by François Baccelli and Jon Crowcroft
The Internet has become critical to everyday life in domains as diverse as education, health, defence, commerce, travel and entertainment. The Internet was not designed for its current level of usage, and there is a need for simple constructs to allow the network to do better in terms of security, mobility and quality of service, among other things.
A number of global and local programmes of research (in the US, EU and further afield, eg Asia) are looking at future network architectures and building testbeds to evaluate new protocols and systems based on these new ideas. The most notable, in the US are the NSF Find programme and the GENI project to build infrastructure. In Europe we find a number of projects, with some interesting high-level thinking (eg the Eiffel project) and a set of testbed initiatives under the FIRE Programme.
While some researchers think there is a need for a 'clean slate' design of the network of the future, much successful work is evolutionary rather than revolutionary, and this can be seen in the articles in this issue.
For this special theme on Future Internet Technology, we issued a call for papers and invited four well known researchers in addition, to cover the following topics as well as possible:
Architectures and Infrastructures
Future core networks will leverage IP over simple, super-fast optical core networks. A major trend is that of virtualization, featuring the construction of optimized virtual networks that answer the needs of a collection of users or applications. A wide variety of wired and wireless access networks are available or being developed: FTTH/FTTO in the wired domain, and 3G/CDMA, LTE, WiFi, WiMax and Satellite in the wireless setting.
Internet Modelling, Simulation and Measurements
It is quite important to comprehend how the complex systems that we build actually operate. This requires basic research on network modelling and simulation, eg, in order to derive the fundamental laws on network dynamics and control or to evaluate the ultimate capacity of self-organized wireless networks. It also requires advances in network measurements: traffic statistics, Internet probing and measurement, network inference and detection of anomalies and attacks.
Internet Algorithms and Software
The future of the Internet will require a wide range of computer science tools: verification; distributed algorithms (eg for consensus, and election and epidemic diffusion); resource management algorithms: resource allocation and scheduling; database algorithms: content storage, update and retrieval, content replication and consistency; search engines and the semantic Web.
A key cost in networking is the operations and management overheads. The scalability of the Internet is well known, but as distributed applications proliferate, a more autonomic approach is increasingly required. Spontaneous and self-organized networks emerge both in the wireless setting (eg with Wifi meshes or infrastructure-less wireless networks [MANETS]) and in the wired network setting (eg in peer to peer),
Embedded Internet, Internet of Things
The Internet started by connecting computers and users to information, and then went on to connect users to each other with audio, video, games and social networking tools. Now the rapid evolution of pervasive, embedded networked devices means that we connect various types of devices to each other. There are far more computers in the world embedded in everyday objects (cars, domestic appliances etc) than there are on desktops. The interconnection of the Internet with the physical world through sensors and agents and the tagging of industrial production by RFIDs will lead to new traffic and architecture challenges, with possibly hundreds of billions of new devices that will collect information and will have to be upgraded and managed remotely and conveniently. This will require new paradigms for routing, search, naming, maintenance, data survival etc.
The Internet of things extends to cars and other vehicles. Here the key potential is in safety: ABS in cars could communicate road surface conditions to following vehicles, setting speed and braking reactions sooner rather than too late. Traffic management and pollution sensing on cars can also make use of networking between cars, and from cars to roadside infrastructure. Tracking goods in transit on the road would allow logistics companies to optimize their freight operations, saving time and energy and perishable goods.
Media Internet/Media-Driven Networks
There is an important diversification of the nature of the content transported across the Internet: initially it was files, then real-time games, video, telephony and whiteboards, and now TV, video on demand etc. There has also been diversification of the localization of contents: with each user potentially a content producer (peer-to-peer applications). New interactions with data are appearing, as in Web 2.0 or the semantic Web, and applications continue to evolve and require new systems, measurements and management tools.
Since its beginnings in 1992, the World-Wide Web has offered remote transactions for goods and services. Recently we have seen a rapid growth in the number of attacks on identity, since acquiring such information allows miscreants to commit fraud that is hard to detect.
Energy saving and the Internet
Estimates vary, but the Internet and all its services consume something on the order of 4% of the energy in the developed world. Only simple measures are required to improve this by a factor of two. Furthermore, the Internet, as we have discussed above, can be used to monitor and control external devices (things, vehicles, services) and significantly reduce their unnecessary power consumption. Figures as high as 30% have been quoted for possible national savings of energy, if unused devices in all homes could be remotely turned off. The investment necessary to achieve this is relatively low, with what seems like a very big potential return.
The Internet Services
We are moving from a Web of documents to a Web of services and Web of knowledge. This has triggered an explosion of new applications such as SecondLife, FaceBook and LinkedIn. And there is more to come, with augmented reality, virtual worlds, real-time games and telepresence. New concepts such as service orchestrations are also emerging, and with them, new business models.
INRIA and École Normale Supérieure, France
François Baccelli is a member of GRIF (Groupe de Réflexion sur l'Internet du Futur), together with Vivek Badrinath (Orange), François Bourdoncle (Exalead), Christophe Diot (Thomson), Serge Fdida (Université Paris 6) and Daniel Kofman (Télécom Paristech). The GRIF was launched by the French Minister of Industry.
University of Cambridge, UK
Articles in this special theme
This special theme of ERCIM News contains 29 articles that are very much in line with the themes above:
Setting The Context
- The invited article 'Starting the Debate: Agreeing on Disagreements' by Dirk Trossen, B.T., U.K., is a meta-discussion note, on how to trigger debate. As such, it is quite interesting as a methodology for encouraging architectural thinking in the long term.
- 'The Internet Engineering Task Force and The Future of Internet' by Emmanuel Baccelli, Thomas H. Clausen and Philippe Jacquet, INRIA Saclay, France, provides a useful reminder of the processes by which standards are created, and an analysis of how future research might be delivered in the short to medium term.
- 'Network Virtualization, a Perspective' by Anja Feldman, Mario Kind, Olaf Maennel, Georg Shaffrath and Christoph Werle: Virtualization, briefly defined above, can be seen as a way to overcome the ossification of the Internet, namely the general idea that one should not change a system or architecture that works well. The technical challenges and the business models associated with overlay networks and more generally with virtualization are thoroughly discussed in this invited paper.
- 'Y-COMM: A New Architecture for Mobile Heterogeneous Communications' by Glenford Mapp suggests that the symmetric architecture we are used to in the Internet is perhaps not well adapted to wireless, and proposes an alternative, useful for promoting discussion (cf the EU ANA Project).
Modelling, Measurement and Management
- 'Epidemic Information Dissemination' by Laurent Massoulie (Invited paper): Epidemic algorithms became popular at the turn of the century with Peer-to-Peer applications. They are now being investigated by major companies designing efficient distributed live streaming mechanisms over the Internet. This beautiful line of research combines the design of innovative algorithms and the modelling of their execution over a large network, which allows one to prove their quasi-optimality.
- Understanding traffic entails delivering information efficiently to humans so that they can decide on actions to take. The article 'iMyNetScope- a Platform for Network Traffic Visualization and Analysis' by Pavel Minarik discusses one approach to this, and presents one particularly interesting topic for today - the element of Cyber Defence.
- 'Unified Access to Internet Measurement Data' by Felix Strohmeier, Martin Nilsson and Demetres Antoniades: like the previous article, this touches on a timely topic - making sure we can all (as researchers or operations) access data easily (see also www.crawdad.org for an example of traffic archives in a standard form). Since the Internet is a federation of a vast number of networks (300 ISPs in the UK alone), we need clear and standard methods for exchanging data about operations. This applies equally to research networks such as those in FIRE.
- 'Privacy Aware Network Monitoring' by Peter Dorfinger, Carsten Schmoll and Felix Strohmeier describes a problem relevant to the research and operations community. Much work on the future Internet requires monitoring of research, but at the same time, real users require privacy. This article reminds us how important and difficult it is to reconcile these two aspects. Indeed, in gathering data about network use there are legal requirements relating to intercept laws and user privacy that must be met.
- 'Breaking the Weakest Link: Becoming a Trusted Authority on the Internet' by Marc Stevens is a reminder of the danger of security failures. The fact that even the basic technology of security is sometimes flawed is interesting. Systems are never perfect, and this is never more true than for security: we can never prove a system is secure, but only discover (eventually) when it no longer is!
Correctness and Robustness (and self-managed to some extent)
- 'ResiliNets: Resilient and Survivable Networks' by David Hutchison and James P.G. Sterbenz: As discussed above, we need secure networks that are efficient and well managed. We also depend on these networks, and since individual components are never 100% immune from faults and failures, we need techniques to make the overall system more resilient than its components.
- 'Standardized Testware for Internet-Based Telecommunication Services' by Bostjan Pintar, Axel Rennoch, Peter Schmitting and Stephan Schulz: As Internet protocols grow in number and complexity each year, we need assurance that systems will behave correctly. This article outlines industrial solutions on the correct operation of the software systems (protocols) that make the Internet and its telecommunication infrastructure work.
- 'Network Description Tools and Standards' by Freek Dijkstra, Jeroen van der Ham and Ronald van der Pol: Setting up high-speed network connections - light paths - is still a manual effort taking two to three weeks. The network engineers need a clear picture of the network topology in order to plan and configure light paths. The Network Markup Language (NML) standardizes network topology and state information. The University of Amsterdam and SARA Computing & Networking Services in the Netherlands are contributing to this effort, with the ultimate goal being to automatically set up and manage light paths.
Novel Applications, especially multimedia
- 'Delay-Tolerant Bulk Internet Transfers' by Nikolaos Laoutaris and Pablo Rodriguez (invited paper): Using disruption-tolerant networking (DTN) techniques, but on the Internet rather than on challenged networks, produces very interesting results in terms of capacity. Throughput can be significantly increased for some applications, and there are some very well-motivated examples, such as backing up information between large data centres, using spare capacity in a smart and autonomic way.
- 'Resource Management for IPTV Distribution' by Henrik Abrahamsson and Per Kreuger: IPTV on AT&T and Telefonica's networks now has in excess of 12 million users, each with hundreds of channels of broadcast quality content. This makes for non-trivial deployments, and the importance of resource management for the various different technologies is unquestionable.
- 'VISTO: Visual Storyboard for Web Video Browsing, Searching, and Indexing' by Marco Pellegrini describes exciting work on future Internet killer applications.
- 'The New Role of Humans in the Future Internet' by Daniel Schall and Schahram Dustdar: Current service-oriented architectures typically orchestrate Web services. This article addresses architectures that combine Web and human services, where the latter are not limited to service consumption but may also undertake service production. The main emphasis is on the interfaces required in such architectures.
In the Home (Internet of Things, and energy!)
- 'Orchestrating your Surroundings' by Pau Giner, Carlos Cetina, Joan Fons and Vicente Pelechano, UP Valencia, Spain, is an example of the 'Internet of Things' being at the centre of future Internet visions.
- 'An Autonomic Home Networking Infrastructure' by Thomas Luckenbach, Mario Schuster and Marc-Oliver Pahl: in line with the previous article but articulated with a vision of autonomic computing, this is on the important topic of home net automation. Home users are non-technical: they really do not want to become network managers, operators or system administrators. It is thus essential to make home networks 'plug and play'.
- 'Load-Balancing Energy-Usage of Household Appliances' by Lennart E. Fahlen: Energy is a highly relevant topic, as described above, and using the Internet to control consumer electronics in the household is a promising way to reduce unnecessary consumption.
On the Road, Vehicular
- 'The Internet of Vehicles or The Second Generation of Telematic Services' by Markus Miche and Thomas Michael Bohnert: Like the Internet of Things, vehicular networking is just taking off. There are car-to-car and car-to-infrastructure communications standards as well as integration of satellite navigation, traffic monitoring, tracking, and cellular map systems.
- 'Web-Enabled Tracking Operations in Distributed Supply-Chains' by Zsolt Kemény and Elisabeth Ilie-Zudor: Using the Web to track logistical information about goods in the real world is already a huge business. This is a very useful article on the topic, which also relates to the previous article.
Awareness, Identity and Society
- 'Knowledge-Based Collaboration Patterns in Future Internet Enterprise Systems' by Gregoris Mentzas and Keith Popplewell is supporting virtual community patterns, a bit like eScience community work.
- 'Experiences from the Public Safety Arena for the Future Internet' by Socrates Varakliotis, Peter Kirstein and Steve Hailes: Since the Boxing Day tsunami and Hurricane Katrina, it has become clear that the Internet might be better used in support of communications during disasters. This is a useful note on work in the area.
- 'With Joint Forces - Establishing Situation Awareness for the Future Internet' by Tanja Zseby and Thomas Hirsch: Like Mentzas' article, we can build systems to track patterns of use. This can be used to support collaboration or to detect misbehaviour. Increasingly, though, agencies are beginning to monitor social networks for this purpose.
- 'Social Networking for a Pervasive Future Internet: the SOCIALNETS Project' by Stuart M. Allen, Marco Conti, Andrea Passarella and Roger M. Whitaker: A timely article on an increasingly important topic. This outlines several new aspects of social networking that our readership should know about.
- 'Owner-Centric Networking: A New Architecture for a Pollution-Free Internet' by Claude Castelluccia and Mohamed Ali Kaafar proposes a new Owner-Centric Networking architecture that provides individuals control over their contents. This architecture would considerably improve privacy on the Internet by limiting data pollution.
- 'Semantic Web for e-Commerce' by Bernd Gruber: The Web is the basis for most new applications (Web mail, photo sharing, social nets, maps) and is the basis for commerce, which drives real economies.
- 'Management of Future Mesh-Based Radio Access Networks' by Vangelis Angelakis, Vasilios Siris and Apostolos Traganitis: Community mesh wireless nets are becoming pretty widespread in some areas as an alternative to expensive cellular data services. Automating management is important; related work at Intel and Microsoft in the last couple of years has covered some aspects (AP selection) but the topic continues to be relevant.
- 'Testing Mobile Data Applications on Smartphoes with SymPA' by Almudena Díaz and Pedro Merino discusses further the shortcomings of operating system platforms for data networking on cell phones and related devices. This is important, since the largest growth in Internet-capable devices is currently in this area.