Ultra reliabilityUltra low latencyVery high availability
NB-IoT + Cat-M1 (LTE and 5G NR) LTE + 5G NR 5G NR 5G NR
Massive IoT
CriticalIoT
Broadband IoT
Industrial AutomationIoT
Commercial – Growth Early pilots & standardization
Asset management
Fleet Management
AutomotiveC-ITS
VR/AR
Drones/UAV Smart Grid Automation
Industrial Application &
Control
RemoteSurgery
RemoteManufacturing
SmartMetering
1 3 7 0
One network – multiple use cases and industries
Presenter
Presentation Notes
3 examples – Smart Metering - T-Mobile (refer slides 7&8) – consumer facing plan looks simple, but to manage aspects like scalability, normalize traffic from meters, ip v/s non-ip devices, additional monetization models based on API calls, throughput rate – handling settlement with partners, notifications VR/AR – Singapore F1 Grand Prix – (refer slides 9, 10 & 11) – 5G n/w slice – CAF, monetization based on the SLA/KPI being offered, ongoing 3GPP release 16 specs which will include SLA/ KPI based paraneters on the basis of which aspects like penalties would need to be automatically calculated. Partner ecosystem – F1, advt, video feed. Operator takes the role of API provider Remote Manufacturing – Smart Factory Texas (refer slide 12) ----------------------------------------------------------------------------------------------- Beyond what NB-IoT and Cat-M can deliver and address Re-use higher performing LTE and NR networks to tackle more complex use cases in the same verticals Introduce concept of slicing and also cases where a total separate large enough slice or even dedicated network is needed to make a non-public network Mention now even in utilities, beyond metering, LTE is needed to connect some grid stations elements How LTE can and is being used in automotive to provide consumer services but also starting to enable C-ITS Example, Vehicle to Traffic lights communication in real time … better safety and traffic flow Mention network from Swisscom for trains Mention how LTE can enable commercial drone usage Industrial IoT: To use 5G networks and technology to facilitate simultaneous product customization and maximum production output – without sacrificing flexibility, traceability, sustainability, or safety. We did this by creating a network of connected machines that allow manufacturers to collect, analyze, and distribute data in real time. Enhancing connectivity and keeping workers continuously in the loop will enable manufacturers to acquire and access much larger amounts of data – at far greater speeds – more efficiently than ever before.
Ericsson Internal | 2018-02-21
Move fast without breaking things – Assemble rather than create services
Streaming interface with 5G CHF for CDR generation
— 5G 3GPP CHF compliance
— New Access CAF— Slices, resources,
connectivity & mobility based charging
— Offer Personalization— Partner product
creation— Models for 5G slices— 5G parameters — B2B Templates
Presenter
Presentation Notes
No one knows what the future will bring Use cases and requirements are emerging, even for 5G All we know is that there will be rapid change Move faster than ever before Grow current business Expand to 5G Try out new services Succeed or fail fast Assemble rather than create services Reduce time between idea and revenue Flexible business models Decrease operations cost
Ericsson Internal | 2018-02-21
Intent driven Blueprint for Digital Telco
Monetization Management
Ericsson Internal | 2018-02-21
Increased OPEX & complexity in management
Massive networkcomplexity
Distributedcloud
Heterogenousnetworks
New usecases
Network slices
Ultra coverageand bandwidth
Ultrareliable
Ultralowcost RAN
densification
OpEx & Complexity
Service Management
New Industries
SDN/NFV &Network Slicing
CEM
IoT
5G
Dealing with OPEX and network performance in this environment will go beyond the reach of humans
Operations to address the complexity and reduce costs by using Intent Based Orchestration
Target is to reach a level of automation where the network can be considered zero touch
2G Time3G 4G 5G1990s 2000s 2010s 2020s
Presenter
Presentation Notes
The complexity of today’s networks is massive and will increase further with the introduction of 5G and IoT. Intent Based Orchestration is a vital technology for us to handle this complexity going forward and will be the way for us to break the complexity curve. In 5G, the number of functionality dimensions and number of data points have increased significantly, for example the number of parameters per network node is significantly more than in 4G. The complexity of running a 5G network, including data points in IoT, makes it close to impossible for a human brain to comprehend and find an optimized solution. In this situation Intent based orchestration with use of algorithms, AI/ML which can adopt to the complex environment and the constant changes in user and traffic behavior will outperform any human driven solution.
Ericsson Internal | 2018-02-21
Intent based orchestra tion
Intent based orchestration – Declarative approach to service modeling to specify the intent and not the ho10-20% TCO & TTM improvements 30—50%
Automation of …1. VNF onboarding & testing2. VNF lifecycle management
Automation of …3. VNF onboarding & Testing4. VNF lifecycle management5. Datacenter networking6. Service topology visualiza tion
Automation of …7. Service orchestra tion8. AI/ ML closed loop assurance9. Transport orchestra tion10.Network slicing
A combination of assurance closed loops and intelligent workload placement functions, enable a real Intent based orchestration
Presenter
Presentation Notes
"Intent-based orchestration or networking is kind of a northbound interface between an infrastructure controller (such as ODL, ONOS, or OpenStack) and a higher-level business system (such as an order management system within OSS). This northbound abstraction is similar to the MANO layering: NFV Orchestrator, VNF Manager and Virtual Infrastructure Manager. At the top layer, instructions are quite generic -- like create a new VPN or create a slice. At the bottom layer, instructions are very granular and device specific. Intent isn’t concerned with the low-level APIs used to control the infrastructure but is instead focused on higher-level business needs. What makes Intent Based different is that it uses a declarative language for this interface rather than an imperative one. A declarative language uses statements to express goals but doesn't specify how to accomplish those goals. In contrast, an imperative language focuses on how the goals are achieved. Imperative is a set of programmed rules while declarative uses an optimizer or planner to make decisions. " Which this means in our case is that we should expose only declarative service definitions probably in a GUI that map to a TOSCA service definition with business goals. The details on how these goals map to actual workflow operations are not shown and in the best of the cases is that the actual workflows are created on demand based on generic workflow templates that get 'specialized' i.e. parametrized based on the intents defined declaratively. At run time, once a service template has been selected, you need capabilities in the orchestration engine to enable this approach – we believe that it is a combination of intelligent workload placement and AI/ML driven closed loop assurance that enables a true intent based orchestration.
Ericsson Internal | 2018-02-21
Intelligent workload placement
Screens everywhere
On demand information
Process automation
Metering and smart grid
Connected doctors and patients
New tools
Real-time information vehicle to vehicle
Flow management and remote supervision
Resource management and automation
Monitoring and medication e-care
Immersive experience
Autonomous control
Cloud robotics and remote control
Machine intelligence and real time control
Remote operations
What
Slic
es to
polo
gy te
mpl
ates
eMBB
Critical MTC
MassiveMTC
ApplicationVirtualizedRAN vEPC
vEPC
Application
VirtualizedRAN
VirtualizedRAN vEPC Application
Radio processing
Radio processing
Radio processing
Radio cloudWAN
Edge CloudWAN
Regional/Central cloud
WANInternet/Public
cloud
The user specifies what service he wants… 1
Where & How— Location— Latency— Bandwidth— QoS (SLA
KPIs)— Reliability &
availability
Design and Assign for Intelligent Workloads placement
…tuning few high level characteristics (Where and how) 2
OUT (Deployment)IN (Definition)
Review the proposed workload placement solution for approval 4 Determine best place to put the workloads based on service characteristics 3
Presenter
Presentation Notes
Intelligent Workload Placement algorithm will decide the placement. The service template simply defines the list of workloads (NFs) and links needed to create a service. Workload and links are described with no information of their own implementation so that the algorithm can decide the best implementation possible given the network topology and its state. At the time of instantiation of the service, user specifies the parameters of the service such as location, bandwidth, latency. The orchestrator runs an “instance design” operations which is pre-requisite for any other LCM operations so that the algorithm is invoked to generate the best possible service graph. The service graph is then stored in the orchestrator instance repository and the orchestrator applies the instantiation part of the TOSCA execution plan to instantiate the workload and trigger
Ericsson Internal | 2018-02-21
Open and closed loop automation – Leverage AI & ML
Open loop: — Decision loop is partly automated.— Humans make the decisions
Closed loop:— Decision loop is fully automated— Humans supervise the overall performance.
Presenter
Presentation Notes
In order to support both partial and full automation, Analytics, Policy and COM provide different interaction points to external systems and to users. The network management environment makes sure that data from the network is mediated and normalized. The data can also be augmented by combining different pieces together (Analytic Session Records is one example of this). The network data can then be sent to an external system or be presented to a user without further refinement but it can also be fed directly to Analytics. The role of Analytics is to create deeper insights on the situation at hand, which again could be sent to an external system or a user or it could trigger Policy. Based on the insight, Policy can make a recommendation to an external system or a user as for which actions to be taken given the current situation. It may also request COM to execute these actions. Whether the decision has been made by Policy, a user or an external system, COM is specialized in implementing the requested action by transforming it into a series of actual changes and interactions with the network. The network management environment can mediate such changes and interactions from COM and directly from a user or an external system. To maintain consistency, the network management environment prohibits unauthorized changes to take place, i.e. it protects changing of locked parameters unless initiated by the configuration template handling.
Ericsson Internal | 2018-02-21
Key takeaways
5G/IoT and NFV/SDN represent new monetization opportunities – but also complexity
Intent Based Orchestration to drive automation, leading to TCO and TTM reduction
Available standards (TM Open API, TMF SID) drive integration and interoperability of heterogeneous multi-technology solutions
1 It’s a Journey – Start Now !! Incrementa l capability building
2 New Competencies – Build/ Invest NOW !!Agile Software factory
3 It will be an interdependent Eco-systemUse flexible, standards-based integra tion
Presenter
Presentation Notes
Orchestration is everywhere in a virtualized and 5G/IoT world. Orchestration happens across multiple network and IT systems Key to managing the complexity is to adopt a catalog-driven approach to orchestration A catalog-driven approach to orchestration is essential to success A master, enterprise catalog underpins PSR models, and identifies fulfillment workflows