Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Ericsson Technology Review - Issue 2, 2018

297 views

Published on

Technology development keeps getting faster and more interconnected, with new innovations appearing every day. As a result, we’re swiftly moving toward the realization of the “Augmented Connected Society” – a world characterized by ubiquitous internet access for all, self-learning robots and truly intuitive interaction between humans and machines. But how can our industry best prepare for this future?

In my role as CTO, I have the challenging and exhilarating annual task of identifying the five technology trends of the future that are (or will be) most relevant to our industry. You can find my insights and reflections in the Technology Trends article included in this issue of the magazine.

It is my hope that the Technology Trends article, together with the other five articles in this issue, will generate a variety of stimulating future-focused discussions in your workplace. Please feel free to share links to the magazine and/or individual articles with your colleagues and other contacts via e-mail or social media.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Ericsson Technology Review - Issue 2, 2018

  1. 1. ERICSSON TECHNOLOGY C H A R T I N G T H E F U T U R E O F I N N O V A T I O N | V O L U M E 9 7 I 2 0 1 8 – 0 2 TECHNOLOGYTRENDS AUGMENTINGTHE CONNECTEDSOCIETY FIXEDWIRELESS ACCESSINLTEAND5G DIGITAL CONNECTIVITY MARKETPLACES
  2. 2. FEATURE ARTICLE Five technology trends augmenting the Connected Society Ericsson’s CTO Erik Ekudden presents the five technology trends that are most relevant to the future development of network platforms capable of supporting the continuous evolution of industries and societies around the world. The use of machines to augment human intelligence will play a key role. 30 CONTENTS ✱ 08 OPEN, INTELLIGENT AND MODEL-DRIVEN: EVOLVING OSS Ericsson is leveraging the open-source approach to build open and intelligent operations support systems (OSS) that are designed to support autonomic networks and agile services. Our concept benefits from our ability to combine the open-source approach with our unique network and domain knowledge. 20 LOW LATENCY, HIGH FLEXIBILITY: VIRTUAL AQM Bufferbloat – the excessive buffering of packets – can cause significant delays in mobile networks for the application consuming the data, which is why it is so important for the network to be able to detect and properly manage it. Virtual Active Queue Management is an innovative approach to buffer management that is deployed as a centralized network function. 42 DIGITAL CONNECTIVITY MARKETPLACES TO ENRICH 5G AND IoT VALUE PROPOSITIONS The ongoing digitalization of industry represents a key growth opportunity for the telecom sector. The main challenge is to fully understand the needs of a new market. Our solution is to establish a platform model where capabilities from many providers can be effectively packaged and exposed in attractive ways to buyers from different industries. 52 COGNITIVE TECHNOLOGIES IN NETWORK AND BUSINESS AUTOMATION While the terms artificial intelligence and machine learning are often used synonymously, achieving automation that goes beyond low-level control of network parameters also requires machine reasoning. The best way to create the intelligent agents that are necessary to automate operational and business processes is to capitalize on the strengths of both technologies. 64 LEVERAGINGLTEAND5GNRNETWORKSFOR FIXEDWIRELESSACCESS The high-speed mobile broadband coverage that LTE and 5G New Radio (NR) enables has created a commercially attractive opportunity for operators to use fixed wireless access (FWA) to deliver broadband services to previously unserved homes and small and medium-sized enterprises around the world. Induced models Inferred knowledge Knowledge transfer Knowledge extraction Training examples Expert knowledge P F A Reasoning Planning Actions Machine learning (Numeric) Machine reasoning (Symbolic) 52 Consumer Producer Consumer Consu Digital connectivity platf Digital connectivity marke Cloud services market Ope se Bu se Application & enablement market Information market Connectivity services market 42 08 Dequeuing scheduler QFI and DSCP tagged IP packets GBR QOS flows Non-GBR QOS flows Flow queueing Flow hashing 20 64 30 64
  3. 3. #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 7 Ericsson Technology Review brings you insights into some of the key emerging innovations that are shaping the future of ict. Our aim is to encourage an open discussion about the potential, practicalities, and benefits of a wide range of technical developments, and provide insight into what the future has to offer. a d d r e s s Ericsson se-164 83 Stockholm, Sweden Phone: +46 8 719 00 00 p u b l i s h i n g All material and articles are published on the Ericsson Technology Review website: www.ericsson.com/ericsson-technology-review p u b l i s h e r Erik Ekudden e d i t o r Tanis Bestland (Nordic Morning) tanis.bestland@nordicmorning.com e d i t o r i a l b o a r d Håkan Andersson, Anders Rosengren, Mats Norin, Erik Westerberg, Magnus Buhrgard, Gunnar Thrysin, Håkan Olofsson, Dan Fahrman, Robert Skog, Patrik Roseen, Jonas Högberg, John Fornehed and Sara Kullman f e at u r e a r t i c l e Five technology trends augmenting the Connected Society by Erik Ekudden a r t d i r e c t o r Liselotte Eriksson (Nordic Morning) p r o d u c t i o n l e a d e r Susanna O’Grady (Nordic Morning) l ay o u t Liselotte Eriksson (Nordic Morning) i l l u s t r at i o n s Nordic Morning Ukraine c h i e f s u b e d i t o r Ian Nicholson (Nordic Morning) s u b e d i t o r s Paul Eade and Penny Schröder-Smith (Nordic Morning) issn: 0014-0171 Volume: 97, 2018 ■ technology development keeps getting faster and more interconnected, with new innovations appearing every day. As a result, we’re swiftly moving toward the realization of the “Augmented Connected Society” – a world characterized by ubiquitous internet access for all, self-learning robots and truly intuitive interaction between humans and machines. But how can our industry best prepare for this future? In my role as CTO, I have the challenging and exhilarating annual task of identifying the five technology trends of the future that are (or will be) most relevant to our industry. You can find my insights and reflections on this year’s trends on page 30. The augmentation of human intelligence is one of the key themes in this year’s trends article. Creating the highly automated environment that network operators and digital service providers will need in the future requires the support of intelligent agents that are able to work collaboratively. The two proofs of concept presented in the Cognitive Technologies article in this issue demonstrate how the combination of machine reasoning and machine learning techniques makes it possible to create intelligent agents that are able to learn from diverse inputs, and share or transfer experience between contexts. I believe that fixed wireless access (FWA) is likely to play an important role on our journey toward a world characterized by ubiquitous internet access for all. Already today, LTE and 5G New Radio are opening up significant commercial opportunities for operators to use FWA to bring the internet to many of the more than TRANSFORMING TO FIT THE FUTURE REALITY one billion households around the world that are still unconnected, as well as to many small and medium- sized businesses. The FWA article in this issue highlights the key principles for combined mobile broadband and FWA deployments, and presents a use case that illustrates the recommended deployment approach. In my discussions with people from different industries this past year, I’ve noticed a growing awareness that telecom infrastructure can bring significant value to the digital transformation of industries. However, many providers need help to understand the requirements of this new market and to create solutions for it. The Digital Connectivity Marketplaces article in this issue explains our concept for a platform model in which capabilities from many providers can be effectively packaged and exposed in attractive ways to buyers from different industries. There’s no doubt that operations support systems (OSS) have a critical role to play in the future of our industry. Check out the OSS article in this issue to learn about our future OSS concept, which is built on a solid implementation architecture that enables the use of industrydefined interfaces and open-source modules, as well as integration with full component compatibility. Finally, I want to encourage those of you who are concerned about bufferbloat to check out the Virtual Active Queue Management (vAQM) article in this issue. In our innovative concept, vAQM is centralized and applied per user and per flow, adapting to link rate fluctuation and considering the current bottleneck link bandwidth to the user. Our testing has shown that when vAQM is centralized upstream, rather than being deployed in the bottleneck nodes as it is in classic AQM, there is a substantial reduction in bufferbloat. I hope you get a lot of value out of this issue of Ericsson Technology Review, and that the articles can serve as the basis for stimulating future-focused discussions with your colleagues and business partners. If you would like to share a link to the whole magazine or to a specific article, you can find both PDF and HTML versions at https://www.ericsson. com/en/ericsson-technology-review ✱ EDITORIAL EDITORIAL ✱ THE AUGMENTATION OF HUMAN INTELLIGENCE IS ONE OF THE KEY THEMES IN THIS YEAR’S TRENDS ARTICLE ERIK EKUDDEN SENIOR VICE PRESIDENT AND GROUP CTO
  4. 4. 8 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 9 EVOLVING OPERATIONS SUPPORT SYSTEMS ✱✱ EVOLVING OPERATIONS SUPPORT SYSTEMS toolchainsthatsupportautomation.Microservice architecturesupportstheDevOpsparadigmand enableshighlyscalable,extendableandflexible systems. Ericsson’sapproachtoevolvingOSSisbasedon combiningthesetechnologiesandsolutionstocreate amodernOSSarchitecturethatisoptimizedtomeet therequirementsofserviceprovidersinthemidto longterm. OurOSSarchitectureprinciples ThekeyprinciplesguidingEricsson’sOSS architectureconceptarethatitisserviceoriented, thattheautomationisanalyticsandpolicydriven, andthatitusesvirtualizationandabstractionof networkfunctions.Byfollowingtheseprinciples, wecanevolveOSStobecomeprogrammableand abletoleveragetheinterfacesofferedbythe productiondomains. Terms and abbreviations AI–artificialintelligence|API–applicationprogramminginterface|BSS–businesssupportsystems| ETSI–EuropeanTelecommunicationsStandardsInstitute|MANO–ManagementandOrchestration| MI–machineintelligence|ML–machinelearning|NFV–NetworkFunctionsVirtualization|ONAP–Open NetworkAutomationPlatform|OSS–operationssupportsystems|PNF–PhysicalNetworkFunction| SDN–software-definednetworking|SLA–ServiceLevelAgreement|TOSCA–TopologyandOrchestration SpecificationforCloudApplications|UDM–UnifiedDataManagement|VNF–VirtualNetworkFunction MALGORZATA SVENSSON, MUNISH AGARWAL, STEPHEN TERRILL, JOHAN WALLIN The long-standing paradigm of bundling vendor and domain-specific management with network functions has been challenged in recent years by a new approach that is built on the concept of an open and model-driven platform. This new approach leverages open source and uses standardized interfaces to enable a horizontal management platform. ■Today’sserviceprovidersexpectrapidnetwork deployments,agileintroductionofnewservices andcost-efficientoperationandmanagement– requirementsthatareevenmoreimportantwhen planningforfuture5Gcapabilities.Recent technologytrendsmakeitpossibletoredefine traditionalsupportsystemsatthesametimeas theyenablegreaterefficiencyindevelopmentand operations. Forexample,model-drivenautomationeliminates theneedformanualinteractionbyexternalizing behavioralaspectsofspecificdomainsfrom applicationlogic.Real-timeandnear-real-time analyticsredefinetheimplementationandscopeof supportsystemfunctionslikeperformance,fault management,assuranceandoptimization,where analyticsmodelsactonthecollectedandcorrelated data,producinginsightsthatarefarricherinscope thantheoutcomeofthetraditionalfunctions. Machineintelligence(MI)algorithmsenable the transitionfromreactivetoproactivedecisionmaking. TheDevOpsparadigmsimplifiesdevelopment andoperationalprocesses.Itrequiresdevelopment anddeployment-friendlysoftwarearchitectureand The simplicity required to deploy and manage services in future network and infrastructure operations is driving the need to rethink traditional operations support systems (OSS). To unleash the potential of 5G, Network Functions Virtualization (NFV) and software-defined networking (SDN), future OSS need to be open, intelligent and able to support model-driven automation. OPEN, INTELLIGENT AND model-driven: evolvingOSS Definitions of key concepts 〉〉 Analytics is the discovery, interpretation and communication of meaningful patterns in data. It is the umbrella for AI, ML and MI. 〉〉 Cloud native is a term that describes the patterns of organizations, architectures and technologies that consistently, reliably and at scale take full advantage of the possibilities of the cloud to support cloud- oriented business models. 〉〉 Microservice is a small service supporting communication over network interfaces. Several such services constitute an application. Each microservice covers a limited and coherent functional scope and is independently manageable by fully automated life-cycle machinery. A microservice is small enough to be the responsibility of a single, small development team. 〉〉 OSS business layer is a conceptual layer of the OSS that guides the service provider through the operational support subset of its business process. 〉〉 OSS control layer is a layer of the OSS containing functionality that controls and manages production domains. 〉〉 Policy is a function that governs the choices in the behavior of a system. Policy can use a declarative or imperative approach. 〉〉 Programmability is the ability to externally influence the behavior of a system in a defined manner. 〉〉 Resource is a means to realize a service, and can be either physical or non-physical. 〉〉 Service is a means to deliver value to a customer or user; it should be understandable to a customer. Service is also a representation of the implementation aspects to deliver the value.
  5. 5. 10 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 11 EVOLVING OPERATIONS SUPPORT SYSTEMS ✱✱ EVOLVING OPERATIONS SUPPORT SYSTEMS Abstraction,programmability anddomain-specificextensions Networksarecomplex.Asthephysicalresources withinthemtransformintovirtualresourcessuch asVirtualNetworkFunctions(VNFs),virtual switches,virtualnetworksandvirtualcloud resources,theywillexposeonlywhatisnecessary andbemanagedinauniformway.Thisleadstoa logicalrepresentationofresourcesandnetworks, whichisreferredtoasabstraction.The capabilitiesofthesevirtualresourcesareexposed throughtheirinterfaces,whiletheirbehavioris influencedbytheseinterfaces.Thisisreferredto asprogrammability.Themethodologyof abstractionandprogrammabilityiscontinually andrecursivelyappliedatalllevelsoftheOSSand networks,fromtheresources’interfacestothe exposedservices. Abstractionandprogrammabilityarethemain prerequisitesforuniformmanagementof automation,easingtheshiftfrommanaging networkstomanagingautomation.Theyenable simplificationbyexposingonlytherequired capabilitiesofadomain,whilestillsupportingan efficientinteractionbetweenmanagementsystem scopes,andmanagementsystemsandnetworks. Regardlessofthegeneralizationofthebehavior betweenvariousproductiondomains,thedomain- specificaspectsremain.Someexamplesof domain-specificextensionsare:managingand optimizingradiofrequency,designingandhandling datacentercapacity,ordesigningL2/L3service overlay.Specificproductiondomaincompetencies andprocessesarerequiredtoprovidethenecessary capabilitiesintheseexamples,eventhoughthe capabilitiesarerealizedbythesameOSS. Fromautomaticmanagementto managementofautomation Thegoalofautomationistoprovideazero-touch network,whichmeansthatautomationmoves fromautomatingviascriptingwhatcanbedone manually,toanautonomicnetworkthattakescare ofitselfandhandlespreviouslyunforeseen situations.Thischangestheapproachto management,frommanagingnetworksmanually tomanagingautomation. Akeypartofmanagingautomationisthecontrol loopparadigm,asshowninFigure1.Thisisachieved bydesigningtheinsightsandpoliciesrequiredfora usecase,andthedecisionsthatneedtobemade. Insightsaretheoutcomewhendataisprocessedby analytics.Policiesrepresenttherulesgoverningthe decisionstobemadebythecontrolloop.Thedecisions areactuatedbyCOM(Control,Orchestration, Management),whichinteractswithproduction domainsbyrequestingactionssuchasupdate, configureorheal.Thecontrolloopsareimplemented bymultipleOSSfunctionsandcanactina hierarchicalway. Byleveraginganalytics,MIandpolicy,control loopsbecomeadaptiveandarecentraltoassuring andoptimizingthedeployedservicesandresources. MIandautonomicnetworks Automationcanbefurtherenhancedbyapplying MI,whichisthecombinationofmachinelearning (ML)andartificialintelligence(AI).MIadaptsto thesituationinanetworkandlearnstoprovidethe bestinsightsforagivennetworksituation. AsthelevelofMIincreases,theroleofpolicyshifts fromgovernanceofthedecisionstobemade,to ensuringthattheinsightsproducedfromMIare appropriate.Inthissense,MIisanenablerfor autonomicnetworks. AUTOMATIONCANBE FURTHERENHANCEDBY APPLYINGMI,WHICHISTHE COMBINATIONOFMACHINE LEARNINGANDARTIFICIAL INTELLIGENCE Figure 1 Analytics and policy-driven automation: closed control loop Analytics Policy Control orchestration management Insights Decisions Actions Data OSS business layer OSS control layer Production domains Design Onboard Test Deploy Operate Retire PNF PNF VNF VNF VNF VNF VNF Cloud system infrastructure Transport
  6. 6. 12 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 13 EVOLVING OPERATIONS SUPPORT SYSTEMS ✱✱ EVOLVING OPERATIONS SUPPORT SYSTEMS Figure 2 Model-driven OSS architecture Network slice life-cycle management Studio suite VNF scaling SLA assurance ...RAN optimization OSS business layer OSS control layer OSS information layer Network slice deployment model VNF scaling policy model RAN optimization policy model Design Onboard Test Deploy Operate Retire Orche- stration PolicyAnalyticsCatalog Inventory Mediation Model-drivenarchitecture Tosecureefficientandconsistentmanagement, thebehavioralaspectsofspecificdomainsare externalizedfromapplicationlogictomodels. Therearemodelsforconfiguration,orchestration, policyrulesandanalytics. Resourceandservicelife-cyclemanagementisa goodexampleofamodel-drivenapproach. AsshowninFigure2,themodelsdescribethe servicesandtheresources.Thesemodelsare designedandonboardedinastudiosuite.Theyare thenusedtotest,deploy,operateandretirethe servicesandtheresources.Themodelsarebased onstandardmodelingapproaches.Forexample, deploymentmodelsaredescribedusingHEAT (specifically,orchestrationinOpenStack[1])–and TOSCA[2].The“operate”processcancover activitiessuchasscale,upgrade,heal,relocate,stop andstart. Theconfigurationandinstrumentationofthe executingresourceandserviceinstancesrequire resourceandserviceviewsexpressedalsoas models.IETFYANG[3]hasemergedasthe mainstreammodelinglanguageforthispurpose. Policyrulesetsdefinethebehavioralaspectsof serviceassurance,resourceperformanceand scaling.Sincethemodelsarethewaytocapture businesslogicandintellectualproperty,theywill becomesellableobjectsthemselvesandbesubject tocommercialagreements. Figure 3 Ericsson’s OSS architecture Service and resource life-cycle management User service and resource life-cycle management Assurance and optimization Interaction management Studiosuite ServicesandAPIexposure Capacity management OSS catalogs and inventories Mediation and analytics Production domains BSS Knowledge management Security management Procurement Forecasting Development Marketplace Packet core Communi- cation services Cloud system infrastructure TransportRAN UDM Modulararchitectureand implementationcapabilities TheoptimalOSSofthefuturewillbebasedbothon theprinciplesoutlinedaboveandonmodular architectureandimplementationcapabilities. Studiosuite Figure3illustratesthestudiosuite,whichprovides asingle,coherentenvironmenttodesign,test, operate,controlandcoordinateallthebusiness, engineeringandoperationalprocessesofaservice provider.Thisisdonebyexpressingtheprocesses, interfaceextensions,policies,deploymentsand configurationsasmodelsandusingthemfor automationofalltheOSSfunctionsandthe managedproductiondomains.Thestudiosuite frameworkincludesbusiness,engineeringand operationalstudios. Thebusinessstudioistheenvironmentwhere businessandoperationalprocessesaremaintained. Thebusinessstudioautomaticallyorchestratesthe processmodelsbyinvokingoperationsonthe appropriatefunctionsintheOSSstack. Theengineeringstudioprovidesamodel-driven platformwheretheOSSfunctionsandtherelevant informationandinformationmodelsassociated withthosefunctionsareconstructedandextended. Itletsauser(humanormachine)realizevarious managementtasksbyextendingtheOSSfunctions withuse-casespecificmodelsforconfiguration,
  7. 7. 14 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 15 EVOLVING OPERATIONS SUPPORT SYSTEMS ✱✱ EVOLVING OPERATIONS SUPPORT SYSTEMS orchestration,dataandpolicies.Itprovidessoftware developmentkitsfordefiningtheextensionmodels andtheninstantiatestheneededcomponents, provisionsthecomponentswiththemodelsand wiresthemtogethertorealizethedesiredusecase. AnexamplecouldbeasetofMLfunctionsthat providescalablemachinerythatcanbeassignedto multipleusesbyapplyingdifferentconfigurationand datamodels. Theoperationalstudioofferscapabilitiesto manageresourcesandservicesthroughouttheirlife cycle.Resourcesrepresenttheinfrastructureandthe networksofproductiondomains.Servicesusethe infrastructureandnetworkstodelivervalueto customers.Oneofthefirstactivitiesintheresource andservicelifecycleistoconstructaspecificationof aresourceoraservice,thenassigntherequired policyrulesandconfigurationsdefiningthebehavior ofthenetwork,ortheinfrastructureortheservice. Oncetheyaredefined,thespecifications,policies andconfigurationsarestoredintheOSScatalogs anddistributedinrealtimetobeusedbytheOSS functions. Managingservicesandresources Serviceandresourcelife-cyclemanagement functionsprovideaprogrammablewayofmanaging theservicesandresourcesoftheproduction domainsbasedonspecifications,policiesand configurationsstoredinthecatalogs.Thecore capabilitiesareorchestrationandpolicyengines thatparsethedefinedmodelstorealizeallthelife- cyclestepsoftheservicesandresources.Thepolicy andorchestrationenginesautomatethenetwork managementbytakingovertheburdenoflow-level decisionmakingandexecutionoflife-cycle operationsfromhumanusers. Thesefunctionsaretriggeredbyeventsgenerated bysurroundingfunctionslikeinteraction management,assuranceandoptimization.They providebothclosedloopoptimizationcapabilities (wheredecisionmakingisdonebasedonpoliciesand MI)andanopenloopoptimizationfeature(wherea humanuserdecidesbasedonasetofautomatic recommendations).Thisdynamic,model-drivenand real-timeinteractioniswhatwillmakethisfuture OSSsolutiondifferentfromthetraditionalone. Wecanalreadyseetheneedforseamless interactionbetweenOSSsoftwaredevelopmentand operations,whichsuggeststhatthecoreDevOps capabilitiesofcontinuousintegration,delivery, releaseanddeploymenton-demandwillbean integralpartofthefutureOSS. OSSinventorieswillbeusedtostoreinstantiated resourcesandservices.Theytrackthepresence, capacityandconfigurationofresourcesandservices, aswellastheirrelationshipstootherresourcesand services.Theyprovidethedataforcapacity management,whichenablescapacityplanning, designandmonitoringovertime.Byusinganalytics incapacitymanagement,itispossibletomakeitwork inrealtimeandbepredictiveandself-learning. Thepurposeoftheuserserviceandresourcelife- cyclemanagementfunctionistomanageresourcesand servicesthathavebeenassignedtoauser,tenantor subscriber.Thiscapabilityalsosupportsuser,tenant andsubscriberprovisioning.Theclosedandopen controllooparchitecturepatternsapplyhereaswell. Theuserserviceandresourcelife-cyclemanagement functionhasthesamepropertiesastheserviceand resourcelife-cyclemanagementfunction. Mediationprovidesanopenandadaptable interfacetocollectdata,processeventsgeneratedby theproductiondomains,anddotherequired translation,transformationandfiltering.Italso providesasetofflexibleinterfacestoinvokelife-cycle operationsonthenetworksandinfrastructuresofthe productiondomains. Theassurancefunctionmakesitpossibletocheck thatresourcesandservicesbehaveastheyhavebeen designed,publishedandoffered.Assurancemakes useofanalyticstodiscoverinsightsaboutthe performanceofservicesandresources.Policy enginesarethentriggeredtoactontheinsightsto determineactionsthatensurethedesired performancelevelofferedonservicesandresources. OurOSSarchitectureisdesignedtouse interactionmanagement,servicesandapplication programminginterface(API)exposuretointeract withexternalfunctions.Afewexamplesofthose externalfunctionsaredevelopment,marketplaces, businesssupportsystems,advancedsecurityand knowledgemanagementsystems,procurementand forecasting. TheproductiondomainexampleshowninFigure 3includesaRAN,packetcoreandcommunication services,UnifiedDataManagement,cloudsystem infrastructure,fixedaccessandtransportnetworks. OurOSSmanagesmulti-vendornetworkand infrastructure. TheOSSfunctionsarebasedonanAPI-first approachthatmandatesthatanAPIbedefined beforethefunction-exposingservicescanbe implemented.Theapproachenablesparallel developmentandprovidestheabilitytoadaptto externalimplementations.TheAPImanagement frameworkfacilitatestheeffectiveuseofAPIsduring development,discoveryofinternalAPIendpointsat runtimeandefficientlycontrolledexposureofAPIs toexternalapplications. Cloudnative Cloudnativeisthenewcomputingparadigmthatis optimizedformodern,distributedsystems environments,capableofscalingtotensofthousands ofself-healing,multi-tenantnodes.Cloudnative systemsarecontainerpackaged,dynamically managedandmicroserviceoriented.TheEricsson OSSofthefuturewillbebasedoncloud-native architecturetoenablebetterreuse,simplified operationsandimprovedefficiency,agilityand maintainability. Themicroservicearchitectureenablesfine-grained reusebyhavingseveralsmallcomponentsinsteadofa singlelargeone.Bybeinglooselycoupledand backwardcompatible,theservicescanbedeveloped independently,enablingefficientDevOps.Italso allowseachservicetochoosethemostefficient developmentandruntimetechnologyforitspurpose. Themicroservicearchitecturemakesitpossibleto composevariousbusinesssolutions,thendeploythem automatically,significantlyreducingdeploymentcost andtimetomarket.Thearchitecturewillusethe cloud-nativeecosystemasaportabilitylayerto supportmultipledeploymentoptions. Security Supportingmultipledeploymentmodelsrequires multiplesecuritycapabilitieslikecredential management,securecommunication,encryption, authenticationandauthorization,whichcanbe selectedbasedonthedeployment.OurOSS architecturewillincludethecapabilitiesneededto supportthemultipledeploymentmodelsandto protectthemodels. Embracingindustryinitiatives DuetoNFVandprogrammablenetworkssuchas SDN,thereishugeindustryinterestinautomationto decreasecomplexityandachieverapidintroduction ofservicesandresources.Thishasresultedinseveral significantindustryinitiatives,mostnotably
  8. 8. 16 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 17 EVOLVING OPERATIONS SUPPORT SYSTEMS ✱✱ EVOLVING OPERATIONS SUPPORT SYSTEMS ETSI-MANO[4]andtheOpenNetwork AutomationPlatform(ONAP)[5]. Initiativesthatcallforstandardizationprovidean environmenttodevelopstrongconceptsandidentify importantinterfaces,whileopensourceprovides referenceimplementations,aswellas implementationinterfaces. ONAP Bybringingtogethertheresourcesofitsmembers, ONAPhasspedupthedevelopmentofaglobally sharedarchitectureandimplementationfor networkautomation.ONAPprovidesbothdesign timeandruntimereferenceimplementations, coveringservicedesignandcreation,catalog, inventory,orchestrationandcontrol,policiesand analytics.Italsoincludesservicemanagementand automationcapabilitiesforPhysical/Virtual NetworkFunctions(PNFs/VNFs),transportand cloudinfrastructure. Ericsson’smodularandmodel-drivenOSS approachiswellalignedwithONAPandextendsthe ONAPvisiontoautomationbeyondnetwork managementtoincludeinteractionmanagement, workforcemanagement,userserviceandresource life-cyclemanagement,advancedservice optimizationandassuranceasafewexamples. TointeractwithONAP,weareintheprocessof adoptingthenecessaryinterfacesandmodel definitionsinourproductportfolio(bothinOSSand networkfunctions).EricssoniscommittedtoONAP andheavilyengagedwithitacrossabroadrangeof topics.WeaimtousetheONAP-providedopen sourceimplementationsinaccordancewith Ericsson’sarchitectureprinciples. Managingmultipleproductiondomainsrequiresa model-drivenarchitecturebecauseachieving domain-specificbehaviorrequirestheapplicationof appropriatemodelsonacompatibleOSSplatform. Sinceamodel-drivenarchitectureisessentialto driveautomation,weareworkingtowardensuring thatourmodelsareusablewithinONAP. Conclusion AsbothanestablishedinfrastructureandOSS supplier,EricssonhasavisionforOSSinwhich autonomicnetworksenableintelligentand automatedserviceandresourcelife-cycle management.Weareusinganalytics,MI,policy andorchestrationtechnologiestocreatetheOSSof thefuture,andtooffergreaterefficiencyin developmentandoperations.WeapplyaDevOps paradigmtooursoftwaredevelopmenttokeepus closetocustomersandspeedupoursoftware delivery. Asourapproachismodel-driven,OSSbehavioris externalizedfromthemanagementplatformand materializedthroughacombinationofmodelsand configurations.Asaresult,thedevelopedmodelsare independentoftheexecutionplatform. OurOSSconceptismodularbydesignandfollows microservicearchitectureprinciplesthatmakeit possibletoreplacesoftwarecomponentswithopen sourceimplementations.Theconceptisbuiltona solidimplementationarchitecturethatenablesthe useofindustry-definedinterfacesandopensource modules,aswellasintegrationwithfullcomponent compatibility. Ericssonhasembracedopensource implementationsthroughactivecontributionsand usage,aswellasdrivingimportant standardizations.Wearecommittedtodriving industryalignmentthatwillhelpcreateahealthy ecosystem. Further reading 〉〉 Ericsson Technology Review, Technology trends driving innovation – five to watch, 2017, Ekudden, E: https://www.ericsson.com/en/publications/ericsson-technology-review/archive/2017/technology-trends-2017 〉〉 Ericsson Technology Review, Generating actionable insights from customer experience awareness, September 2016, Niemöller, J; Sarmonikas, G; Washington, N: https://www.ericsson.com/en/publications/ ericsson-technology-review/archive/2016/generating-actionable-insights-from-customer-experience-awareness 〉〉 EricssonTechnologyReview,DevOps:fuelingtheevolutiontoward5Gnetworks,April2017,Degirmenci, F;Dinsing,T;John,W;Mecklin,T;Meirosu,C;Opsenica,M:https://www.ericsson.com/en/publications/ericsson- technology-review/archive/2017/devops-fueling-the-evolution-toward-5g-networks 〉〉 Ericsson,Gearingupsupportsystemsforsoftware-definedandvirtualizednetworks,June2015: https://www. ericsson.com/en/news/2015/6/gearing-up-support-systems-for-software-defined-and-virtualized-networks- 〉〉 Ericsson,Zerotouchnetworkswithcloud-optimizednetworkapplications,Darula,M;Más,I: https://www. ericsson.com/assets/local/narratives/networks/documents/zero-touch-networks-with-the-5g-cloud-optimized- network-applications.pdf 〉〉 EricssonTechnologyReview,Architectureevolutionforautomationandnetworkprogrammability,November 2014,Angelin,L;Basilier,H;Cagenius,T;Más,I;Rune,G;Varga,B;Westerberg,E: https://www.ericsson.com/ en/publications/ericsson-technology-review/archive/2014/architecture-evolution-for-automation-and-network- programmability References 1. OpenStack,HeatOrchestrationTemplate(HOT)Guide,availableat:https://docs.openstack.org/heat/ocata/ template_guide/hot_guide.html 2. OASIS,OASISTopologyandOrchestrationSpecificationforCloudApplications(TOSCA)TC,availableat: https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=tosca 3. InternetEngineeringTaskForce(IETF),October2010,YANG–ADataModelingLanguagefortheNetwork ConfigurationProtocol(NETCONF),availableat:https://tools.ietf.org/html/rfc6020 4. ETSI,NetworkFunctionsVirtualization,availableat:http://www.etsi.org/technologies-clusters/technologies/nfv 5. ONAP, available at: https://www.onap.org/
  9. 9. 18 ERICSSON TECHNOLOGY REVIEW ✱ #02 2018 ✱ EVOLVING OPERATIONS SUPPORT SYSTEMS Malgorzata Svensson ◆ is an expert in OSS. She joined Ericsson in 1996 and has worked in various areas within research and development. For the past 10 years, her work has focused on architecture evolution. She has broad experience in business process, function and information modelling; information and cloud technologies; analytics; DevOps processes; and tool chains. She holds an M.Sc. in technology from the Silesian University of Technology in Gliwice, Poland. Munish Agarwal ◆ is a senior specialist in implementation architecture for OSS/BSS. He has 19 years of experience in telecommunications working with product development, common platforms and architecture evolution. He has hands-on experience with mediation, rating, order management and interactive voice response. Agarwal holds a Bachelor of Technology degree from the Indian Institute of Technology Kharagpur. Stephen Terrill ◆ has more than 20 years of experience working with telecommunications architecture, implementation and industry engagement. His work has included both architecture definition and posts within standardization organizations such as ETSI, 3GPP, ITU-T (ITU Telecommunication Standardization Sector) and IETF (Internet Engineering Task Force). In recent years, his work has focused on the automation and evolution of OSS, and he has been engaged in open source on ONAP’s Technical Steering Committee. Terrill holds an M.Sc., a B.E. (Hons.) and a B.Sc. from the University of Melbourne, Australia Johan Wallin ◆ is an expert in network management. He joined Ericsson in 1989 and has worked in several areas in RD in various positions covering design, product management and system management. He has broad experience of systems architecture in various mobile telephony systems including GSM, TDMA and CDMA. For the past 10 years, he has been working with network management architectures from an overall Ericsson perspective. He holds an M.Sc. in computer engineering and computer science from the Institute of Technology at Linköping University, Sweden. theauthors Theauthors wouldliketo acknowledge thecontribution theircolleagues JohnQuilty,Bo Åström,Ignacio Más,JanFriman, andJacoFourie madetothe writingofthis article.
  10. 10. 20 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 21 ✱ VIRTUAL ACTIVE QUEUE MANAGEMENT VIRTUAL ACTIVE QUEUE MANAGEMENT ✱ findtheidealsendrateforagivenend-to-endlink. Mobilenetworksareprovisionedwithlargebuffers tohandleburstytraffic,userfairnessandradio channelvariability.Eventhoughtheseoversized bufferspreventpacketloss,theoverallperformance degrades,asbufferbloathasmajorimpactson congestioncontrolatendpoints.Itmisleadsloss- basedcongestioncontrolalgorithms,resulting inpacketovershootingonthesenderside.Over- bufferingalsoresultsinlargejitterofend-to-end latency,whichpotentiallybringsintimeoutevents andeventuallyleadstothroughputdegradation. Applicationimpact Highlatencyandjittercausesignificantproblems forinteractiveapplicationssuchasVoIP,gaming andvideoconferencing.Webbrowsingisalso highlydependentonlatency,andpageloadtimes growlinearlyasround-triptime(RTT)increases. Videostreamingwithadaptivebitratethrougha continuoussequenceofsmallHTTPdownloads sufferswhenpacketsarelost.Undergood conditions,videoistransferredoverthenetwork atthesamerateasitisplayedback.Ifthenetwork can’tkeepupwithvideoplayback,theplayer usuallyswitchestoalowervideoraterepresentation andreasonablevideoqualitycanbeachieved withlessbandwidth. Themainchallengesinvideostreamingare packetlossandexcessivebuffering.Whenapacket islost,thevideoplayerstopsreceivingdatauntilthe lostpackethasbeenretransmitted,eventhoughdata packetscontinuetoarrive.Videoplayersnormally buffervideosegments,allowingthemtocontinue playingfromtheirbufferwhiledownloadingvideo segmentsthatmightinvolveretransmissions. Bufferingisnormallyusedforvideoondemandand timeshiftedvideobutislimitedforlivevideo. Terms and abbreviations ACK–acknowledgement|AMBR–aggregatemaximumbitrate|AQM–ActiveQueueManagement|CoDel –ControlledDelay|CPE–Customer-premisesequipment|DN–DataNetwork|DNS–DomainNameSystem |DSCP–DifferentiatedServicesCodePoint|DRR–DeficitRoundRobin|EPC–EvolvedPacketCore| EPS –EvolvedPacketSwitchedSystem|FQ–FlowQueue|FQ-CoDel–FlowQueueControlledDelay|GBR– guaranteedbitrate|IETF–InternetEngineeringTaskForce|PDNGW–publicdatanetworkgateway|PDU– protocoldataunit|QFI–QoSFlowIdentifier|QUIC–QuickUDPInternetConnections|RTT–round-triptime| SDF–ServiceDataFlow|SMF–SessionManagementFunction| SYN–synchronization|TCP–Transmission ControlProtocol|UE–userequipment|UPF–userplanefunction|vAQM–virtualActiveQueueManagement MARCUS IHLAR, ALA NAZARI, ROBERT SKOG To minimize the effects of bandwidth fluctuations in mobile networks, radio base stations are typically provisioned with large buffers. However, since excessive buffering of packets causes a large perceived delay for the application consuming the data, it is important for the network to detect and properly manage cases in which buffering becomes excessive. ■ActiveQueueManagement(AQM)isawell- establishedtechniqueformanagingthefrequent andsubstantialfluctuationsinbandwidthand delaythatarecommoninnetworks.Ithasbeen appliedtonetworkbuffersformanyyearswiththe purposeofdetectingandinformingdatasenders aboutincipienttrafficcongestion.TraditionalAQM tendstoworkwellinfixednetworks,wherebottle- necklinkbandwidthvarieslittleovershorttime durations.However,thedeploymentoftraditional AQMisnotwidespreadinmobilenetworks. Theproblemofbufferbloat Bufferbloatisanundesirablephenomenon prevalentinpacketnetworksinwhichexcessive bufferingresultsinunnecessarilylargeend-to- endlatency,jitterandthroughputdegradation.It occursduetolargequeuesthatabsorbahuge amountoftrafficinacongestedpathandusually causesalongqueuingdelay.Topreventpacket losses,networksnodesrelyonlargebuffers. Bufferbloatoccursmostlyatedgenetworknodes anddefeatsthebuilt-inTCPcongestionavoidance mechanism,whichreliesondroppedpacketsto Virtual Active Queue Management is an innovative approach to buffer management that does not require deployment in the network nodes where the buffering takes place. Instead, it is deployed as a centralized network function, which allows for simpler configuration and increased flexibility compared to traditional Active Queue Management deployments. LOW LATENCY, HIGH FLEXIBILITY Virtual AQM
  11. 11. 22 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 23 ✱ VIRTUAL ACTIVE QUEUE MANAGEMENT VIRTUAL ACTIVE QUEUE MANAGEMENT ✱ However,ifmultipleTCPpacketsarelost,TCP mayslowdownitssendingratetowellbelowthe datarateofthevideo,forcingtheplayerintothe rebufferingstate. Inshort,bufferbloatdegradesapplication performanceandnegativelyimpactstheuser experience.Thereisthereforeapressingneed tocontrollatencyandjitterinordertoprovide desirableQoStousers. Mitigation TraditionalAQMmitigatesbufferbloatby controllingqueuelengththroughdroppingor markingpacketsfromthebottleneckbufferwhen itbecomesfullorwhenthequeuingdelayexceeds athresholdvalue.AQMreactstocongestionby droppingpacketsandtherebysignalingtothe senderthatitshouldrestrictthesendingdatarate duetotheimminentoverload.Theshortwaittime alsoimprovestheresponsetimeforthetransport protocolerrorhandling. ControlledDelay ControlledDelay(CoDel)isanemergingAQM schemedesignedbytheIETF(Internet EngineeringTaskForce)[1].Itsgoalistocontain queuinglatencywhilemaximizingthethroughput. CoDelattemptstolimitbufferbloatandminimize latencyinsaturatednetworklinksbydistinguishing goodqueues(theonesthatemptyquickly)from badqueuesthatstaysaturatedandslow.CoDel featuresthreemajorinnovationsthatdistinguish itfrompriorAQMs.First,itusesthelocalminimum queueasameasureofthestanding/persistent queue.Second,itusesasinglestate-tracking variableoftheminimumdelaytoseewhere itisrelativetothestandingqueuedelay. Third,insteadofmeasuringqueuesizeinbytes orpackets,itmeasuresthepacketsojourntime inthequeueandusesitasametrictodetect incipientcongestion. CoDelworksasfollows:uponarrival,apacket isenqueuedifthereisroominthequeue.While enqueuing,atimestampisaddedtothepacket. Whendequeuing,thepacketsojourntimeis calculated.CoDeliseitherinadroppingstateor anon-droppingstate.Ifthepacketsojourntime remainsabovethetargetvalue(default5ms)fora specifiedintervaloftime(default100ms),CoDel entersthedroppingstateandstartsdropping packetsattheheadofqueue.WhileCoDelisin droppingstate,ifthepacketsojourntimefallsbelow thetargetorifthequeuedoesnothavesufficient packetstofilltheoutgoinglink,CoDelleavesthe droppingstate. TheCoDelparameterstargetandintervalare fixedparametersandtheirvalueshavebeenchosen basedontheobservationsfromseveralexperiments. Intervalisusedtoensurethatthemeasured minimumdelaydoesnotbecometoostale.Itshould besetatabouttheworst-caseRTTthroughthe bottlenecktogiveendpointssufficienttimetoreact. Thedefaultvalueis100ms.Thetargetparameteris theacceptablequeuedelayforthepackets,including capacitytocopewithtrafficspikes.Atargetof5ms workswellinmostsituations;lowervaluescan reducethethroughput.Ifthetargetismorethan 5ms,thereisminorornoimprovementinutilization. FlowQueueControlledDelay FlowQueueControlledDelay(FQ-CoDel)[2]isa standardAQMalgorithmthatensuresfairnessin CoDel.ItisusefulbecauseAQMalgorithms workingonsinglequeuesexacerbatethetendency ofunfairnessbetweentheflows.FQ-CoDelisa hybridschemethatclassifiesflowsintooneof multiplequeues,applyingCoDeloneachqueue andusingmodifiedDeficitRoundRobin(DRR) schedulingtosharelinkcapacitybetweenqueues. WithFQ-CoDel,itispossibletoreduce bottleneckdelaysbyseveralordersofmagnitude. ItalsoprovidesaccurateRTTestimatesto elephantflows,whileallowingpriorityaccessto shorterflowpackets[3]. IncombinationwithimprovementsinLinux networkstacks,CoDelandFQ-CoDelcan eliminatetheproblemofbufferbloatonEthernet, cable,digitalsubscriberlinesandfiber,resultingin networklatencyreductionsoftwotothreeordersof magnitude,aswellasimprovementsingoodput[4]. Inmobilenetworks,however,AQMneedsto adjusttolinkbandwidthfluctuationthatvariesfor differentusers.Thisisduetothefactthatbottleneck linkbandwidthcanvarysignificantlyinmobile networksevenovershorttimedurations,which causesqueuebuild-upthatcanaddsignificant delayswhenlinkcapacitydecreases. OurinnovativeconceptforvirtualAQM(vAQM) isappliedperuserandperflow,adaptingtolinkrate fluctuationandconsideringthecurrentbottleneck linkbandwidthtotheuser.Bottlenecklink bandwidthisdeterminedbycorrelatingthenumber ofbytesinflightwithRTTmeasurements.vAQMis similartoFQ-CoDelbutitusesthemeasuredtime- varyinglinkbandwidthtotheuserasthedequeuing rateinsteadofnon-varyingdeficit/quantumasin FQ-CoDel. VirtualAQM TraditionalAQMistypicallydesignedtooperate onqueuesdirectly.However,duetodeployment constraintsandotherfactors,itisnotalways straightforwardtodeploynewAQMsatthe networkedgeswherecongestionislikelytobe experienced.Onewayofgettingaroundthis problemistomovethebottlenecktoamore centralpointinthenetworkandapplytheAQM there.Inafixednetwork,itiseasytoachievethis byshapingtraffictoarateslightlybelowthe bottleneckpeakcapacity.Sincemobilenetworks exhibitvastlydifferentpropertiesoffluctuating bandwidth,amorerefinedapproachisrequired toapplycentralizedAQM. VirtualAQMconsistsoftwoseparatebut interrelatedcomponents:abottleneckmodeler andaqueuemanager.AseparatevAQMinstance isrequiredforeachbottleneck.InLTE,this correspondsroughlytoabearer(dedicatedor default),whilein5GitcorrespondstoaProtocol DataUnit(PDU)sessionoraQoSflow.Toeffectively modelthebottleneck,thevAQMmeasurestheRTT ofpacketsbetweenitselfandtheuserequipment (UE),aswellastheamountofinflightdata. VirtualAQMmaintainsfourvariables:PathRTT, AverageRTT,TargetQueueDelayandTarget RTT.PathRTTrepresentsthepropagationdelay betweenthevAQMandtheUEatthereceivingend, withoutqueuingdelay.Itiscalculatedbyfeeding RTTmeasurementsamplestoaminfilter.Duetothe varyingnatureofradionetworks,thePathRTTmust beoccasionallyrevalidated. AverageRTTisanexponentiallyweighted movingaverageofthemeasuredRTT,which representsthecurrentobserveddelay.Theuseof aweightedmovingaveragereducestheimpactof inherentnoiseintheRTTsignal,whichtypically arisesfrompathlayercharacteristicssuchasRadio LinkControllayerretransmissions. Acertainbottleneckqueuedelaymustbe toleratedtoensuresmoothdeliveryofdataoverthe radio.TargetQueueDelayrepresentsthedelaythat istoleratedontopofPathRTTbeforevAQMtakes somesortofaction.TargetQueueDelayisdefinedas afractionofPathRTT.ThesumofthePathRTTand theTargetQueueDelayistheTargetRTT. Thesefourvariablesareusedtodetermine whetherqueuingistakingplaceinnetwork bottlenecks.Theinformationcanbeusedintwo distinctways:eitherasdirectinputtoanAQM algorithmorasinputtoabitrateshaper. OURINNOVATIVE CONCEPTFORVIRTUAL AQMISAPPLIEDPERUSER ANDPERFLOW WITHFQ-CODEL, ITISPOSSIBLETOREDUCE BOTTLENECKDELAYSBY SEVERALORDERSOF MAGNITUDE
  12. 12. 24 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 25 ✱ VIRTUAL ACTIVE QUEUE MANAGEMENT VIRTUAL ACTIVE QUEUE MANAGEMENT ✱ vAQMwithdynamicbitrateadjustment InvAQMwithdynamicbitrateadjustment,the trafficshaperdynamicallyadjustsitsdequeuing ratebasedonthemeasuredsendingrateandthe relationbetweenAverageRTTandPathRTT.The rateofincomingdataismeasuredandappliedtoa maxfilter,whiletheaverageRTTislessthanor equaltoTargetRTT.WhenAverageRTTexceeds theTargetRTT,theshaperwillsetthedequeue ratetothelargestbitrateobservedwhileAverage RTTwasbelowtarget.Furthermore,the differencebetweenAverageRTTandTargetRTT isusedasanadditionalboundingfactoronthe dequeuerate. Internalqueueswillstarttoformifpacketsarrive atthevAQMatahigherratethanthedequeuerate. Insuchcases,astandardAQMalgorithmsuchas FQ-CoDelcanbeappliedtothosequeues.Path RTTcanbeusedasinputtotheAQMalgorithmto determinetheCoDelintervalandothervariables. Figure1demonstrateshowvAQMassigns differentflowstodifferentqueuesmanagedby CoDelanddequeueswithDRRusingthecurrent ratetotheUE. vAQMwithoutbitrateadjustment ThevAQMwithoutbitrateadjustmentapproach sacrificestheflowisolationthatisachievedbyFQ forreducedimplementationandruntime complexity.ThisapproachusestheRTT measurementsasdirectinputtoanAQM algorithmwithoutapplyingtheshapingstep. TheAQMmakesitsdrop/markdecisionsbased ontherelationbetweenAverageRTT,PathRTT andtheTargetQueueDelay.IftheAverageRTT continuouslyexceedstheTargetRTTforan intervalamountoftime,theAQMentersa droppingstate,likethatinCoDel.Theintervalis determinedasafunctionofPathRTT.Ifmultiple flowsaretraversingthesamebottleneck,adrop willoccurontheflowthatisdeterminedtobethe mostsignificantcontributorofqueuedelay.Thisis effectivelydeterminedbycomparingtheflight sizesoftherespectiveflows. Handlingofnon-responsiveflows ThebasicprincipleofAQMisthatitprovides feedbacktothecongestioncontrolmechanismat thesendingendpoint.However,thereisa possibilitythatsenderswillnotreactasexpected tothesignalsthattheAQMprovides.ThevAQM candetectsuchflowsandapplyflow-specific shapingtoalevelwheretheflowdoesnotgenerate congestioninthebottleneck.Thisfeaturemakesit safetodeployandexperimentwithnew congestioncontrolalgorithms,asitreducesthe potentialharmtothenetworkandotherflows sharingthesamebottleneck. RateandRTTmeasurement VirtualAQMreliesonpassivemeasurements ofthroughputandRTTtodetectcongestion. TCPisunencryptedatthetransportleveland canbemeasuredtriviallybystoringsequence numbersandobservingacknowledgements (ACKs)onthereversepath. Whenthetransportlayernetworkprotocol QUIC(QuickUDPInternetConnections)isused, acryptographicenvelopehidesmosttransport mechanisms.Mostnotably,theACKframeis completelyhiddenfromon-pathobservers. Aspecificmechanismthatallowspassiveon-path measurementofRTTisbeingproposedinQUIC standardization[5].Thismechanismmakesuseof asinglebitintheQUICshortheader,thevalueof thebitcycleswithafrequencycorrespondingto theRTTbetweentwocommunicationsendpoints. However,atthetimeofwriting,neitherthe QUICprotocolnorthedescribedmechanism isafinishedstandard. TheadditionalbenefitsofvAQM ThebenefitsoftraditionalAQMarewellknown, namely:fastdeliveryofshort-livedflowsand increasedperformanceoflatency-critical applications.Thereisanincreasedresponsiveness toshort-livedflowsduetoschedulingatflowlevel andminimalqueuedelay.Forexample,Domain NameSystem(DNS)responsesandTCPSYN/ ACKpacketsdonothavetoshareaqueuewith aTCPdownloadofbulkdata. Inrecentyears,ithasbecomeincreasingly commonfordifferenttypesofapplicationsto shareasinglebottleneckbuffer.Latency-critical applicationsthatruninparallelwithbulkdownload applicationssuchasvideostreamingorfile downloadstypicallyexperiencesevereperformance degradation.Byensuringthatthestandingqueuesat thebottleneckaresmallornon-existent,largeflows willnotdegradelatency. InadditiontothebenefitsoftraditionalAQM, vAQMisadaptivetolinkratefluctuationand considersthecurrentbottlenecklinkbandwidthto theuser.Italsoprovidesthebenefitsofcentralized AQMandmodularityforincreasedflexibility. vAQMisabletosimplifythedeploymentofAQM inanoperatornetworkbecauseembeddingitina single(aggregation)nodeissufficienttoenablethe managementoftheentirenetworkdownstream fromthepointofdeployment.Thismeansthat legacyroutersandradioschedulersdonotneed tobeupgradedtoincludeAQM.WhilevAQM currentlyusesaschemeresemblingFQ-CoDel,it isbuilttosupportpluggableAQMmodules,sothat advancesinthefieldofAQMandcongestioncontrol canhaverapidimpactinlivenetworks. vAQMtesting WehavetestedourvAQMconceptonthe EricssonlabLTEnetwork,embeddingitina multi-serviceproxythatinspectstrafficto determineoptimalbottlenecklinkcapacitytoUEs. Figure 1: vAQM with dynamic bitrate adjustment and flow separation DRR scheduling Flows hashed into separate queues Dequeue rate determined by measurement of RTT and flight size vAQM Different flows ... AQM algorithm per queue Origin server Origin server Origin server
  13. 13. 26 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 27 ✱ VIRTUAL ACTIVE QUEUE MANAGEMENT VIRTUAL ACTIVE QUEUE MANAGEMENT ✱ Weconductedthetestingaccordingtotwo multiusercellscenarios:atfull-signalstrength andatweak-signalstrength. Thetestapplicationconsistedofloadinga GoogleDrivepageanddownloadinga16MBimage. Weusedthreemobiledevicestogeneratethe backgroundtrafficforthemultiusercell,with eachofthemdownloading1GBfiles,sothat wecouldtestlatencyunderloadformultiplestreams. ThepagewasfetchedusingbothTCPandQUIC, andtheeNodeBwasconfiguredwithadrop-tail queue.Theserverwasconfiguredtousethe CUBICcongestioncontrolalgorithmforboth TCPandQUIC. Weranthetestsforeachscenariointhree differentways: 〉〉 TCPwithvAQMdisabled 〉〉 QUICwithvAQMdisabled 〉〉 TCPwithvAQMenabled Theresults,showninFigure2,demonstratea reductionofthemedianRTTbyroughlythree timeswithvAQMenabledinthestrong-signal scenario.Intheweak-signalscenario,themedian RTTwithvAQMisnotmuchdifferentfromthe strong-signalscenario,whereasthetrafficwithout AQMdisplaysincreasedlatencywitha95th percentileRTTofapproximatelyonesecond forbothQUICandTCPtraffic. vAQMinpacketcore Figure3illustratesvAQMintwoscenarios:Packet DataNetworkGateway(PDNGW)ofEvolved PacketCore(EPC)in4GandintheUserPlane Function(UPF)of5Gcore. In5G,vAQMwillbedeployedasanetworkfunction bytheUPFofthe5Gcore.Itwillbeconfiguredper PDUsession,perSDF(ServiceDataFlow)/QoS flow–thatis,theguaranteedbitrate(GBR)ornon- GBRQoSflow.Thenon-GBRQoSflowsofone PDUsessionwillbeaggregatedandhandledby onevAQM,whereaseachGBRQoSflowwillbe allocatedavAQMwithadedicatedvirtualbuffer. TheSessionManagementFunction(SMF) willconfigurevAQManditsvirtualbuffersforthe QoSflowsduringthePDUsessionestablishment/ modificationphase.Configurationwilloccurwhen SMFprovidestheUPFwithaQoSEnforcement RulethatcontainsinformationrelatedtoQoS enforcementoftrafficincludinganSDFtemplate (inotherwords,packetfiltersets),theQoS-related informationincludingGBRandmaximumbitrate, andthecorrespondingpacketmarkinginformation –inotherwords,theQoSFlowIdentifier(QFI), thetransport-level-packet-markingvalue.vAQM isself-learning,makingitfullyautomatedwithout anyparametrization.Whensendingthepackets onthePDUsessiontotheUE,vAQMwillread outthepacketsfromthequeues,recalculatethe ratetotheUEandadjustthebottleneck.Incases Figure 2: vAQM in full signal strength testing scenario (top) and in weak signal strength scenario (bottom) TCP QUIC Median 95th percentile TCP vAQM RTT (ms) 200 400 600 800 1000 0 TCP QUIC Median 95th percentile TCP vAQM RTT (ms) 200 400 600 800 1000 0 Figure 3: vAQM at PDN GW of EPC and at UPF of 5GC 4G UE PDN EPC 5GC Data network Local Data Network 4G RAN Serving GW PDN GW vAQM 5G RAN 4G RAN 5G RAN Wi-Fi access 5G UE 4G UE 5G UE Wi-Fi UE CPE Fixed access vAQM at edge UPF per session and per QoS flow for local breakout PDU session vAQM at centralized UPF per PDU session and per QoS-flow vAQM per PDN connection and per EPS bearer UPF vAQM UPF vAQM
  14. 14. 28 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 29 ✱ VIRTUAL ACTIVE QUEUE MANAGEMENT VIRTUAL ACTIVE QUEUE MANAGEMENT ✱ wherethePDUsessionhasalocalbreakout,vAQM isconfiguredattheedgeUPFservingthelocal breakout;inotherwords,attheuplinkclassifieror branchingpoint. Indownlink,asshowninFigure4,theUPFbinds theIPflowsofthePDUsessiontoSDFs/QoSflows basedontheSDFtemplatesusingtheIPpacketfilter set(forexample,5tuples).TheUPFthenenforces thePDUsessionaggregatemaximumbitrate (AMBR)acrossallnon-GBRQoSflowsofthePDU session.ItalsoenforcestheGBRfortheGBRQoS flows.TheUPFfinallytagsthepacketswithQoS FlowIdentities(QFIs)andhandsoverthepacketsto vAQM,whichisresponsiblefortheQoSflow. ThedifferentvAQMinstancesassociated withQoSflowscancontainimplementations ofcompletelydifferentAQMalgorithmsorbe configuredwithdifferentdefaultparameters. Conclusion Theongoingevolutionofnetworksandapplications isincreasingtherequirementsonend-to-end performance.Specifically,itisimportanttobe abletoservedataathighrateswithoutcausing unnecessarydelayduetobloatedbuffers.Our testinghasshownitispossibletomitigate bufferbloatsubstantiallythroughtheuseofavirtual formofAQM(vAQM)thatiscentralizedupstream ratherthanbeingdeployedinthebottleneck nodes,asitisinclassicAQM.Thisapproachgreatly simplifiesthedeploymentandconfigurationof AQMinmobilenetworks,anditensuresconsistent behavioracrossbottlenecknodes,whichinmany casesaresuppliedbyamultitudeofvendors.Inlight ofthesebenefits,vAQMwillbeanimportantuser planefunctionin5Gcore. Further reading 〉〉 Ending the Anomaly: Achieving Low Latency and Airtime Fairness in WiFi (conference paper), Toke Høiland-Jørgensen et al., USENIX ATC (2017), available at: https://www.usenix.org/system/files/conference/ atc17/atc17-hoiland-jorgensen.pdf 〉〉 Ending the Anomaly: Achieving Low Latency and Airtime Fairness in WiFi (slide set), Toke Høiland- Jørgensen et al., USENIX ATC (2017), available at: https://datatracker.ietf.org/meeting/99/materials/slides- 99-iccrg-iccrg-presentation-1/ References 1. ControlledDelayActiveQueueManagement,RFC8289,January2018,availableat: https://tools.ietf.org/html/rfc8289 2. TheFlowQueueCoDelPacketSchedulerandActiveQueueManagementAlgorithm,RFC8290,January 2018,availableat:https://www.ietf.org/mail-archive/web/ietf-announce/current/msg17315.html 3. CoDelOverview,availableat:https://www.bufferbloat.net/projects/codel/wiki 4. MakingWiFiFast(blog),DaveTäht,availableat:http://blog.cerowrt.org/post/make-wifi-fast/ 5. TheAdditionofaSpinBittotheQUICTransportProtocol,IETFDatatracker(2017),availableat: https://datatracker.ietf.org/doc/draft-trammell-quic-spin Figure 4: vAQM for the PDU session QoS flows at the UPF Dequeuing scheduler QFI and DSCP tagged IP packets GBR QOS flows Non-GBR QOS flows DN Flow queueing Flow hashing UE-AMBR policing on non-GBR SDFs MBR policing (MBR for GBR SDFs and session AMBR for non-GBR SDFs) Packet filter sets PDU session IP flows PF1 PF2 PF3 PF4 PF5 PF6 SDF1SDF1 SDF2SDF2 SDFSDF3 Marcus Ihlar ◆ is a system developer in the field of traffic optimization and media delivery. He joined Ericsson in 2013 and initially did research on information- centric networking. Since 2014, he has been working on transport layer optimization and media delivery. Ihlar also participates actively in IETF standardization in the Transport Area Working Group. He holds a B.Sc. in computer science from Stockholm University in Sweden. Ala Nazari ◆ is a media delivery architecture expert. He joined Ericsson in 1998 as a specialist in datacom, working with GPRS, 3G and IP transport. More recently, he has worked as a senior solution architect and engagement manager. Prior to joining Ericsson, Nazari spent several years at Televerket Radio working with mobile and fixed broadband and transport. He holds an M.Sc. in computer science from Uppsala University in Sweden Robert Skog ◆ is a senior expert in the field of media delivery. After earning an M.Sc. in electrical engineering from KTH Royal Institute of Technology in Stockholm in 1989, he joined Ericsson’s two-year trainee program for system engineers. Since then, he has mainly been working in the service layer and media delivery areas, on everything from the first WAP solutions to today’s advanced user plane solutions. In 2005, Skog won Ericsson’s prestigious Inventor of the Year Award. theauthors
  15. 15. #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 201830 31 ✱ FEATUREARTICLE FEATUREARTICLE ✱ Rapid advancements in the use of machines to augment human intelligence are creating a new reality in which we increasingly interact with robots and intelligent agents in our daily lives, both privately and professionally. The list of examples is long, but a few of the most common applications today are found in education, health care, maintenance and gaming. My vision of the future network is an intelligent platform that enables this new reality by supporting the digitalization of industries and society. This network platform consists of three main areas: 5G access, automation through agility, and a distributed cloud. A set of intelligent network applications and features is key to hiding complexity from the network’s users, regardless of whether they are humans or machines. The ability to transfer human skills in real time to other humans and machines located all around the world has the potential to enable massive efficiency gains. Autonomous operation by machines with self-learning capabilities offers the additional advantage of continuous performance and quality enhancements. High levels of cooperation and trust between humans and machines are essential. Building and maintaining trust will require decision transparency, high availability, data integrity and clear communication of intentions. The network platform I envision will deliver truly intuitive interaction between humans and machines. In my view, there are five key technology trends that will play critical roles in achieving the vision: by erik ekudden, cto #1 The realization of zero touch #2 The emergence of the Internet of Skills #3 Highly adaptable, cyber-physical systems #4 Trust technologies for security assurance #5 Ubiquitous, high-capacity radio five technology trends augmenting the connected society
  16. 16. ✱ FEATUREARTICLE the realization of zero-touch #1 TH E Z ERO -TO U CH networks of the future will be characterized by the fact that they require no human intervention other than high-level declarative and implementation- independent intents. On the road to zero touch both humans and machines will learn from their interactions. This will build trust and enable the machines to adjust to human intention. Computeandintelligencewillexistinthe device,inthecloudandinvariousplacesin thenetwork.Thenetworkwillautomatically computetheimperativeactionstofulfill givenintentsthroughaclosedloop operation.Today’scomplexnetworks aredesignedforoperationbyhumansand thecomplexityisexpectedtoincrease.As machinelearningandartificialintelligence continuetodevelop,efficientlyintegrating learningandreasoning,thecompetence levelofmachineintelligencewillgrow. AUGMENTATIONOFHUMAN INTELLIGENCE Therealizationofzerotouchisaniterative processinwhichmachinesandhumans collaboratereciprocally.Machinesbuild intelligencethroughcontinuouslearning andhumansareassistedbymachines intheirdecision-makingprocesses. Inthiscollaboration,themachines gatherknowledgefromhumansandthe environmentinordertobuildmodels ofthereality.Structuredknowledgeis createdfromunstructureddatawiththe supportofsemanticwebtechnologies, suchasontologies.Themodelsare createdandevolvedwithnewknowledge tomakeinformedpredictionsandenhance automateddecisionmaking. Tomaximizehumantrustandimprove decisionquality,thereisaneedfor transparencyinthemachine-driven decision-makingprocess.Itispossible togaininsightsintoamachine’sdecision processbyanalyzingitsinternalmodeland determininghowthatmodelsupported particulardecisions.Thisservesasabasis forgeneratingexplanationsthathumans canunderstand.Humanscanalsoevaluate decisionsandprovidefeedbacktothe machinetofurtherimprovethelearning 32 33 process.Theinteractionbetweenhumans andmachinesoccursusingnatural languageprocessingaswellassyntactical andsemanticanalysis. ROBOTSANDAGENTSCOLLABORATE WITHHUMANS Inacollaborativescenario,arobotwill beabletoanticipatehumanintentions andrespondproactively.Forexample,an assembly-linerobotwouldautomatically adaptitspacetotheskillsofitshuman coworkers.Suchinteractionsrequire theintroductionofexplainableartificial intelligencetocultivatehumantrustin robots.Robotswillworkalongsidehumans toaidandtolearn.Robotscanalsointeract withotherdigitalizedcomponentsor digitaltwinstoreceivedirectfeedback. However,furtheradvancementsinrobot designandmanufacturingwillbeneeded toimprovetheirdexterity. Asoftwareagentinazero-touch networkactsinthesamewayasahuman operator.Theagentshouldbeabletolearn theroleinrealtime,aswellasthepattern andtheproperactionsforagiventask. Inparticular,itshouldbeabletohandle awiderangeofrandomvariationsinthe task,includingcontaminateddatafrom therealworldthatoriginatesfrom incidentsandmistakes.Theseagents willlearnthroughacombinationof reinforcementlearning(wheretheagent continuallyreceivesfeedbackfrom theenvironment)andsupervised/ unsupervisedlearning(suchas classification,regressionandclustering) frommultipledatastreams.Anagentcan bepre-trainedinasafeenvironment, aswithinadigitaltwin,andtransferred toalivesystem.Domainknowledgeisa keysuccessfactorwhenapplyingagents tocomplextasks. Techniquessuchasneuralnetworks offersignificantadvantagesinlearning patterns,butthecurrentapproachistoo rigid.Differentialplasticityisanother techniquethatlookspromising. #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 FEATUREARTICLE ✱
  17. 17. the emergence of the internet ofskills #2 TH E I NTERN E T O F S KI LL S allows humans to interact in real time over great distances – both with each other and with machines – and have similar sensory experiences to those that they experience locally. Current application examples include remote interactive teaching and remote repair services. A fully immersive Internet of Skills will become reality through a combination of machine interaction methods and extended communication capabilities. Internet of Skills-based systems are characterized by the interplay of various devices with sensing, processing and actuation capabilities near the user. Currentsystemslacktheaudio,visual, hapticandtelecommunicationcapabilities necessarytoprovideafullyrealistic experience.ToenabletheInternetofSkills, theinterplaybetweenhumansandrobots, andbetweenhumansandvirtualcontent, isofparticularimportance.Bothindustry andconsumersareshowinggreatinterest andopennessinusingthesenew capabilities. HUMANSKILLSDELIVERED WITHOUTBOUNDARIES Anauthenticvisualexperiencerequires real-time3Dvideocapturing,processing andrendering.Thesecapabilitiesmakeit possibletocreatea3Drepresentationof thecapturedworldandprovidethe experienceofbeingimmersedinaremote orvirtualenvironment.Whiletoday’suser devicesdon’tyetprovidethenecessary resolution,fieldofview,depthperception, wearabilityandpositioningcapabilities, thequalityandperformanceofthese technologycomponentsissteadily improving. Spatialmicrophoneswillbeusedto separateindividualsoundsourcesinthe spacedomain.Thisimpliesthattherewill beanincreasedamountofdataneededto capturetheaudiospatialaspects.Spatial audiorenderingperformanceisverymuch tiedtoefficienthead-relatedfiltermodels. Newformatsforexchangingspatialaudio streamshavebeenspecifiedand compressiontechniquesarebeing developed. Hapticcomponentsallowuserstofeel shapes,textures,motionandforcesin3D. Deviceswillalsotrackthemotionsand forcesappliedbytheuserduring interaction.Withcurrenttechnologiesthe userneedstowearorholdaphysical device,butfutureultrasoundbasedhaptic deviceswillofferacontact-freesolution. Standardizationeffortsforhaptic communicationwillallowforaquicker adoptionofhapticcapabilities. INSTANTINTERACTION ANDCOMMUNICATION Communicationbetweenhumansand machineswillbecomemorenatural,tothe pointthatitiscomparabletointerpersonal interaction.Naturaluserinterfacessuchas voiceandgesturewillbecommonplace. Theuseofvision-basedsensorswillallow foranintuitivetypeofinteraction.Tobetter understandhuman-machineinteraction thereisaneedtoevolvetheunderstanding ofkinesiology,ergonomics,cognitive scienceandsociology,andtoincorporate themintoalgorithmsandindustrialdesign. Thiswouldmakeiteasiertoconveya machine’sintentbeforeitinitiatesactions, forexample. Largevolumesof3Dvisualinformation imposehighnetworkcapacitydemands, makingultra-lowlatencyandhigh bandwidthcommunicationtechnologies essential.Enablingthebestuser experiencerequirestheuseofnetwork edgecomputerstoprocessthelarge volumesof3Dvisual,audioandhaptic information.Thissetupsavesdevice batterylifetimeandreducesheat dissipation,aswellasreducingnetwork load. ✱ FEATUREARTICLE 34 35#02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 FEATUREARTICLE ✱
  18. 18. highly adaptable cyber-physical systems #3 ✱ FEATUREARTICLE A CYB ER- PHYS I CAL system is a composite of several systems of varying natures that will soon be present in all industry sectors. It is a self-organizing expert system created by the combination of model of models, dynamic interaction between models and deterministic communication. A cyber-physical system presents a concise and comprehensible system overview that humans can understand and act upon. Themainpurposeofacyber-physical systemistocontrolaphysicalprocessand usefeedbacktoadapttonewconditions inrealtime.Itbuildsupontheintegrationof computing,networkingandphysical processes.Anexampleofacyber-physical systemisasmartfactorywhere mechanicalsystems,robots,rawmaterials, andproductscommunicateandinteract. Thisinteractionenablesmachine intelligencetoperformmonitoringand controlofoperationsatallplantlevels. SYNERGISTICINTEGRATIONOF COMPUTATION,NETWORKINGAND PHYSICALPROCESSES Themainchallengeistheorchestration ofthenetworkedcomputationalresources formanyinterworkingphysicalsystems withdifferentlevelsofcomplexity.Cyber- physicalsystemsaretransformingtheway peopleinteractwithengineeredsystems, justastheinternethastransformedthe waypeopleinteractwithinformation. Humanswillassumeresponsibility onawideroperatingscale,supervising theoperationofthemostlyautomated andself-organizingprocess. Acyber-physicalsystemcontains differentheterogeneouselementssuchas mechanical,electrical,electromechanical, controlsoftware,communicationnetwork andhuman-machineinterfaces.Itisa challengetounderstandtheinteractionof thephysical,cyberandhumanworlds. Systemmodelswilldefinetheevolutionof eachsystemstateintime.Anoverarching modelwillbeneededtointegrateallthe respectivesystemmodelswhilecontem- platingallpossibledynamicinteractions. Thisimpliesacontrolprogramthatdelivers adeterministicbehaviortoeach subsystem.Currentdesigntoolsneedto beupgradedtoconsidertheinteractions betweenthevarioussystems,their interfacesandabstractions. MODELOFMODELSCREATES THECYBER-PHYSICALSYSTEM Withinthecyber-physicalsystemall systemdynamicsneedtobeconsidered throughamodelthatinteractswithallthe sub-models.Manyfactorsimpactthe dynamicsoftheinteractionsbetweenthe systems,includinglatency,bandwidthand reliability.Forawirelessnetwork,factors suchasthedevicelocation,thepropagation conditionsandthetrafficloadchangeover time.Thismeansthatnetworksneedtobe modeledinordertobeintegratedinthe modelofmodels. Thetimeittakestoperformataskmay becriticaltoenableacorrectlyfunctioning system.Physicalprocessesare compositionsofmanythingsoccurringin parallel.Amodeloftimethatisconsistent withtherealitiesoftimemeasurementand timesynchronizationneedstobe standardizedacrossallmodels. EXAMPLE:INDUSTRY4.0 Thefactoryofthefutureimplementsthe conceptofIndustry4.0,whichincludesthe transformationfrommassproductionto masscustomization.Thisvisionwillbe realizedthroughlarge-scaleindustrial automationtogetherwiththedigitalization ofmanufacturingprocesses. Humansassumetheroleofsupervising theoperationoftheautomatedandself- organizingproductionprocess.Inthis contextitwillbepossibletorecognizeall thesystemmodelsthatneedtointeract: • Physicalandroboticsystemssuchas conveyors,roboticarmsandautomated guidedvehicles • Controlsystemssuchasrobot controllersandprogrammablelogic controllersforproduction • Softwaresystemstomanageallthe operations • Bigdataandanalytics-based softwaresystems • Electricalsystemstopower machinesandrobots • Communicationnetworks • Sensorsanddevices. Themastermodelconsistsofandinteracts withallthelistedprocessesabove,resulting intherealizationofthefinalproduct. 36 37#02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 FEATUREARTICLE ✱
  19. 19. trust technologies for security assurance #4 TRUS T TECH N O LO G I E S will provide mechanisms to protect networks and offer security assurance to both humans and machines. Artificial intelligence and machine learning are needed to manage the complexity and variety of security threats and the dynamics of networks. Rapidly emerging confidential computing – together with possible future multi-party computation – will facilitate secure cloud processing of private and confidential data. Performance and security demands are driving the development of algorithms and protocols for identities. Theuseofcloudtechnologiescontinuesto grow.Billionsofnewdeviceswithdifferent capabilitiesandcharacteristicswillallbe connectedtothecloud.Manyofthemare physicallyaccessibleandthusexposed andvulnerabletoattackortobeing misusedasinstrumentsofattack.Digital identitiesareneededtoproveownership ofdataandtoensurethatservicesonly connecttoothertrustworthyservices. Flexibleanddynamicauditingand complianceverificationarerequiredto handlenewthreats.Furthermore,thereisa needforautomatedprotectionthatadapts tooperatingmodesandperformsanalytics onthesysteminoperation. PROTECTIONDRIVENBY ARTIFICIALINTELLIGENCE Artificialintelligence,machinelearningand automationarebecomingimportanttools forsecurity.Machinelearningaddresses areassuchasthreatdetectionanduser behavioranalytics.Artificialintelligence assistssecurityanalystsbycollectingand siftingthroughthreatinformationtofind relevantinformationandcomputing responses.However,thereisaneedto addressthecurrentlackofopen benchmarkstodeterminethematurityof thetechnologyandpermitcomparisonof products. Whilethecurrenttrendistocentralize dataandcomputation,security applicationsfortheInternetofThingsand futurenetworkswillrequiremore distributedandhierarchicalapproachesto supportbothfastlocaldecisionsand slowerglobaldecisionsthatinfluencelocal policies. CONFIDENTIALCOMPUTING TOBUILDTRUST Confidentialcomputingusesthefeatures ofenclaves–trustedexecution environmentsandrootoftrust technologies.Codeanddataiskept confidentialandintegrityprotectionis enforcedbyhardwaremechanisms,which enablestrongguaranteesthatdataand processingarekeptconfidentialinthe cloudenvironmentandprevent unauthorizedexposureofdatawhendoing analytics.Confidentialcomputingis becomingcommercialincloudsystems. Researchisunderwaytoovercomethe remainingchallenges,includingimproving theefficiencyofthetrustedcomputing base,reducingcontextswitchoverheads whenportingapplicationsandpreventing sidechannelinformationleakage. Multi-partycomputationenables partiestojointlycomputefunctionsover theircombineddatainputswhilekeeping thoseinputsprivate.Inadditionto protectingtheconfidentialityoftheinput data,multi-partyprotocolsmust guaranteethatmaliciouspartiesarenot abletoaffecttheoutputofhonestparties. Althoughmulti-partycomputationis alreadyusedinspecialcases,itslimited functionalityandhighcomputation complexitycurrentlystandinthewayof wideadoption.Timewilltellifitbecomes aspromisingasconfidentialcomputing. PRIVACYREQUIRESSECURE IDENTITIES Digitalidentitiesarecrucialtomaintaining ownershipofdataandforauthenticating andauthorizingusers.Solutionsthat addressidentitiesandcredentialsfor machinesareequallyimportant.The widespreaduseofwebandcloud technologieshasmadetheneedfor efficientidentitysolutionsevenmore urgent.Inaddition,betteralgorithmsand newprotocolsforthetransportlayer securityprovideimprovedsecurity,lower latencyandreducedoverhead.Efficiency isparticularlyimportantwhen orchestratingandusingidentitiesformany dynamiccloudsystems,suchasthose realizedviamicroservices,forexample. Whenquantumcomputerswithenough computationalpowerareavailable,all existingidentitysystemsthatusepublic- keycryptographywilllosetheirsecurity. Developingnewsecurealgorithmsforthis post-quantumcryptographyeraisan activeresearcharea. 38 39#02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 ✱ FEATUREARTICLE FEATUREARTICLE ✱
  20. 20. TH E WI RELE SS access network is becoming a general connectivity platform that enables the sharing of data anywhere and anytime for anyone and anything. There is good reason to believe that rapidly increasing data volumes will continue in the foreseeable future. Ultra-reliable connectivity requires ultra-low latency, which will be needed to support demanding use cases. The focus will be on enabling high data rates for everyone, rather than support for extremely high data rates that are only achievable under specific conditions or for specific users. Afewtechnologieswillneedtobe enhancedinordertocreateaubiquitous, high-capacityradionetwork.Thecommon denominatorforthesetechnologiesistheir capabilitytoenableandutilizehigh frequenciesandwidebandwidth operations.Coverageisaddressed throughbeamformingandflexibilityin deviceinterworking.Thechallengeisto supportdatavolumesanddemanding- trafficusecases,withoutacorresponding increaseincostandenergyconsumption. DEVICESACTASNETWORKNODES Toenhancedevicecoverage,performance andreliability,simultaneousmulti-site connectivityacrossdifferentaccess technologiesandaccessnodesisrequired. Wirelesstechnologywillbeusedforthe connectivitybetweenthenetworknodes, asacomplementtofiber-basednetworks. Devicecooperationwillbeusedtocreate virtuallargeantennaarraysonthedevice sidebycombiningtheantennasofmultiple devices.Theborderlinebetweendevices andnetworknodeswillbemorediffuse. Massiveheterogenousnetworkswill haveamuchmoremesh-likeconnectivity. Advancedmachinelearningandartificial intelligencewillbekeytothemanagement ofthisnetwork,enablingittoevolveand adapttonewrequirementsandchangesin theoperatingenvironment. NOSURPRISE–EXPONENTIAL INCREASEDDATARATES Meetingfuturebitratedemandswill requiretheuseoffrequencybandsabove 100GHz.Operationinsuchspectrumwill enableterabitdatarates,althoughonlyfor short-rangeconnectivity.Itwillbean implementationchallengetogenerate substantialoutputpowerandhandleheat dissipation,consideringthesmall dimensionsofTHzcomponentsand antennas.Spectrumsharingwillbefurther enabledbybeamforming,whichismade possiblebythehighfrequency. Integratedpositioningwillbeenabledby high-frequencyandwide-bandwidth operationincombinationwithverydense deploymentsofnetworknodes.High- accuracypositioningisimportantfor enhancednetworkperformanceandisan enablerfornewtypesofend-userservices. Thepositioningofmobiledevices,both indoorandoutdoor,willbeanintegrated partofthewirelessaccessnetworks. Accuracywillbewellbelowonemeter. ANEWTRADE-OFFBETWEEN ANALOGANDDIGITALRADIO FREQUENCYHARDWARE Forthepast20yearstherehasbeen acontinuoustrendtowardmoving functionalityfromtheanalogtothedigital radiofrequencydomain.However,the trendisreversedforverywideband transmissionatveryhighfrequencies, incombinationwithaverylargenumber ofantennas.Thismeansthatanew implementationbalanceandinterplay betweentheanaloganddigitalradio frequencydomainswillemerge. Increasinglysophisticatedprocessingis alreadymovingovertotheanalogdomain. Thiswillsoonalsoincludeutilizing correlationsbetweendifferentanalog signalsreceivedondifferentantennas,for example.Thecompressionrequirements ontheanalog-to-digitalconversionis reduced.Thesplitbetweenanalogand digitalradiofrequencyhardware implementationwillchangeovertimeas technologyandrequirementsevolve. ubiquitous, high-capacity radio #5 40 41#02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 ✱ FEATUREARTICLE FEATUREARTICLE ✱
  21. 21. 42 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 43 ✱ DIGITAL CONNECTIVITY MARKETPLACES DIGITAL CONNECTIVITY MARKETPLACES ✱ consumenetworksjustasotherICTcapabilities areconsumed.Thebusinessmodelofthefuture willlookmorelikecurrentaaSmodelsforotherparts oftheICTsector. Thecurrenttelecombusinessmodel mustevolvesignificantlytoworkinthisnewmarket realitytoplacetelecomcompaniesinabetterposition tocapitalizeonnewbusinessopportunities. Thecaseforbusinessmodeltransformation Formanyyears,thetelecomindustryhassuccessfully beenoperatinginwhatisbestdescribedasavertical businessmodel.Inthismodel,telecomoperators havedeliveredend-to-end(E2E)standardized servicestoconsumersunderlong-termcontracts andcompetedwithinnationalboundaries. Thecollaborationbetweencommunication serviceprovidershasgenerallybeenlimitedto interoperabilityandroamingforthepurpose ofglobalservicereachandtodrivetechnology andmarketscale.Itisreasonabletoexpectthe traditionalmodusofcollaborationtocontinue toservetheindustryintheyearsahead,butitis alsoclearthatthisframeworkisnotfittoexplorethe fullvalueofnetworkinfrastructure.Richconfiguration variancesandmorecomplexserviceswillinmany casespushdemandbeyondthecapabilityofasingle Terms and abbreviations aaS–asaService|B2B–business-to-business|B2C–business-to-consumer|BSS–businesssupportsystems |DCP–DeviceConnectionPlatform|E2E–end-to-end|IoT–InternetofThings|OSS–operationssupport systems|SLA–ServiceLevelAgreement MALGORZATA SVENSSON, LARS ANGELIN, CHRISTIAN OLROG, PATRIK REGÅRDH, BO RIBBING Digitalization is undoubtedly one of the most significant forces for change of our time, characterized by a wide variety of smart devices and applications that touch all possible areas of life. Now, new technologies such as the Internet of Things (IoT), artificial intelligence, wearables and robotics are driving an accelerating digital transformation in which core processes across virtually all businesses and industries are becoming more distributed,connectedandreal-timeoptimized. ■Businessesneednetworksthatcanprovide themwiththefoundationtooperatethenew transformingbusinessmodels.Digitaltransformation requiresnetworkcapabilitiestosupportabroad spectrumofdifferentuserscenarios.Coverage, speed,latencyandsecurityaresomeofthekey parameters,alongwithnewnetworkfunctions foreffectiveprocessingofdataanddistributionof applicationsandfunctionsfaroutinthenetwork. 5Gbringstheabilitytorealizeawidevariety ofconnectivityandnetworkservicestomeet theperformancerequirementsoftomorrow’s digitalindustries. Withnetworkscapableofsupportingan unlimitedsetofservicesandusecases,ithasbecome increasinglyclearthattraditionalmass-market serviceswillsoonbeathingofthepast.Looking forward,networksmustbeviewedasahorizontal foundationorplatformonwhichbusinessescan One of the key growth opportunities for the telecom industry is to provide network capabilities that support the digital transformation underway in most businesses and industries. Already today, we have a powerful technology foundation in place, and this will become even stronger with 5G. Now is the ideal time to evolve the business side of the equation toward platform business models, which will enable the telecom industry to prosper in multisided business ecosystems as well. Digital marketplacesTO ENRICH 5G AND IoT VALUE PROPOSITIONS connectivity Definitions 〉〉 Consumer:auserassociatedwithacompanythatpurchasesservices/productsexposedbytheplatform/ marketplace.Aconsumercanbeanindividualand/oranorganization. 〉〉 Producer:asupplierofservices/productstoaplatform/marketplace.Aproducercanbeanindividualand/ oranorganization. 〉〉 Platformprovider:anorganizationthatmanagestheplatform/marketplace. 〉〉 Platform/marketplace:theentitythatexposestheservicesoftheplatform.Theexposedservicesarea compositionofplatformservicesandservicessuppliedbyproducers. 〉〉 Digitalconnectivitymarketplace:themarketplacethatoffersvariousconnectivityservicesandconnectivity serviceenablerstoenterprises.TheenterprisesmayrequireIoTcapabilities. 〉〉 Cloudservices:asetofservicesincludingstorageandcomputation. 〉〉 Digitalization:theuseofdigitaltechnologiestochangeabusinessmodelandprovidenewrevenueandvalue- producingopportunities.Itistheprocessofmovingtoadigitalbusiness.
  22. 22. 44 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 45 ✱ DIGITAL CONNECTIVITY MARKETPLACES DIGITAL CONNECTIVITY MARKETPLACES ✱ provider,aswellasbeyondthescopeofwhatis possibletorealizewithtraditionalinteroperability andfederationmodels. Inlightofthis,itisnecessarytoevolvetoward anewcollaborationframework–onethatoffers serviceprovidersanenvironmentwheretheycan dynamicallyaggregatecapabilitiesfrommultiple sourcesintothetailoredsolutiontheyrequireto enabletheiruniquedigitalizationjourney. Suchamodelmustexcelinservingtheuniqueand varyingneedsofconsumersaseffectivelyaspossible andincludevaluecontributionfrominnovation partners.Itmustalsomeetanewsetofrequirements toeffectivelycreateandcapturevalue.Someofthe mostcriticalaspectsare: 〉〉 simplicityforserviceproviderstodefineandconfigure theirownconnectivityandnetworksolutionregardless ofunderlyinginfrastructureandplayerboundaries. 〉〉 opennessforserviceproviderstoeasilyintegrateand dynamicallyadjusttheirsolution,includingtheneeded management,intotheirdigitalbusinessprocesses. 〉〉 well-structuredexposureofenhancedconnectivity andnetworkcapabilitiessuchasdistributedcloud, edgeanalytics,andsoon,forrapiduse-caseinnovation andvaluecreation. 〉〉 easyexchangeofdigitalassetssuchassafeandsecure tradeofdataacrossdifferentplayers. 〉〉 attractivecomplementforoperatorstoexplorecurrently unaddressablebusiness/marketsbeyondthevertical businesstheytraditionallyruntoday. 〉〉 creationofanenvironmentwhereinnovationthrives tothebenefitofallbusinessmodelparticipants. Toeffectivelymeettheseandotherrequirements,it isobviousthattheestablishmentofsuchabusiness modelisbeyondtheabilityofasingleplayeror company.Instead,itmakesmoresensetolookata businessthatisdesignedlikeaplatformwherea multitudeofplayers,producersandconsumerscan coexistandgainmutualbusinessoutcomes.Inthis scenario,theplatformisbothamarketplacethat supportsthebusinessofitsstakeholdersinthemost effectivewayandalsoaninfrastructurethat facilitatesandsimplifiesthesemultisided relationshipswitharangeofkeyfunctionalities. Multisidedconnectivityplatforms Multisidedplatformsbringtogethertwoormore distinctbutinterdependentgroupsofconsumers andsuppliers.Suchplatformsareofvaluetoone group(consumers,forexample)onlyiftheother group(suppliers,inthiscase)isalsopresent.A platformcreatesvaluebyfacilitatinginteractions andmatchmakingbetweendifferentgroups.A multisidedplatformgrowsinvaluetotheextent thatitattractsmoreusersanditmatchmakesone consumer’sneedswithservicesandproducts offeredbymultiplesuppliers;aphenomenonknown asthenetworkeffect. Thetelecomindustryalreadyoffersavastvariety ofconnectivityservices,and5Gtechnologies willacceleratethedevelopmentofnewones[1] characterizedbythedifferentcapabilitiesthatthe variousindustriesrequire.Theconnectivityservices of5Gwillberealizedbymeansofnetworkslicing. Webelievethattailoredconnectivityserviceswith diversecharacteristicswillbenecessaryfor enterpriseswithintheindustriestodigitalizetheir operationalandbusinessprocesses[2]. Digitalizationofenterprises’businessand operationrequiresnotonlyconnectivityservicesbut alsovariouscloud-basedserviceslikecomputation andstorage,wherecompanies’operationsand businessapplicationscanbehosted,deployedand operated,orwhereenterprisescanbenefitfromthe cloudservices’capabilitiesofferedinthesoftwareas servicebusinessmodel. Thefullpotentialof5Gconnectivityandcloud servicescanbereachedwhentheyareexposed onfullydigitalmultisidedconnectivityplatforms (marketplaces),asshowninFigure1.Inthisway,the services–andconsequentlytheindustries–will benefitfromthepresenceofavastrangeofother servicessuchasinformationandapplication enablementservices,aswellasbusinessand operationsupportcapabilities.Theconnectivity marketplacewillcomplementthefederationmodel forconnectivityandbridgethegapwherethe existingfederationmodelisnotenough. Thedigitalconnectivitymarketplacewillenable newbusinessmodelsbyprovidingcapabilities fortheexchangeofinformationandconnectivity enablingservicestoecosystemsofconsumers andsuppliers. MULTISIDEDPLATFORMS BRINGTOGETHERTWOORMORE DISTINCTBUTINTERDEPENDENT GROUPSOFCONSUMERSAND SUPPLIERS Figure 1: Digital connectivity marketplace: multisided platform Consumer Producer Platform provider Consumer Consumer Consumer Digital connectivity platform Digital connectivity marketplace Cloud services market Operation support services market Business support services market Application enablement market Information market Connectivity services market
  23. 23. 46 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 47 ✱ DIGITAL CONNECTIVITY MARKETPLACES DIGITAL CONNECTIVITY MARKETPLACES ✱ Ourvisionofthefuturedigitalconnectivity marketplaceisthatitwillofferinteractionbetween producersandconsumersbyexposingthe connectivityandcloudservicesononeside,and collectingtherequirementsfromconsumersand matchmakingthemagainstthecapabilitiesofthe servicesontheother.Theplatformwillfacilitatethe interactionsbyprovidingbusinessandoperation supportservicestobothproducersandconsumers. Theservicesexposedinthemarketplacecouldbe suppliedbymultipleproducersowningthenetworks andinfrastructuresrequiredtoproducethese services.Theconsumersrepresentingvarious industrieswillrequireconnectivityandcloud servicestodigitalizetheirbusinessand operationalprocesses[2]. Consumerservicesofferedbyproducers Asdiscussedabove,thefuturemultisided connectivitymarketplacewilloffercapabilitiesto exposeconnectivityandcloudservicesthatmay beprovidedbymultipleproducers.Inthecontext of5G thefollowingtypesofconnectivityservices maybeexposed:massivemachinetype communication(alsoknownasmassiveIoT), mission-criticalmachinetypecommunication (alsoknownasmission-criticalIoT)andextreme mobilebroadbandservices.Therewillbeaneed tocomplementtheconnectivityservicesprovided byotherstobeabletoofferthedesiredvalue propositionandsatisfyindustrydemands. Securitywillremainoneofthetoppriorities associetiescontinuetoconnecteverythingfrom autonomoustruckstoelevatorsandventilation systems.Isolationtechniquessuchasautomatic networkslicingarelikelytoplayakeyrolein reducingsecurityrisks,togetherwithmachine- learning-basedingressfirewalls,possiblyembedded asvirtualnetworkfunctionsinoperators’networks. Makingtheplatformattractive andefficienttooperate AtEricsson,weforeseethedigitalmarketplaceas thefoundationofanecosystemofmanyconsumers andprovidersfromamultitudeofindustries. Thisimpliesthatindustrieswillrequirecapabilities toexpresstheirrequirementsinlanguagesand contextsthatarespecifictotheirbusinessand operationalenvironments,asopposedtoneeding telcoandcloud-domainknowledgetobeableto chooseordefinetherequirementsonthe connectivityandcloudservices. Themultisidedcharacterofaplatform–andits matchmakingabilities–willmakeitwellsuited toprovidethesecapabilities.Theplatformwill matchindustry-specificrequirementswith servicecapabilities. Thematchmakingfunctionusestherelatively oldresearchareaofgraphtheory,boostedbyits explosioninusageandimplementationinsocial networksinrecentyears.ThemodelinFigure2 showsanexampleofhowconsumerrequirements, modeledasgraphedges,arelinkedwithservices, thentechnologiesandconsequentlysubscriptions. Bydefiningeachserviceasavertexinagraph whereedgesconstituteaconsumer/producer relationship,itispossibletomodelcomplexvalue chainsorrelationshipsbetweenservices. Givenasetofrequirementsmodeledasvertexes withconsumptionneedsas“danglingedges,” wecanfindanefficientsubsetofthegraphwhich fulfillstheseneeds,ifatallpossible.Byalsomodeling non-functionalaspectssuchaslocation,latency andthroughput,wegetagenericallyapplicable mechanism.Thisgivesrisetotremendous possibilitieswhereamultitudeoflooselyconnected playerstogetherenableanefficientE2Eservice, asadynamicallyconstructedvaluechain,capable ofreachingfarbeyondwhatiscurrentlypossible. Therearemanyimplementationoptions,including capabilitiesforlife-cyclemanagementtoensurethat consumersandproducers,joinedbymarketforces andtheplatform,provideasuitablemixofstability andinnovation. Figure 2: Graph model: how industry requirements can be linked with connectivity services Digital connectivity platform Matchmaking Subscription Application Capability Technology Technology Capability Capability Assets Service Subscription Producer Producer Consumer Subscription x Subscription y Producer y Data volume Throughput Latency Producer x Subscription Subscription z 4G Connectivity Vehicle Logistics Edge compute 5G SECURITYWILLREMAIN ONEOFTHETOPPRIORITIES ASSOCIETIESCONTINUE TOCONNECT
  24. 24. 48 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 49 ✱ DIGITAL CONNECTIVITY MARKETPLACES DIGITAL CONNECTIVITY MARKETPLACES ✱ Datalakesandanalyticsareusedtoexecute matchmaking,asshowninFigure3.Analytics serviceswillmakethefutureplatformdatadriven, astheywillbeexposedforusebyeitherconsumers orproducerstohelpthemevolvetheirservices. Atthesametime,theconsumersorproducers providetheplatformwithextendedinsightintohow producers’servicesareused,potentiallyhelping improvetherecommendationssupportingservice discoveryandmatchmaking. Acommonsolutionforprovidingidentification isyetanotherimportantcomponentoftheplatform totrulysupportscaleandinnovativecombinations. Federationofidentitiesisakeyservicewhen consideringnewservicecombinations.Authorization foraconnectionbetweentwothingswilllikelytake bothsoftwarefingerprintingandbehavioralanalysis intoaccountifthisfunctionalityisavailableina “thingmanagement”systeminatrustedrelationship. Securitysolutionswillenablethesafeandsecure tradeofdatabetweendifferentplayersandthe exchangeofdigitalassets[3].Trustfacilitieswill enablesecuretransactionsbetweenconsumers andproducers.Analyticsinsightintheshapeof fraudmanagement,combinedwithanovelform ofanalyticsinsightpredictinglikelihoodofsuccessful serviceactivation,willbethekeyfortheplatform providertobeabletomanageriskwhenbridging thetrustdivide. Serviceexposurewillsecureopennessfor consumerstoeasilyintegrateanddynamicallyadjust theirsolutionsandservices,includingtheneeded management,intotheirdigitalbusinessprocesses. Businesssupportplatform Thebusinesssupportplatformcomprisesmanythings forbothproducersandconsumers:theirentirelife cycle,interactionsandmanagementwhileinabusiness relationshipwiththeplatform;theentireservicelife cycleoftheonboardedproducerandconsumer servicessets;andself-servicesfortheaforementioned. Oneofthemostcriticalcapabilitiesiscontracting, duetothefactthatthemixofservices,SLAsand pricingschemesiscomplexandcontractsare frequentlynegotiatedandrenegotiated.Thebusiness supportplatformsimplifiesbusinessbyusingonly bilateralcontractsbetweenparties–thatis,a business-to-business(B2B)contractbetween anytwoofthethree:theproducer,theconsumer andthemarketplaceprovider.B2Ccontractsare applicableiftheconsumerisaprivateindividual andnotanenterprise.Whilebusinessvaluechains areoftenreferredtoasB2B2x,thecontracting arrangementsupportingthemisasetoflinked B2BandB2Ccontracts. Inaddition,themarketplaceallowsforapartyto actasaconsumertopurchaseservices,orchestrate themintoanewservice,andoffertheorchestrated serviceontheplatformasaproducer.Apartyin aproducerrolecanalsoactasaconsumerforits resourceneeds.ThisresultsinaseriesofB2Cand B2Bcontracts,whichistheessenceofavibrant ecosystembuiltontheplatform.Thecontractingand contractmanagementprocessesarefullyautomated. Operationssupportplatform Theconsumerandproducerservicesthatenablethe networkcapabilitiessuchasdistributedcloud andedgeanalyticswillbeexposedusingservice exposure,therebyenablinginnovationand valuecreation. Thedependenciesbetweenservicesare automaticallyinferredthroughservice-level assurance,andthecombinedanalyticsofthe differentservicesprovidesinsightintoperformance atseverallevels.Thismayallowfine-tuningofSLAs basedonheuristicsandhelpfuelinnovationby managingandprovidingtransparencywithregard torisk.Anetworkorserviceoutagemustbedealt withswiftly,astheymaybeusedinconsuming enterprises’mission-criticalproduction.Automatic healing,basedonorchestrationandpolicies,isone waytotomitigateoutagesituations. Theserviceconfigurationandactivationwillmake itpossibleforenterprisestodefineandconfigure theirconnectivityservicessettingssimplyand independentlyofunderlyinginfrastructureand playerboundaries. DeviceConnectionPlatform Oneofthefirstrealizationsofamultisidedplatform isEricsson’sDeviceConnectionPlatform(DCP). Thisisaplatformwheremobilenetworkconnectivity servicesareexposedtobeconsumedbyapplications. Itsfundamentalpurposeistomakeconnectivity availableforenterprisestoincludeintheirofferings andprovidethemeanstomanagethisconnectivity. Twoexamplesoftypicalenterprisesinthiscontext wouldbeanoriginalequipmentmanufacturer ofsomesort(suchasanautomotivecompanyor acameramanufacturer)oraserviceproviderofa servicewhereadeviceisanintegralpartofthe offering(suchasautilitycompanyusingautomatic meterreadingorapoint-of-salesterminalprovider). Asmobilenetworksarealwaysofageographically- limitednature,normallybycountryorlicensearea, thereisaneedtoharmonizethewaytheenterprise’s connectivityisusedacrossseveralnetworks.Ideally, theenterpriseneedstohavethesameconnectivity experiencewhereveritlaunchesandrunsitsservice. Theonlypracticalwaytoachievethisistostandardize onacentralizedplatform,whichbringstogether theconnectivityofallprovidersandexposesitin auniformway.Thismakestheexperiencetruly harmonized,intermsoffunctionalityaswellasservice levelsandoperationalprocedures.Furthermore,it alsominimizesthecostofintegrationbetweenthe enterprisesystemsandtheaccessnetworks,saving moneybothinitiallyandcontinuously,aswellas shorteningtimetomarketfortheenterprise’sservice. Conclusion Theongoingdigitalizationofbusinessesand industriesisoneofthekeygrowthopportunities Figure 3: Digital connectivity platform Digital connectivity platform Matching engine Serviceexposure Service activation Service configuration Service capability matchmaking Consumer/producer service management Consumer/producer managementInteraction management IdentitymanagementSecuritymanagement Assetmanagementintegration CloudserviceintegrationConnectivityserviceintegration Service assurance Business support systems Operations support systems Data lake and analytics
  25. 25. 50 #02 2018 ✱ ERICSSON TECHNOLOGY REVIEWERICSSON TECHNOLOGY REVIEW ✱ #02 2018 51 ✱ DIGITAL CONNECTIVITY MARKETPLACES DIGITAL CONNECTIVITY MARKETPLACES ✱ Further reading 〉〉 Ericsson white paper, November 2015, Operator service exposure – enabling differentiation and innovation, available at: https://www.ericsson.com/en/white-papers/operator-service-exposure--enabling- differentiation-and-innovation 〉〉 Ericsson Technology Review, April 2017, Tackling IoT complexity with machine intelligence, available at: https://www.ericsson.com/en/publications/ericsson-technology-review/archive/2017/tackling-iot-complexity- with-machine-intelligence 〉〉 Ericsson press release, December 2017, 2018 hot consumer trends: technology turns human, available at: https://www.ericsson.com/en/press-releases/2017/12/2018-hot-consumer-trends-technology-turns-human References 1. Ericssonwhitepaper,January2017,5G–enablingthetransformationofindustryandsociety,availableat: https://www.ericsson.com/en/white-papers/5g-systems--enabling-the-transformation-of-industry-and-society 2. Ericsson white paper, October 2017, Telecom IT for the digital economy, available at: https://www.ericsson.com/en/publications/white-papers/telecom-it-for-the-digital-economy 3. EricssonTechnologyReview,November2017,End-to-endsecuritymanagementfortheIoT,availableat: https://www.ericsson.com/en/ericsson-technology-review/archive/2017/end-to-end-security-management-for-the-iot Malgorzata Svensson ◆ is an expert in operations support systems (OSS). She joined Ericsson in 1996 and has worked in various areas within research and development. For the past 10 years, her work has focused on architecture evolution. Svensson has broad experience in business process, function and information modeling, information and cloud technologies, analytics, DevOps processes and tool chains. She holds an M.Sc. in technology from the Silesian University of Technology in Gliwice, Poland. Patrik Regårdh ◆ is head of strategy for Solution Area OSS, where his work focuses on market development, industry dynamics and driving strategies and initiatives for Ericsson’s digital business. He joined the company in 1994 and his previous positions have ranged from strategy and business development to account management. Currently based at the global headquarters in Stockholm, he has also worked extensively in Brazil, Thailand and Germany. Regårdh holds an M.Sc. from KTH Royal Institute of Technology in Stockholm, Sweden. Christian Olrog ◆ is an expert in cloud service delivery architecture. He joined Ericsson in 1999 and has been a technical leader and architect in many different areas, ranging from embedded client development and wireless enterprise connectivity with a security focus to cloud and operations support systems/ business support systems (OSS/BSS). Olrog currently sits in the Technology Industry group at Business Area Digital Services at Ericsson. He holds an M.Sc. in general physics from KTH RoyalInstituteofTechnology. Lars Angelin ◆ is an expert in BSS at Business Area Digital Services. He has more than 30 years of experience in the areas of concept development, architecture andstrategieswithinthe telco and education industries. He joined Ericsson in 1996 as a research engineer, and in 2003 he moved to a position as concept developer in the machine-to-machine and OSS/BSS areas. Since 2006, Angelin has been focusing on BSS – specifically business support, enterprise architectures and the software architectures to implement the systems. He holds an M.Sc. in engineering physics and a Tech. Licentiate in tele- traffic theory from the Faculty of Engineering at Lund University in Sweden, and an honorary Ph.D. from Blekinge Institute of Technology, Sweden. Bo Ribbing ◆ is head of Product Management for Connectivity Management. He joined Ericsson in 1991 and has worked in the networks business since 1995. During this time, he has gained broad international experience, particularly from Latin America and Asia Pacific, and has held several management positions. Ribbing holds an M.Sc. in applied physics from Linköping University in Sweden. theauthors Theauthors wouldliketo thankFrans deRooijforhis contributionto thisarticle. forthetelecomindustry.Telecominfrastructure, encompassingbothcurrentnetworksandsoon-to-be- launched5G,israpidlyevolvingintoaverypowerful resourcethatcanbringsignificantvaluetothedigital transformationofmostindustries.Lookingahead, someofthetelecomindustry’skeyvalueleversare: qualitydifferentiatedconnectivity;distributed computingandstorage;analyticsofdataflows; andsecuritysolutions.Withthetechnologyalready inplace,themainchallengeistofullyunderstand theneedsofanewbreedofconsumers,andusethat informationtoorganizethebusiness,developnew relationshipsandestablishefficientoperations. AtEricsson,webelievethatthebestwayto overcomethischallengeistoestablishaplatform modelwherecapabilitiesfrommanyproviderscan beeffectivelypackagedandexposedinattractive waystobuyersfromdifferentindustries.This approachcreatesanewandmuch-neededrole forthetelecomindustryastheproviderofthe marketplaceandbringsopportunitiesformultisided businessrelationsandtransactionstoprosper. Ononesideoftheplatformarethebusinesses andindustriesthatareconsumingservicesfrom theunderlyinginfrastructure,andontheotherside areprovidersofbothnetworkassetsandenabling functions.Theplatformorganizestherelationships foroptimalfitandusefulnessfortheplayersinvolved. Insodoing,theplatformprovidesanumberofuseful functionswithoneofthekeyvaluesbeingthescale ofbusinessandthenetworkingeffectthatithasto offer.Insteadofcompetingforthesameconsumers withthesameoffer,operatorsthatrestructure accordingtoalogicthatenablesfullparticipation inanecosystemsplatformwillfindthemselves wellpositionedtofullycapitalizeonnewmarket opportunities.

×