Future internet architecture (visit http://trends-in-telecoms.blogspot.com/ for more insights)


Published on

Future internet architecture (visit http://trends-in-telecoms.blogspot.com/ for more insights)

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Future internet architecture (visit http://trends-in-telecoms.blogspot.com/ for more insights)

  1. 1. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F FUTURE INTERNET ARCHITECTURES: DESIGN AND DEPLOYMENT PERSPECTIVES A Survey of the Research on Future Internet Architectures Jianli Pan, Subharthi Paul, and Raj Jain, Washington University ABSTRACT Security needs to be an inherent feature and integral part of the architecture. Also, there is a The current Internet, which was designed significant demand to transform the Internet over 40 years ago, is facing unprecedented chal- from a simple “host-to-host” packet delivery lenges in many aspects, especially in the com- paradigm into a more diverse paradigm built mercial context. The emerging demands for around the data, content, and users instead of security, mobility, content distribution, etc. are the machines. All of the above challenges have hard to be met by incremental changes through led to the research on future Internet architec- ad-hoc patches. New clean-slate architecture tures. designs based on new design principles are Future Internet architecture is not a single expected to address these challenges. In this improvement on a specific topic or goal. A clean- survey article, we investigate the key research slate solution on a specific topic may assume the topics in the area of future Internet architec- other parts of the architecture to be fixed and ture. Many ongoing research projects from unchanged. Thus, assembling different clean- United States, the European Union, Japan, slate solutions targeting different aspects will not China, and other places are introduced and dis- necessarily lead to a new Internet architecture. cussed. We aim to draw an overall picture of Instead, it has to be an overall redesign of the the current research progress on the future whole architecture, taking all the issues (security, Internet architecture. mobility, performance reliability, etc.) into con- sideration. It also needs to be evolvable and flex- INTRODUCTION ible to accommodate future changes. Most previous clean-slate projects were focused on The Internet has evolved from an academic net- individual topics. Through a collaborative and work to a broad commercial platform. It has comprehensive approach, the lessons learned become an integral and indispensable part of and research results obtained from these individ- our daily life, economic operation, and society. ual efforts can be used to build a holistic Inter- However, many technical and non-technical net architecture. challenges have emerged during this process, Another important aspect of future Internet which call for potential new Internet architec- architecture research is the experimentation tures. Technically, the current Internet was testbeds for new architectures. The current designed over 40 years ago with certain design Internet is owned and controlled by multiple principles. Its continuing success has been hin- stakeholders who may not be willing to expose dered by more and more sophisticated network their networks to the risk of experimentation. So attacks due to the lack of security embedded in the other goal of future Internet architecture the original architecture. Also, IP’s narrow waist research is to explore open virtual large-scale means that the core architecture is hard to mod- testbeds without affecting existing services. New ify, and new functions have to be implemented architectures can be tested, validated, and through myopic and clumsy ad hoc patches on improved by running on such testbeds before top of the existing architecture. Moreover, it has they are deployed in the real world. become extremely difficult to support the ever In summary, there are three consecutive steps increasing demands for security, performance leading toward a working future Internet archi- reliability, social content distribution, mobility, tecture: and so on through such incremental changes. As Step 1: Innovations in various aspects of the a result, a clean-slate architecture design Internet This work was supported paradigm has been suggested by the research Step 2: Collaborative projects putting multiple in part by a grant from community to build the future Internet. From a innovations into an overall networking archi- Intel Corporation and non-technical aspect, commercial usage requires tecture NSF CISE Grant fine-grained security enforcement as opposed to Step 3: Testbeds for real-scale experimentation #1019119. the current “perimeter-based” enforcement. It may take a few rounds or spirals to work out a 26 0163-6804/11/$25.00 © 2011 IEEE IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  2. 2. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F future Internet architecture that can fit all the These projects also face challenges such as how requirements. to trade off mobility with scalability, security, Since the primary Future Internet research efforts may be clas- and privacy protection of mobile users, mobile usage of the today’s sified based on their technical and geographical endpoint resource usage optimization, and so diversity. While some of the projects target at on. Internet has changed individual topics, others aim at holistic architec- Cloud-computing-centric architectures: from host-to-host tures by creating collaboration and synergy Migrating storage and computation into the among individual projects. Research programs “cloud” and creating a “computing utility” is a communication to specifically aimed at the design of the future trend that demands new Internet services and content distribution, Internet have been set up in different countries applications. It creates new ways to provide it is desirable to around the globe, including the United States, global-scale resource provisioning in a “utility- the European Union (EU), Japan, and China. like” manner. Data centers are the key compo- change the The geographical diversity of research presents nents of such new architectures. It is important architecture’s narrow different approaches and structures of these dif- to create secure, trustworthy, extensible, and ferent research programs. While dividing the robust architecture to interconnect data, con- waist from IP to the projects by their major topics is also possible, trol, and management planes of data centers. data or content due to the holistic architecture goals, different The cloud computing perspective has attracted distribution. projects may have some overlap. considerable research effort and industry pro- Over the past few years’ future Internet jects toward these goals. A major technical research has gathered enormous momentum as challenge is how to guarantee the trustworthi- evidenced by the large number of research pro- ness of users while maintaining persistent ser- jects in this area. In this article, primarily based vice availability. on the geographical diversity, we present a short Security: Security was added into the original survey limited in scope to a subset of representa- Internet as an additional overlay instead of an tive projects and discuss their approaches, major inherent part of the Internet architecture. Now features, and potential impact on the future. security has become an important design goal We discuss the key research topics and design for the future Internet architecture. The research goals for the future Internet architectures. is related to both the technical context and the Research projects in the United States, Euro- economic and public policy context. From the pean Union, and Asian countries are discussed technical aspect, it has to provide multiple gran- in detail, respectively. Some of our discussions ularities (encryption, authentication, authoriza- and perspectives on future Internet architectures tion, etc.) for any potential use case. Also, it are included later. Finally, a summary concludes needs to be open and extensible to future new the article. security related solutions. From the non-techni- cal aspect, it should ensure a trustworthy inter- face among the participants (e.g., users, KEY RESEARCH TOPICS infrastructure providers, and content providers). In this section, we discuss some key research There are many research projects and working topics that are being addressed by different groups related to security. The challenges on this research projects. topic are very diverse, and multiple participants Content- or data-oriented paradigms: Today’s make the issue complicated. Internet builds around the “narrow waist” of IP, Experimental testbeds: As mentioned earlier, which brings the elegance of diverse design developing new Internet architectures requires above and below IP, but also makes it hard to large-scale testbeds. Currently, testbed research change the IP layer to adapt for future require- includes multiple testbeds with different virtual- ments. Since the primary usage of today’s Inter- ization technologies, and the federation and net has changed from host-to-host coordination among these testbeds. Research communication to content distribution, it is organizations from the United States, European desirable to change the architecture’s narrow Union, and Asia have initiated several programs waist from IP to the data or content distribution. related to the research and implementation of Several research projects are based on this idea. large-scale testbeds. These projects explore chal- This category of new paradigms introduces chal- lenges related to large-scale hardware, software, lenges in data and content security and privacy, distributed system test and maintenance, security scalability of naming and aggregation, compati- and robustness, coordination, openness, and bility and co-working with IP, and efficiency of extensibility. the new paradigm. Besides these typical research topics, there Mobility and ubiquitous access to networks: are several others, including but not limited to The Internet is experiencing a significant shift networked multimedia; “smart dust,” also called from PC-based computing to mobile computing. the “Internet of things”; and Internet services Mobility has become the key driver for the future architecture. However, note that in this survey, Internet. Convergence demands are increasing we are not trying to enumerate all the possible among heterogeneous networks such as cellular, topics and corresponding research projects. IP, and wireless ad hoc or sensor networks that Instead, we focus on a representative subset and have different technical standards and business discuss a few important ongoing research pro- models. Putting mobility as the norm instead of jects. an exception of the architecture potentially nur- Due to length limitations, we are not able to tures future Internet architecture with innovative enumerate all the references for the projects dis- scenarios and applications. Many collaborative cussed below. However, we do have a longer sur- research projects in academia and industry are vey [18], which includes a more complete pursuing such research topics with great interest. reference list for further reading. IEEE Communications Magazine • July 2011 27Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  3. 3. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F network again and again without any traffic Categories Project or cluster names (selected) optimization on the network’s part. The NDN uses a different model that enables the network FIA NDN, MobilityFirst, NEBULA, XIA, etc. to focus on “what” (contents) rather than “where” (addresses). The data are named CABO, DAMS, Maestro, NetSerV, RNA, SISS, etc. (more than 47 instead of their location (IP addresses). Data FIND total) become the first-class entities in NDN. Instead of trying to secure the transmission channel or Spiral1: (5 clusters totally): DETER (1 project), PlanetLab (7 pro- data path through encryption, NDN tries to jects), ProtoGENI (5 projects), ORCA (4 projects), ORBIT (2 pro- secure the content by naming the data through jects; 8 not classified; 2 analysis projects a security-enhanced method. This approach GENI allows separating trust in data from trust Spiral2: over 60 active projects as of 2009* between hosts and servers, which can potentially enable content caching on the network side to Spiral3: about 100 active projects as of 2011* optimize traffic. Figure 1 is a simple illustration of the goal of NDN to build a “narrow waist” * GENI design and prototyping projects can last for more than one spiral. around content chunks instead of IP. NDN has several key research issues. The Table 1. U.S. projects and clusters on the future Internet. first one is how to find the data, or how the data are named and organized to ensure fast data lookup and delivery. The proposed idea is to RESEARCH PROJECTS FROM THE name the content by a hierarchical “name tree” UNITED STATES which is scalable and easy to retrieve. The sec- ond research issue is data security and trustwor- Research programs on future Internet architec- thiness. NDN proposes to secure the data ture in United States are administrated by the directly instead of securing the data “containers” National Science Foundation (NSF) directorate such as files, hosts, and network connections. for Computer and Information Science and The contents are signed by public keys. The Engineering (CISE). third issue is the scaling of NDN. NDN names are longer than IP addresses, but the hierarchi- FIA AND FIND cal structure helps the efficiency of lookup and The Future Internet Architecture (FIA) pro- global accessibility of the data. gram [1] of the National Science Foundation Regarding these issues, NDN tries to address (NSF) is built on the previous program, Future them along the way to resolve the challenges in Internet Design (FIND) [2]. FIND funded about routing scalability, security and trust models, fast 50 research projects on all kinds of design data forwarding and delivery, content protection aspects of the future Internet. FIA is the next and privacy, and an underlying theory supporting phase to pull together the ideas into groups of the design. overall architecture proposals. There are four such collaborative architecture groups funded MobilityFirst — The MobilityFirst [4] project under this program, and we introduce them is led by Rutgers University with seven other here. Table 1 illustrates the overall research universities. The basic motivation of Mobility- projects from the United States, including FIA First is that the current Internet is designed for and FIND. interconnecting fixed endpoints. It fails to address the trend of dramatically increasing Named Data Networking (NDN) — The demands of mobile devices and services. The Named Data Networking (NDN) [3] project is Internet usage and demand change is also a key led by the University of California, Los Angeles driver for providing mobility from the architec- with participation from about 10 universities and tural level for the future Internet. For the near research institutes in the United States. The ini- term, MobilityFirst aims to address the cellular tial idea of the project can be traced to the con- convergence trend motivated by the huge cept of content-centric networks (CCNs) by Ted mobile population of 4 to 5 billion cellular Nelson in the 1970s. After that, several projects devices; it also provides mobile peer-to-peer such as TRIAD at Stanford and DONA from (P2P) and infostation (delay-tolerant network the University of California at Berkeley were [DTN]) application services which offer robust- carried out exploring the topic. In 2009 Xerox ness in case of link/network disconnection. For Palo Alto Research Center (PARC) released the the long term, in the future, MobilityFirst has CCNx project led by Van Jacobson, who is also the ambition of connecting millions of cars via one of the technical leaders of the NDN project. vehicle-to-vehicle (V2V) and vehicle-to-infra- The basic argument of the NDN project is structure (V2I) modes, which involve capabili- that the primary usage of the current Internet ties such as location services, georouting, and has changed from end-to-end packet delivery to reliable multicast. Ultimately, it will introduce a a content-centric model. The current Internet, pervasive system to interface human beings which is a “client-server” model, is facing chal- with the physical world, and build a future lenges in supporting secure content-oriented Internet around people. functionality. In this information dissemination The challenges addressed by MobilityFirst model, the network is “transparent” and just include stronger security and trust requirements forwarding data (i.e., it is “content-unaware”). due to open wireless access, dynamic association, Due to this unawareness, multiple copies of the privacy concerns, and greater chance of network same data are sent between endpoints on the failure. MobilityFirst targets a clean-slate design 28 IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  4. 4. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F directly addressing mobility such that the fixed Internet will be a special case of the general A “narrow waist” around content chunks instead of the IP. design. MobilityFirst builds the “narrow waist” of the protocol stack around several protocols: Web, email, VoIP, eBusiness... Browsers, Skype, online gaming... • Global name resolution and routing service HTTP, RTP, SMTP... File streams... • Storage-aware (DTN-like) routing protocol • Hop-by-hop segmented transport TCP, UDP, SCTP... Security... • Service and management application pro- IP “Narrow waist” Contents gramming interfaces (APIs) The DTN-like routing protocol is integrated with Ethernet, WIFi... Strategies... the use of self-certifying public key addresses for CSMA, ADSL, Sonet... P2P, UDP, IP broadcast... inherent trustworthiness. Functionalities such as context- and location-aware services fit into the Optical fiber, copper, radio... Optical fiber, copper, radio... architecture naturally. An overview of the Mobil- ityFirst architecture is shown in Fig. 2. It shows all the building blocks mentioned above and how Figure 1. The new “narrow waist” of NDN (right) compared to the current they work together. Internet (left). Some typical research challenges of Mobility- First include: Mobility first routers are • Trade-off between mobility and scalability with storage capability • Content caching and opportunistic data delivery Hop-by-hop segment • Higher security and privacy requirements transport • Robustness and fault tolerance NEBULA — NEBULA [5] is another FIA pro- Core network ject focused on building a cloud-computing-cen- tric network architecture. It is led by the Data plane University of Pennsylvania with 11 other univer- Generalized sities. NEBULA envisions the future Internet DTN routing consisting of a highly available and extensible Global name Name to resolution service core network interconnecting data centers to address provide utility-like services. Multiple cloud pro- mapping viders can use replication by themselves. Clouds comply with the agreement for mobile “roam- Control and management plane ing” users to connect to the nearest data center with a variety of access mechanisms such as wired and wireless links. NEBULA aims to Figure 2. MobilityFirst architecture. design the cloud service embedded with security and trustworthiness, high service availability and reliability, integration of data centers and tipath routing and use of new networks. NDP routers, evolvability, and economic and regulato- involves a novel approach for network path ry viability. establishment and policy-controlled trustworthy NEBULA design principles include: paths establishment among NEBULA routers. • Reliable and high-speed core interconnect- Figure 3 shows the NEBULA architecture com- ing data centers prising the NDP, NVENT, and NCore, and • Parallel paths between data centers and shows how they interact with each other. core routers • Secure in both access and transit eXpressive Internet Architecture (XIA) — • A policy-based path selection mechanism Expressive Internet Architecture (XIA) [6] is • Authentication enforced during connection also one of the four projects from the NSF FIA establishment program, and was initiated by Carnegie Mellon With these design principles in mind, the NEB- University collaborating with two other universi- ULA future Internet architecture consists of the ties. As we observe, most of the research pro- following key parts: jects on future Internet architectures realize the • The NEBULA data plane (NDP), which importance of security and consider their archi- establishes policy-compliant paths with flex- tecture carefully to avoid the flaws of the origi- ible access control and defense mechanisms nal Internet design. However, XIA directly and against availability attacks explicitly targets the security issue within its • NEBULA virtual and extensible networking design. techniques (NVENT), which is a control There are three key ideas in the XIA archi- plane providing access to application- tecture: selectable service and network abstractions • Define a rich set of building blocks or com- such as redundancy, consistency, and policy munication entities as network principals routing including hosts, services, contents, and • The NEBULA core (NCore), which redun- future additional entities. dantly interconnects data centers with ultra- • It is embedded with intrinsic security by high-availability routers using self-certifying identifiers for all princi- NVENT offers control plane security with poli- pals for integrity and accountability proper- cy-selectable network abstraction including mul- ties. IEEE Communications Magazine • July 2011 29Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  5. 5. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F ent from common testbeds in that it is a general- purpose large-scale facility that puts no limits on the network architectures, services, and applica- NDP path NVENT tions to be evaluated; it aims to allow clean-slate NDP path Reliable designs to experiment with real users under real Wireless trustworthy Wired conditions. access core network access network (NCore) Data center network The key idea of GENI is to build multiple NVENT virtualized slices out of the substrate for resource sharing and experiments. It contains two key Data center Transit network pieces: • Physical network substrates that are expand- able building block components Figure 3. NEBULA architecture components and their interactions. • A global control and management frame- work that assembles the building blocks together into a coherent facility • A pervasive “narrow waist” (not limited to Thus, intuitively two kinds of activities will be the host-based communication as in the involved in GENI testbeds: one is deploying a current Internet) for all key functions, prototype testbed federating different small and including access to principals, interaction medium ones together (e.g., the OpenFlow among stakeholders, and trust manage- testbed for campus networks [8]); the other is to ment; it aims to provide interoperability at run observable, controllable, and recordable all levels in the system, not just packet for- experiments on it. warding. There are several working groups concentrat- The XIA components and their interactions ing on different areas, such as the control frame- are illustrated in Fig. 4. The core of the XIA is work working group; GENI experiment workflow the Expressive Internet Protocol (XIP) support- and service working group; campus/operation, ing communication between various types of management, integration, and security working principals. Three typical XIA principal types are- group; and instrumentation and management content, host (defined by “who”), and service working group. (defined by what it does). They are open to The GENI generic control framework con- future extension. Each type of principal has a sists of several subsystems and corresponding narrow waist that defines the minimal function- basic entities: ality required for interoperability. Principles talk • Aggregate and components to each other using expressive identifiers (XIDs), • Clearinghouse which are 160 bit identifiers identifying hosts, • Research organizations, including pieces of content, or services. The XIDs are researchers and experiment tools basically self-certifying identifiers taking advan- • Experiment support service tage of cryptographic hash technology. By using • “Opt-in” end users this XID, the content retrieval no longer relies • GENI operation and management on a particular host, service or network path. Clearinghouses from different organizations and XIP can then support future functions as a places (e.g., those from the United States and diverse set of services. For low-level services, it European Union) can be connected through fed- uses a path-segment-based network architecture eration. By doing this, GENI not only federates (named Tapa in their previous work) as the with identical “GENI-like” systems, but also with basic building block; and builds services for con- any other system if they comply with a clearly tent-transfer and caching and service for secure defined and relatively narrow set of interfaces content provenance at a higher level. XIA also for federation. With these entities and subsys- needs various trustworthy mechanisms and pro- tems, “slices” can be created on top of the vides network availability even when under shared substrate for miscellaneous research- attack. Finally, XIA defines explicit interfaces defined specific experiments, and end users can between network actors with different roles and “opt in” onto the GENI testbed accordingly. goals. GENI’s research and implementation plan consists of multiple continuous spirals (currently GLOBAL ENVIRONMENT FOR in spiral 3). Each spiral lasts for 12 months. Spi- NETWORK INNOVATIONS (GENI) ral 1 ran from 2008 to 2009; spiral 2 ran from 2009 to 2010; spiral 3 started in 2011. In spiral 1, GENI [7] is a collaborative program supported the primary goals were to demonstrate one or by NSF aimed at providing a global large-scale more early prototypes of the GENI control experimental testbed for future Internet archi- framework and end-to-end slice operation across tecture test and validation. Started in 2005, it multiple technologies; there were five competing has attracted broad interest and participation approaches to the GENI control framework, from both academia and industry. Besides its ini- called “clusters.” tial support from existing projects on a dedicated Cluster A was the Trial Integration Environ- backbone network infrastructure, it also aims to ment based on DETER (TIED) control frame- attract other infrastructure platforms to partici- work focusing on federation, trust, and security. pate in the federation — the device control It was a one-project cluster based on the Cyber- framework to provide these participating net- Defense Technology Experimental Research works with users and operating environments, to (DETER) control framework by the University observe, measure, and record the resulting exper- of Southern California (USC)/ISI, which is an imental outcomes. So generally, GENI is differ- individual “mini-GENI” testbed to demonstrate 30 IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  6. 6. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F federated and coordinated network provisioning. Cluster A particularly aimed to provide usability Users User-network across multiple communities through federation. Trustworthy network operation The project delivered software “fedd” as the implementation of the TIED federation archi- Application tecture providing dynamic and on-demand feder- ation, and interoperability across ProtoGENI, GENIAPI, and non-GENI aggregate. It included Services an Attribute Based Access Control (ABAC) Network-network mechanism for large-scale distributed systems. It created a federation with two other projects: Host Content Services StarBED in Japan and ProtoGENI in the Unit- support support support ed States. Intrinsic Cluster B was a control framework based on security eXpressive Internet protocol PlanetLab implemented by Princeton University emphasizing experiments with virtualized machines over the Internet. By the end of spiral 2, it included at least 12 projects from different Figure 4. XIA components and interactions. universities and research institutes. The results of these projects are to be integrated into the PlanetLab testbed. PlanetLab provided “GENI- We can see that spirals 1 and 2 integrated a wrapper” code for independent development of very wide variety of testbeds into its control an aggregate manager (AM) for Internet enti- framework. Spiral 2 was the second phase aim- ties. A special “lightweight” protocol was intro- ing to move toward continuous experimentation. duced to interface PlanetLab and OpenFlow Key developments include improved integration equipment. Through these mechanisms, other of GENI prototypes; architecture, tools, and ser- projects in the cluster can design their own sub- vices enabling experiment instrumentation; inter- strates and component managers with different operability across GENI prototypes; and capacities and features. researcher identity management. In spiral 3, the Cluster C was the ProtoGENI control frame- goal is to coordinate the design and deployment work by the University of Utah based on Emu- of a first GENI Instrumentation and Measure- lab, emphasizing network control and ment Architecture. Supporting experimental use management. By the end of spiral 2, it consisted of GENI and making it easier to use are also key of at least 20 projects. The cluster integrated goals. Also, more backbone services and partici- these existing and under-construction systems to pants are expected to join in the GENI frame- provide key GENI functions. The integration work for this spiral. included four key components: a backbone based Another notable and unique characteristic on Internet2; sliceable and programmable PCs offered by GENI is that instrumentation and and NetFPGA cards; and subnets of wireless measurement support have been designed into and wired edge clusters. Cluster C so far is the the system from the beginning since the ultimate largest set of integrated projects in GENI. goal of GENI is to provide an open and extensi- Cluster D was Open Resource Control Archi- ble testbed for experimentation with various new tecture (ORCA) from Duke University and Internet architectures. RENCI focusing on resource allocation and integration of sensor networks. By the end of spiral 2, it consisted of five projects. ORCA tried RESEARCH PROJECTS FROM THE to include optical resources from the existing Metro-Scale Optical Testbed (BEN). Different EUROPEAN UNION AND ASIA from other clusters, the ORCA implementation The European Union has also initiated a bundle included the integration of wireless/sensor proto- of research projects on future Internet architec- types. It maintains a clearinghouse for the tures. In this section, we introduce the research testbeds under the ORCA control framework organized under the European Seventh Frame- through which it connects to the national back- work Programme (FP7) along with that in Japan bone and is available to external researchers. and China. Cluster E was Open-Access Research Testbed for Next-Generation Wireless Networks EUROPEAN UNION (ORBIT) by Rutgers University focusing on The European Future Internet Assembly [19] mobile and wireless testbed networks. It includ- (abbreviated FIA as in the United States) is a ed three projects by the end of spiral 2. The collaboration between projects under FP7 on basic ORBIT did not include a full clearing- future Internet research. Currently, the FIA house implementation. Cluster E tried to brings together about 150 projects that are part research how mobile and wireless work can of FP7. These projects have a wide coverage, affect and possibly be merged into the GENI including the network of the future, cloud com- architecture. WiMAX is one of the wireless net- puting, Internet of service, trustworthy informa- work prototypes in this cluster. tion and communication technology (ICT), A more detailed description of the clusters networked media and search systems, socio-eco- and their specific approaches and corresponding nomic aspects of the future Internet, application features can be found in our previous survey domain, and Future Internet Research and [18]. Even more details can be found from GENI Experimentation (FIRE) [10]. The FIA main- project websites and wikis [7]. tains a European Future Internet Portal [20], IEEE Communications Magazine • July 2011 31Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  7. 7. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F Categories Project names (selected) A significant trait of the “Network of the 4AWARD, TRILOGY, EIFFEL, SPARC, SENSEI, Socrates, CHANGE, Future Architectures and Technologies Future” is that the PSIRP, etc. research projects Services, Software, and Virtualization ALERT, FAST, PLAY, S-Cube, SLA@SOI, VISION Cloud, etc. cover a very wide Network Media 3DLife, COAST, COMET, FutureNEM, nextMEDIA, P2P-Next, etc. range of topics and a number of Internet of Things ASPIRE, COIN, CuteLoop, SYNERGY, etc. commercial Trustworthiness ABC4Trust, AVANTSSAR, ECRYPT II, MASTER, uTRUSTit, etc. organizations, including traditional Testbeds FIRE, N4C, OPNEX, OneLAB2, PII, WISEBED, G-Lab, etc. telecommunication Others HYDRA, INSPIRE, SOCIALNETS, etc. companies, participate in Table 2. EU research projects on future Internet. the research consortiums. which is an important web portal for sharing bound to hosts (somewhat similar to the information and interaction among the partici- goal of the NDN project) pating projects. Multiple FIA working groups • To design the network path to be an active have been formed to encourage collaboration unit that can control itself and provide among projects. resilience and failover, mobility, and secure Of these projects, around 90 of them were data transmission launched following the calls of FP7 under the • To devise “default-on” management capa- “Network of the Future” Objective 1.1. They can bility that is an intrinsic part of the network be divided into three clusters: “Future Internet itself Technologies (FI),” “Converged and Optical • To provide dependable instantiation and Networks (CaON),” and “Radio Access and interoperation of different networks on a Spectrum (RAS).” The total research funding single infrastructure. since 2008 is over €390 million. A subset of the Thus, on one hand, 4WARD promotes the projects is shown in Table 2. innovations needed to improve a single network A significant trait of the “Network of the architecture; on the other hand, it enables multi- Future” [17] is that the research projects cover a ple specialized network architectures to work very wide range of topics and a number of com- together in an overall framework. There are five mercial organizations, including traditional task components in the 4WARD research: telecommunication companies, participate in the • A general architecture and framework research consortiums. Since there are a large • Dynamic mechanisms for securely sharing number of projects, we selected a few represen- resources in virtual networks tative ones and explain them in some detail. • “Default-on” network management system; They are all under FP7 based on a series of a communication path architecture with design objectives categorized by the ICT chal- multipath and mobility support lenge #1 of building “Pervasive and Trusted • Architecture for information-oriented net- Network and Service Infrastructure.” works Due to the large number of projects, for the Note that 4WARD is one of many projects architecture research in this article, we selected under the FP7 framework on future Internet a project named 4WARD (Architecture and architecture research. Readers can find more Design for the Future Internet), and for the information on a complete list of the projects testbed we selected FIRE. We selected them from [19]. Some typical projects focusing on dif- due to the fact that FIRE is often deemed the ferent aspects of future architecture are listed in European counterpart project to GENI, and the Table 2. 4WARD project aims at a general architectural level of redesign of the Internet, and we feel that Future Internet Research and Experimenta- it is representative of the rest. It also involves a tion (FIRE) — FIRE [10] is one of the Euro- large number of institutions’ participation and pean Union’s research projects on testbeds and cooperation. is like a counterpart of GENI in the United In the following, we discuss these two pro- States. FIRE was started in 2006 in FP6 and has jects briefly. continued through several consecutive cycles of funding. FIRE involves efforts from both indus- 4WARD — 4WARD [9] is an EU FP7 project try and academia. It is currently in its “third on designing a future Internet architecture led wave” focusing on providing federation and sus- primarily by an industry consortium. The fund- tainability between 2011 and 2012. Note that the ing is over 45 million dollars for a 2-year period. FIRE project’s research is built on the previous The key 4WARD design goals are: work on the GEANT2 (Gigabit European Aca- • To create a new “network of information” demic Networking Technology) project [11], paradigm in which information objects have which is the infrastructure testbed connecting their own identity and do not need to be over 3000 research organizations in Europe. 32 IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  8. 8. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F FIRE has two interrelated dimensions: • To support long-term experimentally driven Experimentally-driven, multi-disciplinary research research on new paradigms and concepts and architectures for the future Internet • To build a large-scale experimentation facil- ity by gradually federating existing and Testbeds VITAL++ WISEBED future emerging testbeds FIRE also expects not only to change the Inter- net in technical aspects but also in socio-eco- PII nomic terms by treating socio-economic requirements in parallel with technical require- Federica ments. OneLab2 A major goal of FIRE is federation, which by definition is to unify different self-governing testbeds by a central control entity under a com- mon set of objectives. With this goal in mind, Support actions the FIRE project can be clustered in a layered way as depicted in Fig. 5. As shown in the figure, it contains three basic clusters. The top-level Figure 5. FIRE clustering of projects. cluster consists of a bundle of novel individual architectures for routing and transferring data. The bottom cluster consists of projects providing AKARI: AKARI means “a small light in the support for federation. In the middle is the fed- darkness.” The goal of AKARI is a clean-slate eration cluster, which consists of the existing approach to design a network architecture of testbeds to be federated. These existing small the future based on three key design princi- and medium-sized testbeds can be federated ples: gradually to meet the requirements for emerging • “Crystal synthesis,” which means to keep future Internet technologies. Documents describ- the architecture design simple even when ing these sub-testbeds can be found on the FIRE integrating different functions project website [10]. • “Reality connected,” which separates the physical and logical structures ASIA • “Sustainable and evolutional,” which means Asian countries such as Japan and China also it should embed the “self-*” properties have projects on future Internet architectures. (self-organizing, self-distributed, self-emer- gent, etc.), and be flexible and open to the Japan — Japan has broad collaborations with future changes both the United States and European Union AKARI is supposed to assemble five subar- regarding future Internet research. It participates chitecture models to become a blueprint in PlanetLab in the United States, and the NWGN: testbed in Japan is also federated with the Ger- • An integrated subarchitecture based on a man G-Lab facility. The Japanese research pro- layered model with cross-layer collabora- gram on future Internet architecture is called tion; logical identity separate from the data New Generation Network (NWGN) sponsored plane (a kind of ID/locator split structure) by the Japan National Institute of Information • A subarchitecture that simplifies the layered and Communications Technology (NICT). The model by reducing duplicated functions in Japanese research community defines the clean- lower layers slate architecture design as “new generation” and • A subarchitecture for quality of service the general IP-based converged design as “next (QoS) guarantee and multicast generation” (NXGN) design. NWGN started in • A subarchitecture to connect heterogeneous June 2010 and expects to change the network networks through virtualization technologies and Internet community with broad • A mobile access subarchitecture for sensor impact in both the short term (to 2015) and long information distribution and regional adap- term (to 2050). Like the projects in the United tive services States and European Union, NWGN consists of AKARI is currently in the process of a proof- a series of sub-projects collaborated on by of-concept design and expects to get a blueprint academia and industry. The sub-projects range in 2011. Through systematic testbed construction from architecture designs, testbed designs, virtu- and experimentations, it aims to establish a new alization laboratories, and wireless testbeds to architecture ready for public deployment by data-centric networking, service-oriented net- 2016. works, advanced mobility management over net- JGN2plus and JGN-X: JGN2plus is the work virtualization, and green computing. Rather nationwide testbed for applications and networks than enumerating all projects, we briefly discuss in Japan, and also the testbed for international the architecture project AKARI [12] and the federation. It includes broad collaboration from testbed projects JGN2plus [13] and JGN-X (JGN both industry and academia. It evolved as JGN, stands for Japan Gigabit Network). The reason migrated to JGN II in 2004, and then to we selected these projects is similar to the reason JGN2plus in 2008. From 2011, the testbed is we selected FIA and GENI. AKARI is so far the under JGN-X, which targets to be the real biggest architectural research project in Japan; NWGN testbed to deploy and validate AKARI JGN2plus and JGN-X are the testbed research research results. JGN2plus provides four kinds counterparts to GENI and FIRE. of services: IEEE Communications Magazine • July 2011 33Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  9. 9. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F • Layer 3 (L3) IP connection DISCUSSIONS AND PERSPECTIVES Any architecture that • L2 Ethernet connection • Optical testbed Having presented a variety of research projects, requires investment • Overlay service platform provisioning we find that there are several issues worth dis- without immediate There are also five subtopics in the research cussing. In this section, we give our perspective payoff is bound to of JGN2plus: regarding these issues. Of course, there is no • NWGN service platform fundamental tech- agreement among researchers regarding these fail. Of course, the nologies perspectives, and none is implied. payoff will increase • NWGN service testbed federation technolo- Clean-slate vs. evolutionary: Clean-slate gy designs impose no restriction and assumption as the deployment • Middleware and application of lightpath on the architectural design. The key idea is not of the new NWGN to be subjected to the limitations of the existing technology increases, • Component establishment technologies for Internet architecture. It is also called “new gen- NWGN operation eration” by Japanese and Chinese researchers. economies of scale • Verification of new technologies for inter- While the architectures can be revolutionary, reduce the cost, and national operation their implementation has to be evolutionary. JGN2plus expects to create collaboration among Today, the Internet connects billions of nodes eventually the old industry, academia, and government for NWGN and has millions of applications that have been architecture experiments. It also aims to contribute to human developed over the last 40 years. We believe deployed base will resource development in the ICT area via these any new architecture should be designed with experiments. this reality in mind; otherwise, it is bound to diminish and fail. Legacy nodes and applications should be disappear. China — The research projects on future Inter- able to communicate over the new architecture net in China are mostly under the 863 Program, without change (with adapter nodes at the 973 Program, and “12th Five-Year Plan Pro- boundary), and new nodes and applications jects” administrated by the Ministry of Science should similarly be able to communicate over and Technology (MOST) of China. Currently the existing Internet architecture. Of course, there are several ongoing research projects, the services available to such users will be an which include: intersection of those offered by both architec- • New Generation Trustworthy Networks tures. Also, the new architecture may provide (from 2007 to 2010) adaptation facilities for legacy devices at their • New Generation Network Architectures boundary points. Various versions of Ethernet (from 2009 to 2013) are good examples of such backward compati- • Future Internet Architectures (from 2011 to bility. Some variations of IP are potential exam- 2015) ples of missing this principle. Project 1 was still IP - network research instead New architecture deployment will start in a of a clean-slate future Internet. It consists of very small scale compared to the current Inter- research sub-projects on new network architec- net. These early adopters should have economic ture, next generation broadcasting (NGB), new incentives for change. Any architecture that network services, a national testbed for new requires investment without immediate payoff is generation networks and services, new routing/ bound to fail. Of course, the payoff will increase switching technology, a new optical transmis- as the deployment of the new technology increas- sion network, and low-cost hybrid access equip- es, economies of scale reduce the cost and even- ment. tually the old architecture deployed base will Besides the research projects on future Inter- diminish and disappear. net architecture, there are also ongoing research Integration of security, mobility, and other projects for building a China Next Generation functionalities: It is well understood and agreed Internet (CNGI) testbed. It is based on the pre- that security, mobility, self-organization, disrup- vious infrastructure network testbed of the tion tolerance, and so on are some of the key China Education and Research Network (CER- required features for the future Internet. How- NET [14] and CERNET2 [15]) and the China ever, most of the projects, even for those col- Science and Technology Network (CSTNET). A laborative ones like in FIA program, put more terabit optical, terabit WDM, terabit router plus emphasis on a specific attribute or a specific set IPTV testbed called (3T-NET) was also of problems. It seems to be a tough problem to announced on July 2009 as NGB. The testbed handle many challenges in a single architecture projects are mostly industry oriented with specif- design. Currently, for the collaborative projects ic interest in IPv6 related protocols and applica- such as FIA, they are trying to integrate miscel- tions. laneous previous research results into a coher- We observed that the current future Internet ent one trying to balance some of the issues. architecture research in China leans heavily Although different projects have different toward IPv6 related testbed, which is relatively emphases, it is beneficial to create such diversi- short-term. To some extent, it reveals the pain ty and allow a bunch of integrated architectures China felt due to the collision between the to potentially compete in the future. However, extreme shortage of IPv4 address space in we believe that there is still a long way to go China and the ever expanding demands from before there is a next-generation architecture increasing customers and novel services. unifying these different lines of designs. For Longer-term research projects on innovative example, we observe that the four U.S. FIA architectural research are still in the cradle projects concentrate on four different specific compared to those of the United States and issues. Self-certifying and hash-based addresses European Union. are effective tools for security. However, securi- 34 IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  10. 10. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F ty needs much more consideration on both our perspective, such challenges also mean a micro and macro scopes. Content- and informa- valuable opportunity for research on sharing and Internet 3.0, which is tion-centric features are also important trends, virtualization over diverse platforms. but how to integrate these differing require- Service delivery networks: The key trend a project in which ments and resulting architectures is still a pend- driving the growth of the Internet over the last the authors are ing problem. We expect that more integration decade is the profusion of services over the involved, includes research will be required when such issues Internet. Google, Facebook, YouTube, and simi- emerge in the future. It is therefore desirable lar services form the bulk of Internet traffic. developing an open for different projects to create some synergy for Cloud computing and the proliferation of mobile and secure service the integration process. devices have lead to further growth in services Architectures built around people instead of over the Internet. Therefore, Internet 3.0 [16], delivery network machines: It has been widely realized that the which is a project in which the authors of this architecture. This will usage pattern of the Internet has changed, and article are involved, includes developing an open allow telecommuni- the trend of building future Internet architecture and secure service delivery network (SDN) archi- around the contents, data, and users seems to be tecture. This will allow telecommunication carri- cation carriers to justifiable and promising in the future. Design ers to offer SDN services that can be used by offer SDN services goal changes naturally lead to design principle many application service providers (ASPs). For changes. Different patterns may emerge without example, an ASP wanting to use multiple cloud that can be used by any further synthesis. Current existing projects computing centers could use it to set up its own many application on future Internet architectures sort out differ- worldwide application-specific network and cus- service providers. ent principles according to their own design tomize it by a rule-based delegation mechanism. emphases. From our perspective, it is essential These rules will allow ASPs to share an SDN and important to form a systematic and compre- and achieve the features required for widely dis- hensive theory in the research process rather tributed services, such as load balancing, fault than designing based only on experiences. It may tolerance, replication, multihoming, mobility, take several continuous spirals between theoreti- and strong security, customized for their applica- cal improvement and practical experience to tion. One way to summarize this point is that achieve a sound architecture. We believe more service delivery should form the narrow waist of research in this area may be desirable and mean- the Internet (Fig. 1), and content and IP are ingful for future Internet research. special cases of service delivery. Interfaces among stakeholders: Future Inter- net architectures are required to provide extensi- ble and flexible explicit interfaces among SUMMARY multiple stakeholders (users, Internet service In this article, we present a survey of the current providers, application service providers, data research efforts on future Internet architectures. owners, and governments) to allow interaction, It is not meant to be a complete enumeration of and enforce policies and even laws. A typical all such projects. Instead, we focus on a series of example is Facebook, which creates a complex representative research projects. Research pro- situation for data, privacy, and social relation- grams and efforts from the United States, Euro- ships. Societal and economic components have pean Union, and Asia are discussed. By doing become indispensible factors in the future Inter- this, we hope to draw an approximate overall net. The transition from the academic Internet picture of the up-to-date status in this area. to a multifunctional business-involved future Internet puts much higher requirements on the architectural supports to regulate and balance REFERENCES the interests of all stakeholders. In both techni- [1] NSF Future Internet Architecture Project, http://www. cal and non-technical aspects, the future Inter- nets-fia.net/. ______ [2] NSF NeTS FIND Initiative, http://www.nets-find.net. net architectures are required to provide [3] Named Data Networking Project, http://www.named- ___________ extensible and flexible explicit interfaces among data.net. _____ multiple actors to allow interaction, and enforce [4] MobilityFirst Future Internet Architecture Project, policies and even laws. The deep merging of the http://mobilityfirst.winlab.rutgers.edu/. [5] NEBULA Project, http://nebula.cis.upenn.edu. Internet into everyone’s daily life has made such [6] eXpressive Internet Architecture Project, endeavors and efforts more and more urgent http://www.cs.cmu.edu/~xia/. and important. From our perspective, significant [7] Global Environment for Network Innovations (GENI) research efforts are still needed in aspects such Project, http://www.geni.net/. [8] OpenFlow Switch Consortium, http://www.open- ___________ as economics, society, and laws. flowswitch.org/. ________ Experimental facilities: Most of the current [9] The FP7 4WARD Project, http://www.4ward-project.eu/. testbeds for future Internet architecture research [10] FIRE: Future Internet Research and Experimentation, in different countries are results of previous http://cordis.europa.eu/fp7/ict/fire/ [11] GEANT2 Project, http://www.geant2.net/. research projects not related to future Internet [12] AKARI Architecture Design Project, http://akari-pro- _________ architectures. The networks use different tech- ject.nict.go.jp/eng/index2.htm ________________ nologies and have different capabilities. [13] JGN2plus- Advanced Testbed Network for R&D, Although the federation efforts are meaningful, http://www.jgn.nict.go.jp/english/index.html. [14] China Education and Research Network, they may be restricted in both manageability and http://www.edu.cn/english/. capability by such diversity. Testbeds from differ- [15] CERNET2 Project, http://www.cernet2.edu.cn/ ent countries are also generally tailored or spe- index_en.htm. _______ cialized for the architectural design projects of [16] Internet 3.0 project, http://www1.cse.wustl.edu/ ~jain/research/index.html. ______________ those countries, with different features and [17] The Network of the Future Projects of EU FP7, emphases. Federation and creating synergy http://cordis.europa.eu/fp7/ict/future-networks/ among such testbeds may be challenging. From home_en.html. ________ IEEE Communications Magazine • July 2011 35Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F
  11. 11. Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F [18] S. Paul, J. Pan, and R. Jain, “Architectures for the student in the Department of Computer Science and Future Networks and the Next Generation Internet: A Engineering at Washington University. His primary Survey,” Comp. Commun., U.K., vol. 34, issue 1, 15 research interests are in the area of future Internet Jan. 2011, pp. 2–42. architectures. [19] Future Internet Assembly, http://www.future- ____________ internet.eu/home/future-internet-assembly.html. _________________________ RAJ JAIN [F] (jain@cse.wustl.edu) is a Fellow of ACM, a win- __________ ner of the ACM SIGCOMM Test of Time award and CDAC- BIOGRAPHIES ACCS Foundation Award 2009, and ranks among the top 50 in Citeseer’s list of Most Cited Authors in Computer Sci- J IANLI P AN [S] (jp10@cse.wustl.edu) received his B.E. in ____________ ence. He is currently a professor in the Department of 2001 from Nanjing University of Posts and Telecommunica- Computer Science and Engineering at Washington Universi- tions (NUPT), China, and his M.S. in 2004 from Beijing Uni- ty. Previously, he was one of the co-founders of Nayna versity of Posts and Telecommunications (BUPT), China. He Networks, Inc., a next-generation telecommunications sys- is currently a Ph.D. student in the Department of Computer tems company in San Jose, California. He was a senior con- Science and Engineering at Washington University in Saint sulting eEngineer at Digital Equipment Corporation in Louis, Missouri. His current research is on future Internet Littleton, Massachusetts, and then a professor of computer architecture and related topics such as routing scalability, and information sciences at Ohio State University, Colum- mobility, mulithoming, and Internet evolution. His recent bus. He is the author of Art of Computer Systems Perfor- research interests also include green building in the net- mance Analysis, which won the 1991 Best-Advanced working context. How-to Book, Systems award from Computer Press Associ- ation. His fourth book, High-Performance TCP/IP: Concepts, S UBHARTHI P AUL [S] (pauls@cse.wustl.edu) received his ____________ Issues, and Solutions, was published by Prentice Hall in B.S. degree from the University of Delhi, India, and his November 2003. Recently, he co-edited Quality of Service Master’s degree in software engineering from Jadavpur Architectures for Wireless Networks: Performance Metrics University, Kolkata, India. He is presently a doctoral and Management, published in April 2010. 36 IEEE Communications Magazine • July 2011Communications A BMaGS IEEE Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page E F