| NSF Org: |
CNS Division Of Computer and Network Systems |
| Recipient: |
|
| Initial Amendment Date: | April 26, 2012 |
| Latest Amendment Date: | April 26, 2012 |
| Award Number: | 1219557 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Darleen Fisher
CNS Division Of Computer and Network Systems CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | May 1, 2012 |
| End Date: | April 30, 2016 (Estimated) |
| Total Intended Award Amount: | $483,103.00 |
| Total Awarded Amount to Date: | $483,103.00 |
| Funds Obligated to Date: |
|
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
77 MASSACHUSETTS AVE CAMBRIDGE MA US 02139-4301 (617)253-1000 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
32 Vassar Street Cambridge MA US 02139-4309 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): | Networking Technology and Syst |
| Primary Program Source: |
|
| Program Reference Code(s): | |
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
This research is intended to build on, complement, and enhance the NSF future Internet architecture programs through cross-cutting and integrative comparative analysis of architectural proposals. Over the last 6 years, the National Science Foundation has created a series of programs focused on new Internet architecture and the embedding of networking into the larger social context: these include the Future Internet Design (FIND) program, the Network Science and Engineering (NetSE) program, and Future Internet Architecture (FIA) solicitation. This sequence of programs address a bold objective?to look into the future and challenge the research community to design the global network that the world will need fifteen years from now. In the current stage of the overall program, a number of teams have been funded to develop integrated proposals for the design of a future Internet. Each of these projects can be expected to produce a coherent justification for one or another approach to the architecture of a future network, but the overall outcome of the project should be much more than that: it should also result in overarching insights into design preferences and alternative approaches, general lessons about network system design, and a more systematic rationale for designing networks and concluding that one response to requirements is preferable to another. The FIA project should push the frontiers of principled network design, and this phase of the programs has laid the foundation for this exciting step.
This award addresses these larger project goals through a program of research that is complementary to the current awards, by looking across the projects (and other relevant architectural work) to seek more integrative and overarching conclusions. The challenge is difficult, given the diversity of goals and requirements that might be posed for a system as complex as the Internet, as well as the diversity of design approaches. In contrast to much of computer science research, this agenda is focused on comparative analysis, not engineering design. As appropriate, the research will be done in collaboration with the individual projects through cross-project studies on specific topics related to network design. The investigator will seek to involve others in the broader community who have given thought to high-level issues in network architecture, and will carry out his own research into principals and methods for network design. The investigator will also participate in the FIA PI meetings, since those meetings will explore specific topics in a cross-project way.
Intellectual merit: This NSF program should contribute to the long-rang goal of making network architecture design more methodical and systematic, and less an art; this research sets that goal as its highest-level objective. This research has the goal of discovering and articulating fundamental insights and conclusions with respect to issues in network design, results that arise from looking across the various different projects with their different approaches. Topics will include core network functions such as addressing and forwarding, dealing with important requirements such as security and availability, new areas of architectural study such as information-centric networks, and the relation between architectural design choices and larger social, economic and contextual requirements.
Broader impact: This research will provide a structure and direction for future research in network design, and provide a new and more principled way of teaching about networks. Most courses in networking today deal with the subject by teaching large numbers of facts: the formats of headers, operation of protocols, and so on, with less emphasis on the ?why? of design. Prior working papers of the investigator have been used in at least one advanced course on network architecture, and one ambition of this research is to develop materials suitable to the teaching of network architecture. The outcomes of this research can also help with the transfer of results from the FIA program into the commercial world. The investigator will also continue to provide support to the community toward the program goals.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
This effort was part of the the NSF Future Internet Architecture program, and was intended to build on, complement, and enhance that program through cross-cutting and integrative comparative analysis of architectural proposals. The goal of the Future Internet Architecture program was to stimulate long-term thinking about the future of the Internet by challenging the research community to propose what an Internet of fifteen years from now might be if the design were not constrained by the details of the present Internet--if the design could be conceived "clean slate". NSF funded a number of teams to develop integrated proposals for the design of a future Internet. Each of these projects produced a justification for their approach to the architecture of a future network, but the overall outcome of the project was also intended to result in overarching insights into design preferences and alternative approaches, general lessons about network system design, and a more systematic rationale for designing networks.
This award addressed these larger program goals through research that was complementary to the awards that supported design efforts, by looking across the projects (and other relevant architectural work) to seek more integrative and overarching conclusions. In contrast to much of computer science research, this effort was focused on comparative analysis, not engineering design. This research had the goal of discovering and articulating fundamental insights and conclusions with respect to issues in network design, results that arise from looking across the various different projects with their different approaches. The framework developed to organize this analysis started with the core network functions of addressing and forwarding, and then covered various important requirements a network architecture must address, including security and availability, longevity, economic viability in the real world, network management and control, and the relation between architectural design choices and larger social, policy and contextual requirements.
Some of the key conclusions of the study are summarized below:
The word "security" is so general that it is aspirational in character--it does not actually specify what problem is to be solved. The high-level goal must be broken into more specific challenges, such as protecting the network itself from attack, protecting the communication among parties from attack by other parties, and protecting one party to a communication from being attacked by another. When viewed at this level of detail, some of these goals turn out to be potentially in conflict with each other--security is not the perfection of any single attribute such as privacy, but rather the balancing of different consideration that represent the interests of different parties. The research also clarified the role that different actors must take in improving security: the network designer, the designer of applications, the operating system and so on.
Availability is one of the most important requirements on a network, but the field lacks a general theory of availability. This research suggests a very simple formulation of availability, which is that for a network to be available, the user must not be required to depend on any components that are not trustworthy. This almost simplistic framing turns out to have significant implications for system design.
One measure of success for the Internet is that it has survived for over thirty years--much longer than most computing artifacts. Longevity may be an important requirement for an Internet, but there are a number of different theories of longevity, some in apparent conflict. This research identified at least 15 different theories of longevity.
For a network design to be a success, it must not just meet technical requirements, but also be successful in the real world, which means that it must be economically viable. This research developed a framework to assess economic viability of an architecture, using the work of economist Ronald Coase to map the modularity of the architecture to the resulting industry structure, and the features in an architecture that may facilitate or block the flow of money among the economic actors in the Internet.
Network management and control has been the subject of a great deal of research (for example research in different congestion control algorithms) but has perhaps been under-studied from an architectural perspective--what elements in the packet header and what state in the router can enable classes of control. Architectural proposals that specify more state in the router (per flow or per packet) will not just have different data forwarding capabilities but different capabilities to implement control algorithms.
The results of this research are the subject of a book that should be published some time in 2017, and which is available as a pre-publication now for the reader interested in further information. The pre-publication version, which elaborates the points summarized here, can be obtained at http://groups.csail.mit.edu/ana/People/DDC/archbook.
Last Modified: 10/31/2016
Modified by: David D Clark
Please report errors in award information by writing to: awardsearch@nsf.gov.
