that customers demand. which coordinates the RAN automating the RAN orchestration with other domains In this context, Deloitte Telecoms therefore orchestrating end-to-end network from the end-to- Engineering Centre of Expertise believes service assurance. end orchestration layer automation is the foundational pillar to tackle the challenge of increased However, this challenge goes beyond down to the abstraction complexity, changing business models technology. Companies will have to point. and easing investment constraints. transform internally, creating and Sharing this vision, some players have maintaining an open mentality, sharing moved ahead and are already reaping responsibilities between IT and Networks Network complexity in a post-5G world is benefits. teams or thinking in organisation-wide beyond the control of legacy networking automation use cases. Externally, getting practices. Deloitte Telecom Engineering In this paper we focused on RAN the benefit of a greater and dynamic Centre of Expertise structured approach to automation and in the role the key ecosystem, leveraging a new open, API automation takes care of this by employing automation blocks in a 5G world, from a based, data driven and cloud-native a fluid agile-based, intelligently phased high-level perspective: architecture. implementation with the technical backing necessary to comfortably work on a broad • The RAN orchestration/ Deloitte Tee CoE has been working with spectrum of environments with different mediation/abstraction layer, multiple telecom operators across the needs. Ensure your place in the future of which being northbound to the OSS world and, together with an ecosystem flexible digital providers by starting your abstracts and orchestrates diverse of partners, developing use cases to journey now. (v)RAN resources in the system, bring important and tangible benefits for

20 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

An ecosystem of automation specialist partners

Deloitte Tee CoE is a telecom network different levels (network and service) and we recognize the value of specialized specialized group, focused in enabling a a diverse potential of automation, there automation companies in order to jointly smooth transition into the automated 5G is no organization who masters the whole deliver end-to-end use cases and reap the networks of the future as CSP move to multitude of skills necessary to cover full benefits of engineering and operations DSP. all the required fields that maximize the automation and our partnerships with automation benefits. TUPL and Cellwize are living examples of In this multi-domain 5G world, which relies this. on a federation architecture that includes Being specialized in engineering and in the multiple technologies, orchestration at transformation brought by automation,

TELCO AUTOMATION EXPERTS A NETWORK AUTOMATION AND ORCHESTRATION to design, deploy and implement specific use cases PLATFORM, based on an open architecture

Who TUPL is What Cellwize Does

Experts in Networks, Software and AI Chime, a cloudified RAN automation and orchestration platform, can accelerate Backgrounds from multiple Mobile deployment of 5G networks with ease and face Operators, Big OEMs and other start-ups the future with confidence

Chime eliminates complexity and introduces smart automation and management, enabling What TUPL Does next generation networks System for complex task automation. Abstracts Chime allows to code in any language, and network complexity automatically for minimum enabling the creation of any needed application human supervision, leveraging AI

Cellwize Operates globally, has a rich partnership Builds digital super engineers via SW and ML catalogue, with dozens of projects running today, to address the top three problems for CSPs: enabling leading service providers worldwide efficiency, CSAT , and performance with a unique RAN automation experience

All of this is achieved via Intelligent Process Field-proven value Automation

An open architecture for handling all types of Field-proven value RAN vendor data, abstracting it, and connecting to any application via open APIs Make use of existing Big Data and AI tech with TuplOS Agile & scalable for handling the multiple complexities of rolling out and managing 5G Respond to customers 10x faster networks

Improve operations and engineering team Full automation of complex heterogeneous efficiency by offloading repetitive tasks to networks, including both traditional network kit automation players as well as new vRAN players

Access network data instantly via centralized interface 21 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Who we are

Our focus is The Telecom Engineering Centre of Excellence (TEE) has a footprint spanning 4 continents and has delivered projects in over 50 on making countries being currently supported by 1 headquarters and 5 an impact branches with circa 80 telecom engineers. that matters. We deliver professional telecommunications engineering consulting services globally supporting our customers via a global network of offices from Europe to Australia, having delivered over 200 projects globally in over 50 telecom operators.

Working together with other Deloitte practices when needed, our integrated business ecosystem provides a differentiated set of core foundations beyond telecoms engineering which, when combined, enable a portfolio that assures end-to-end business impact beyond the deployment of technical solutions.

Amsterdam Netherlands

London UK Lisbon Portugal

Luanda Bangkok Angola Thailand

Melbourne Australia

22 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

Glossary

APP – Application API – Application Programming Interface AI – Artificial Intelligence BSS – Business Support System BTP – Business Process Transformation CAPEX – Capital Expenditure CoE – Centre of Excellence CSP – Communication Service Provider DWDM – Dense Wavelength Division Multiplexing DSP – Digital Service Provider E2E – End-to-End FOA – First Office Application IT – Information Technology IoT – Internet of Things IP – Internet Protocol KPI – Key Performance Indicator LTE – Long Term Evolution ML – Machine Learning MANO – Management and Network Orchestration MVP – Minimum Viable Product MNO – Mobile Network Operator NOC – Network Operations Centre NFV – Network Functions Virtualization ORAN – Open Radio Access Network OPEX – Operational Expenditure OSS – Operations Support System QoE – Quality of Experience QoS – Quality of Service RAN – Radio Access Network RFP – Request for Proposal ROI – Return on Investment SON – Self Organizing Network SDN – Software Defined Network SA – Standalone Architecture VNF – Virtual Network Function vRAN – Virtual Radio Access Network VoLTE – Voice over Long Term Evolution

23 The Age of Telecom Network Automation | Automating Engineering and Network Operation RAN Processes

References

Endnotes

1 https://tractica.omdia.com/research/artificial-intelligence-for- telecommunications-applications/ "Artificial Intelligence for Telecommunications Applications" Report 2 https://www.researchandmarkets.com/reports/4621704/son-self- organizing-networks-in-the-5g-era "SON (Self-Organizing Networks) in the 5G Era: 2019 - 2030 - Opportunities, Challenges, Strategies & Forecasts" Report 3 https://www.marketstudyreport.com/reports/network-function- virtualization-nfv-market "Global CSP NFV Management and Orchestration Solutions Market Size, Status and Forecast 2019-2025" Report 4 https://www.cisco.com/c/en/us/products/collateral/cloud- systems-management/network-services-orchestrator/white- paper-c11-738289.html "The Business Benefits of Automation and Orchestration White Paper" White Paper 5 Deloitte Internal Research 6 https://www.gsma.com/futurenetworks/wiki/ai-automation-an- overview/ "AI & Automation: An Overview" Online Document 7 https://www.ericsson.com/en/managed-services/supercharging- customer-experience-through-ai-and-automation “Supercharging customer experience through AI and automation” Report 8 https://www.tmforum.org/resources/whitepapers/open-digital- architecture/ "Open Digital Architecture" Report 9 https://www2.deloitte.com/content/dam/insights/us/articles/techtrends-2020/DI_TechTrends2020.pdf "DEloitte Tech Trends 2020" Report 10 https://www.atlassian.com/dam/jcr:d9ce12ec-c1a1-4495-b7dc- 16c38d6b594e/Atlassian_Harvard_Business_Review_Agile_Report. pdf "AGILE PRACTICE: THE COMPETITIVE ADVANTAGE FOR A DIGITAL AGE" Report

24 Contacts

Sponsor Experts

Pedro Tavares Hugo Pinto Maurício Pereira Pedro Sanguinho Telecom Engineering Telecom Engineering Telecom Engineering Telecom Engineering Centre of Excellence Centre of Excellence Centre of Excellence Centre of Excellence (TEE) Leader (TEE) Senior Manager (TEE) Manager (TEE) Senior Manager [email protected] [email protected] [email protected] [email protected]

Acknowledgements

Special thanks to the Deloitte TEE Team who contributed to this publication in terms of researching, providing expertise, and coordinating:

Catarina Santos | Filipe Leonardo | Guilherme Oliveira Henrique Piedade | Miguel Dias | Thiago Silva Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited (“DTTL”), its global network of member firms, and their related entities. DTTL (also referred to as “Deloitte Global”) and each of its member firms are legally separate and independent entities. DTTL does not provide services to clients. Please see www. deloitte.com/about to learn more.

Deloitte is a leading global provider of audit and assurance, consulting, financial advisory, risk advisory, tax and related services. Our network of member firms in more than 150 countries and territories serves four out of five Fortune Global 500® companies. Learn how Deloitte’s approximately 312,000 people make an impact that matters at www.deloitte.com.

This communication contains general information only, and none of Deloitte Touche Tohmatsu Limited, its member firms or their related entities (collectively, the “Deloitte network”) is, by means of this communication, rendering professional advice or services. Before making any decision or taking any action that may affect your finances or your business, you should consult a qualified professional adviser. No entity in the Deloitte network shall be responsible for any loss whatsoever sustained by any person who relies on this communication.