FP7 ICT ADVISORY GROUP. Working Group on "Future Internet Infrastructure" Version 8, 23 th January 2008

Transcription

1 FP7 ICT ADVISORY GROUP Wrking Grup n "Future Internet Infrastructure" Versin 8, 23 th January Intrductin The face f the Internet is cntinually changing, as new services and applicatins emerge and becme glbally significant at an increasing pace. Recent develpments include Scial netwrking interactivity and cmmunicatins User generated cntent wikis, blgs, pdcasts and vide Shift f bradcast cntent t the internet internet radi, dwnlads & IPTV Emergence f a rich set f web based services and applicatins resurces, service building blcks and applicatins Migratin f the internet frm the desktp int the wrld, thrugh wireless cnnectivity and mbile cmputing devices Increasing ptential fr the integratin f the physical wrld with the digital wrld f the internet, thrugh identificatin labels (RFID), sensrs and embedded systems and actuatrs New business and perating mdels fr prvisining f sftware and hardware (Sftware-as-a-Service, Clud Cmputing) Better access t knwledge fr humans and machine agents thrugh standardized semantic mark-up languages (RDF, OWL) The infrastructure f the Internet has and will cntinually evlve t supprt and enable new services, trends and businesses. This is best exemplified by the grwth (in this first phase f the Web) f the businesses f Ggle, Amazn, ebay and thers AND the cmputing infrastructures and services that are needed t supprt them. Each f them, alng with ther glbal service prviders, are perating netwrk and service infrastructures f increasing scale and diversity f services, with very large numbers f users. Over the next decade, develpments and innvatins in the netwrk and service infrastructure will cntinue, with significant investment n a glbal scale. Fr example, the multi-billin Eur investments that leading cmmunicatins service prviders are making in IP based Next Generatin Netwrks will simplify netwrk infrastructure, but the greater impact will cme frm the much richer and mre flexible service capability. Thugh the infrastructure f the internet has evlved, its underlying architecture has nt. This was nt created t functin in the rle that it currently undertakes, as the glbal critical infrastructure, and it has a number f fundamental limitatins. Mrever, the piecemeal grwth and additin f functinality has created a set f structural anmalies. It has prved t be very 1

2 successful as a unique platfrm fr glbal innvatin, but this success tends t bscure sme f the prblems, and t make them harder t fix. Sme believe that we are very clse t crisis, and this has mtivated research initiatives acrss the wrld, explring appraches fr a successr t, r evlutin f, the current netwrk architecture. In Eurpe, the EIFFEL Think-Tank reprt ( was released in December 2006, and identifies a number f drivers fr the next generatin Internet. Amngst the technlgical drivers are much mre emphasis n mbility, the anticipated changes in scale frm millins t billins f cnnected devices, increases in bandwidth, increase in digitised media, increasing imprtance f security and evlutin f services t mre adaptability, and awareness f user cntext and preferences. The reprt als identifies a number f sci-ecnmic drivers; these include the effect f ICT n sci-ecnmic develpment, the need fr privacy, security, the cncept f an nline identity implemented with credentials such as credit cards, but als avatars and the desire fr end-user empwerment. The reprt identifies technical challenges t be vercme; the main ne being t understand the architectural fundatins and design f the Future Internet. The aspiratin that we describe here is that the Future Internet will be a glbal, pen platfrm that will fix the shrtcmings f the current architecture, and als supprt a set f emerging Visins fr the future impact f the Internet. Each f these Visins take different perspectives n the underlying infrastructure, and ne f the key research challenges is t derive a rbust technical design fr the future netwrk and service infrastructure that supprts these. This will prbably replace the current layering paradigm with a mre general mdel. Eurpe is well-placed t take a leading rle in defining and develping the netwrk and service infrastructures fr the future. In the fllwing, we first lk at sme f the emerging trends and services that are likely t drive the evlutin f these infrastructures, fcussing n tw aspects that we believe are the mst significant and ptentially disruptive, and where there is the ptential fr Eurpean value: Internet f Services the service infrastructure Internet f Things the integratin f the physical and the digital wrld A third visin addresses sme majr enhancements in the user experience by the intrductin f an Internet f 3D Wrlds. We als discuss the trends, requirements and challenges fr the underlying netwrk and service infrastructure, since this is a critical underlying resurce fr the services and applicatins f the Future Internet. It is currently the fcus f majr reginal research initiatives, and it is imperative that Eurpe carefully cnsiders the scpe and bjectives f its research prgramme in relatin t this glbal activity. Finally, we examine the majr research challenges in realising these visins, and in designing the netwrk and service infrastructure that is required t supprt it. We clse with sme specific recmmendatins. 2. Emerging Visins fr the Internet In this sectin, we describe three f the Visins fr the Future Internet that will drive the requirements fr its netwrk and service infrastructure. Sme f the develpments and innvatin in these Visins can take place within existing architectures and business structures, and alng 2

3 well sign-psted trajectries (e.g. Service Oriented Architectures). In ther cases, change may require architectural revlutin, r may result in business disruptins. The Visin fr the Internet f Services. Lw entry barriers fr prvisining, brkering and cnsumptin f services are crucial fr large and in particular small enterprises acting as service prviders and trying t access a wrldwide market f ptential service users. In the last few years, Service Delivery Platfrms (SDP) have emerged in the telecmmunicatins sectr t manage delivery f cmmunicatins, media and entertainment services frm cmmunicatins service prviders t mbiles, PDAs and IPTV. SDPs integrate class-carrier platfrms, cnventinal middleware and enterprise services fr service delivery (CRM, billing, master data management), and nw frm a critical part in the technlgy stack f majr vendrs. With substantial grwth in a shrt time since inceptin and with massive frecasts f hand-held cmmunicatins devices, analysts predict a revenue surge as SDPs engage n data services as value-added cnsumables. A bradened cncept f glbal and pen Service Delivery Platfrm is prpsed fr the Internet f Services. It will g beynd the client-server mdel f service delivery t supprt rich mechanisms f glbal service supply, where third parties have the capability t aggregate services, act as intermediaries fr service delivery and prvide innvative new channels fr cnsuming services. This reflects the future requirements f the mainstream enterprise service cmmunities and the glbalizatin f these enterprise services. T realize this visin, an pen platfrm fr tradable, cmpsable, value-added services n the internet is required. Such a platfrm will need t build upn and extend: Web 2.0 cncepts t allw fr cmmunity-driven service innvatin and engineering n a large scale; glbal repsitries fr value-added services; and, semantic supprt t enhance enable autmatic cmpsitin f value-added services. This will enhance reusability f services, and allw fr reasning t derive further knwledge. Legal, security, lgistics, business and technical aspects must be simultaneusly addressed fr an integrated apprach t the Internet f Services, where the service cnsumer will get custmized services within Business Webs. This is why EDA (Event Driven Architecture) will cmplement SOA in custmize cmplex services either fr creating added-value mash-up services r fr sensing and reacting t (pssibly unexpected) situatins typical f cmpliance, lgistics r finance services. The Visin fr the Internet f Things The Internet f Things is ging t include infrmatin abut bjects f the real wrld and their respective surrundings. This infrmatin will be prvided by the things, as they btain and reveal the infrmatin thrugh a huge variety f sensrs and wireless cmmunicatin devices munted in different envirnments, embedded in systems, wrn by users, r even swallwed r injected under the skin. The increasing cmputing capabilities f such devices will als allw the implementatin f cmpletely nvel prcesses. Physical bjects include nt nly sensrs t detect a variety f infrmatin frm the physical wrld as well as lcal data strage and devices t help prcess the data, but als (mechatrnic/rbtic) machines that can act n the physical wrld. The Internet f Things can then integrate many different sensrs, actuatrs, micrsystems, mechatrnic systems, and rbts at its ends, thus 3

4 allwing the Internet nt nly t exchange and prcess infrmatin but als t cntrl actins in the real wrld. When cnnected t the Internet f Things, mechatrnics, smart prducts, and rbtics technlgy are becming increasingly pervasive in ur daily envirnments In Japan and Suth Krea, scenaris where the Internet cnnects smart envirnments, bjects, and rbts have been clearly envisaged, and strategic lines f research and develpment alng this directin are given high pririty. The Internet f Things can help integrate pervasive cmputing with pervasive rbtics, truly filling the gap between the ICT wrld and the physical wrld. Frm a strategic pint f view, this visin f The Internet f Things has many ptential advantages: it adds an enrmus range f new industrial pprtunities t ICT market, makers and prviders; it ffers new pprtunities fr a strategic alliance between ICT and nn-ict industry in Eurpe. Huge markets, such as autmatin, energy generatin and management, medical, envirnment, and s frth, can be better catered fr; it increases the perceived and real usefulness f the Internet t the majrity f EU citizens, wh are interested nt nly in navigating and retrieving infrmatin, but als in getting physical supprt t their daily needs (fr all citizens, and particularly fr citizens with special needs, such as disabled and elderly); it invlves nt nly the Internet geeks, but virtually all citizens; This visin is in full synergy with the visin f imprving ICT pprtunities by advancing the Internet infrastructure and services. A Visin f an Internet f 3D Wrlds The advent f 3D envirnments has been pineered thrugh games and is rapidly gaining mmentum. This technlgy places stretching requirements n the Internet infrastructures. Secnd life is a typical example, but mre cmplex applicatins are expected t emerge which will pen new classes f issues such as including high-availability and realtime requirements. Increased technlgical capabilities (see Justin Ratner's speech at last Intel's develpers frum r are prbably needed, whilst issues such as interactin between virtual and real wrld, interperability acrss multiple virtual envirnments, multiple identities r standards are still pen. In additin, such envirnments are expected t put additinal requirements n search (and find) technlgies which will have t cpe nt nly with text based infrmatin, but with unstructured and structured multimedia representatin f bjects, with r withut semantics built in. Grand Challenges In preparatin fr FP7, a previus incarnatin f ISTAG prduced the reprt "Grand Challenges in the Evlutin f the Infrmatin Sciety" (September, 2004). 4

5 The reprt ges n t identify 11 grand challenges. Of these, the 100% Safe Car, the Internet Plice Agent, the Pervasive Cmmunicatin Jacket and the Intelligent Retail Stre are the mst pertinent t the Future Internet theme. Cmmn themes which run thrugh the grand challenges include: The need fr rbust design principles fr dependable systems. The issues f trust, privacy and security. The challenges created by dramatic changes in scale (frm millins f wired cmpnents t billins f wireless cmpnents). The user experience Frm the user perspective, there are a number f ways in which the Future Internet must supprt r deliver richer capabilities: Intelligence T the user, the Future Internet will present Itself as an intelligent infrastructure which supprts humans as well as machine agents and allws fr seamless interactin f bth f them It will need t understand the semantics f cntent and t deliver knwledge rather than infrmatin t the user, in a pr-active manner. With the increased cupling between the digital and physical wrlds, it will prvide effect and actin as well as infrmatin. The greater part f this intelligent actin will be prvided by the intelligent services and applicatins running n the infrastructure, but the Future Internet Infrastructure must prvide the apprpriate supprt Quality f Service Quality f service will be measured in a deeper way, relating t the quality f the infrmatin prvided, r the user experience Service Repsitry The Future Internet shuld ffer like the Naming service (DNS/ICAN) als a Web Services repsitry and gvernance service (UDDI/???). Trust In the Future Internet, Trust will be critical. Can we trust the peple we meet, r the rganisatins and cmpanies that we deal with? Understanding trust in its multiple facets and manifestatins: trust between peple, between peple and the cyber-infrastructure, amng peple using the cyber-infrastructure, in a sciety f human and sftware entities, finally in mre cmplex value netwrk and ecsystem settings. Can we trust the data r the knwledge? Trusting data r knwledge passes thrugh engineering trust management, access cntrl and 5

6 plicy systems. They shuld be flexible enugh t understand and accmmdate the dynamic nature f trust. They shuld integrate seamlessly with existing and future security systems, fr example detectin f fraud, intrusin attacks, ther malicius behaviur, trust and risk assessments and relatinship and access cntrl plicies. Mrever cnsidering the huge number f data and knwledge circulating in the Internet f the Future, certify the prvenance f data r knwledge will becme a critical issue. 3. Requirements fr the Future Internet Infrastructure The visins briefly sketched ut abve begin t define sme f the characteristics f the netwrk and service infrastructure in this Future Internet. We expect an infrastructure that will prvide: Cnnectivity services hrizntal services such as identity, ttrust, lcatin, brkering Cmputing resurces as services These will need t be prvided in an intelligent manner, persnalised t the needs and cntext f the user, and at the desired quality level. And supprt the vertical services : Infrmatin & knwledge services Business and applicatins services Sense and actin n the real wrld Als, the infrastructure must supprt dynamic business relatinships and value chains. Service prviders need an infrastructure where they are able t manage and balance revenue, csts and risk. The vertical services will als need the supprt f a rich set f hrizntal enabling services such as identity, trust, lcatin, brkering etc. We can glimpse sme f the first signs f this future infrastructure tday, in advanced SOA implementatins within enterprises, in the NGN implementatins f leading telecmmunicatins prviders, and with the emergence f web services and utility cmputing services n the Internet. Hwever, there are frmidable challenges in scaling up t a glbal pen and dynamic infrastructure, These lead implementatins are built n tp f the current Internet architecture, and wuld seem t ffer evidence that a new service infrastructure culd be built n tp f the current netwrk architecture, in the small at least. Hwever, this bscures sme very significant issues: The current Internet architecture has sme very significant shrtcmings, and can be cnvincingly described as perating n the edge. The current architecture des nt have the hks t supprt flexible business mdels that will reward investment in the infrastructure which will be required t supprt the type f capabilities envisaged in the earlier visins. 6

7 There is als a strng desire t virtualise cnnectivity, and prvide this as a smart and dynamic cnverged service. This wuld require radical architectural changes, including embedding sme key hrizntal enabling services in the netwrk infrastructure. Fixing these issues with radically different netwrk architecture culd pen up cnnectivity t the same kind f dynamic innvatin that we are expecting t see in ther virtualised services. It als ensures that we are nt cnstructing a new service industry n the sandy fundatins f the current architecture. There is a large prgramme within the Framewrk prgramme n the Future Internet (at the level f the netwrk infrastructure), and ne pprtunity fr this paper is t prvide sme guidance t that prgramme. 4. Research Challenges Internet f Services Overall, we see as main challenges: Enablement f Business Webs where services are tradable gds n the internet by develping a next generatin service infrastructure fr engineering, discvery, brkerage, cmpsitin and executin f value-added services. Key ingredients are Web 2.0, Semantic Technlgies, Service-Oriented Architectures (SOA) and their integratin with Event Driven Architectures (EDA). Harmnizatin f system architectures (SOA),service infrastructure architectures (SOI), and netwrking architectures t advance the structure f multi-tier, federated and Internet scale architectures, supprt all kinds f business mdels, applicatins and emerging hardware and netwrking envirnments and prvide transparent and integrated access fr all relevant stakehlders Advanced system lifecycle appraches including engineering, deplyment, cmpsitin, prvisining, management and decmmissining phases that supprt transparent knwledge tracking, feedback lps, predictin and simulatin, allw fr a clear separatin f cncerns between different stakehlders and supprt the full variety f business scenaris while adhering t verarching sustainability requirements. Advanced infrastructure technlgies in terms f hardware, middleware and related prgramming mdels that meet the required flexibility f the netwrked ecnmy. Repsitry service and gvernance (UDDI/???) Internet f Things Technical research challenges t turn the visin f Internet f Things int reality have t be addressed at multiple layers f the entire technlgical framewrk and infrastructure (discussed bttm-up): Further research n and develpment f basic enabling r s called edge technlgies, such as sensrs and actuatrs, passive/active identificatin tags, embedded systems etc. that are attached t real wrld bjects and making bjects smart enugh t participate in nvel Internet f Things applicatin scenaris. Are there missing standards? 7

8 Research must als cncentrate netwrking technlgies, such as fixed, mbile, wired and wireless netwrks allwing the highly available bidirectinal cmmunicatin n different levels, i.e. between real wrld bjects, applicatins and services that ffer functinality specific t Internet f Things. Scalable, secure and semantically enriched middleware systems (lw level SOAs) play a key rle putting real wrld data int the cntext f varius Internet applicatins. Research will have t cncentrate n hw t bridge gaps that ccur due t the usage f hetergeneus device, netwrk, sensrs and ther technlgies. Platfrm services that run in the backgrund have t supprt a superir management f all invlved technical cmpnents (i.e. trillins f lsely cupled devices!) in an integrated way ensuring scalability, high availability, and the safe and secure executin f the requested functinality. Further t the abve technical questins, intensive research has t be cnducted in rder t accelerate the adptin f Internet f Things in varius applicatins dmains. This cvers the analysis f current and future demands and trend in varius industries, public and gvernmental rganizatins, etc. t find and justify the ecnmic and scietal impact. Sme ecnmic benefits that may mtivate the implementatin f the Internet f Things are: imprved prcess perfrmance, visibility, and scalability, higher level f autmatin, increased cst efficiency thrugh real-time, high-reslutin data, enhanced prcess, prduct r service perfrmance management, better transparency f physical flws and detailed status infrmatin, etc. Cnverged Infrastructure challenges acrss the visins There are a set f challenges frm the tw visins that shuld be brught tgether, but addressed in different ways in different timescales. In the nearer term, in the cntext f the current netwrk architecture (IP and the end-t-end architecture), the fllwing issues need t be cnsidered: Cnverge the visins f the Internet f Services and the Internet f Things int a seamless single cncept. Especially the Internet f Things layers f platfrm service structures and semantically enriched middleware cmpnents might prfit frm sphisticated service prvisining methds described in the respective sectins abve. Hrizntal and vertical Interperability f Trust, Security Dependability attributes in envirnments where the bundaries between the netwrk and service prviders are n lnger clear and multi-service interactins amng service prviders can span a plyinfrastructure cnsisting f varied unrelated netwrk prviders. This makes nt feasible at the mment delivery f services with run-time agreements f trust security and dependability attributes ver these envirnments. Then, these cnverged visins can be used as a cmmn scenari t drive the design f next generatin netwrk architecture. The majr challenges in this are: Develping an architecture that fixes the shrtcmings f the current Internet, which include: Security, Privacy, Trust, Identity Management Hks fr business and incentive mdels 8

9 Supprt fr Semantics Supprt fr mbility Adequate resilience (e.g. DDOS) Creating an architecture that is flexible enugh t supprt a range f applicatin Visins and business mdels in a dynamic way. Replacing the traditinal layering paradigm with a mre general mdel (which may have impact n wider system architectures) Ensuring cnvergence between technlgy, business and regulatry cncerns There is a further general challenge, which is hard t assess at this stage. This relates t the peratin f service ecsystems at massive scale, as is envisaged in the Visins. 5. Recmmendatins Future Internet Infrastructure The visin frm the previus ISTAG (and frm the cmpanin paper n Web services) is that a Eurpean fcus will be in utilizing ICT t slve scietal challenges. This creates an asset that can be sld in Eurpean and glbal markets. This is greatly facilitated by cmmn glbal standards in the netwrk and service infrastructure, which wuld encurage a cllabrative apprach t wrking with the Future Internet research initiatives in ther majr regins. The visins fr the Future Internet presented here, and thers, all fresee an envirnment where what the user receives is the result f a cmpsitin f a number f elements. It is essential that the architecture develped is capable f sharing the reward frm prviding the end service dwn the value chain t the cntributrs f thse elements used. In rder t avid creating new stve-pipes, such a mechanism must be capable f stretching beynd the elements envisaged by any particular visin. They als need t include the return fr the cntributin f the infrastructure upn which all the capabilities will. In develping a new netwrk architecture (as well as with cnverging between and amngst vertical scenaris), the Internet experience has been that there are prblems with wrking n t narrw a scenari ( "internet f services" r "internet f things") AND there are well knwn limitatins with hrizntal layers. Many f these prblems stem frm the apprach that tries t identify the full set f requirements up frnt and t embed the reslutin f all resurce cnflicts in the design phase. The ne thing that is clear frm the histry f the current Internet is that we cannt predict hw it will be used in the future. We therefre recmmend that there needs t be a fcus n prgrammable / run time defined appraches t reslving resurce cnflicts. We are suspicius f the value f large scale and generic test-beds early in the research life cycle. Test-beds shuld be t test 9

10 In parallel with the technlgy research, it is imprtant t create frums and dialgues with all the stakehlders in the Future Internet, n technlgy, business and regulatin (vendrs, service prviders ld and new, users, develpers and regulatrs). This needs t be encuraged at as early a stage as pssible. Cmpared t the US prgramme n the Future Internet, r Krea, r China, the EU are perhaps t cnstrained by the instruments f Framewrk, They are able t prvide supprt fr a prtfli f research prjects, but it is difficult t prvide real directin, leadership r fcus. In the Future Internet, Eurpe is in danger f being left behind thrugh this. It is an area where industry culd have mre ability t create an atmsphere cnducive t leadership, but (again) the current instruments dn t lend themselves t an bvius way frward. 10