IRTF Update

By: Aaron Falk

Date: September 7, 2009

line break image

Since IETF 74 the Internet Research Task Force (IRTF) has been working on IRTF RFC Stream desired copyrights. The intent is to maximize commonality with the IETF process while permitting unlimited derivative works (with attribution) or no derivative works at all. Issues to be addressed include working with the IETF Trust and establishing a correct Internet Draft boilerplate, among others. The topic is being addressed on the RFC-interest mailing list.

No new RFCs have been published, but five documents are on hold pending resolution of the aforementioned copyright issues.

A new research group (RG) has been chartered called the Public Key Next-Generation Research Group (PKNG), chaired by Paul Hoffman. The group will be looking into alternate certificate formats, semantics, and public-key services that could eventually replace Public-Key Infrastructure (X.509), if deployed. Discussions for an RG on network virtualization continue.

During the IETF technical plenary in Stockholm, I gave a short overview of a few active RGs. The overview was intended to introduce people in the IETF to the work going on in the IRTF and to encourage more participation. The following two sections are introductions to the Host Identity Protocol RG and the Internet Congestion Control RG, as presented during the IETF 74 technical plenary.

Host Identity Protocol Research Group (hiprg)

At present, IP addresses serve two roles on the Internet: the first is to identify the host during communication, and the second is to locate the host within the Internet routing system. This overloading of the address can cause failures in transport protocols such as TCP or in applications when the IP address changes-for example, because of host mobility.

HIP is a host-based protocol devised to split identifiers from locators-at roughly the endpoint sublayer of the IP layer. HIP identifies host using a self-generated public-private key. ESP (Encapsulating Security Payload) encryption and a key exchange protocol provide secure support for mobility and multihoming. RFC 4423 describes the HIP architecture in more detail. The first commercial HIP products have been announced.

The IRTF HIP RG was established in parallel with the IETF HIP working group (WG) in 2004 and has since matured and migrated several drafts to the HIP WG-on such topics as network address translation traversal, native API definition, certificates, and support for legacy applications. The focus of the RG is on discussing ideas that are not yet ready for the IETF process. The RG also provides a forum to present HIP extensions and experiments. The RG is now working on an experiment report summarizing experiences with HIP use.

The RG is currently discussing a number of HIP extensions:

  • The use of HIP for object identification in what is called the Internet of Things (see page 20)
  • Hierarchical host identity tag and host identity revocations
  • ID-to-locator resolution using distributed hash table (DHT) and DNS
  • Mobile router extensions

Future plans for the HIP RG include the adoption of a number of Internet-Drafts as RG items based on the sustained interest of participants. The RG is also interested in helping the WG move HIP to proposed standards, and IETF participants are encouraged to install and try HIP on their computers and to provide feedback. Current HIP implementations can be found at:,

A more detailed article on HIP was published in the IETF Journal in March 2009. There is also a book

Internet Congestion Control Research Group (iccrg)

The Standard TCP congestion control is not always fit for today’s Internet. It tends to make inefficient use of high bandwidth links and with wireless connections, especially those with long round-trip times. It is also not the right choice for all applications. There is increasing deployment of nonstandardized, high-speed TCP variants such as C-TCP (included in Windows Vista), CUBIC (default in some Linux distributions), and H-TCP. The ICCRG considers new directions and techniques for congestion control in the context of the Internet. The RG is composed mostly of people involved in the IETF Transport area, people implementing operating systems, and people from the research community.

The ICCRG is organizing the existing congestion-control-related RFCs, and it created a road map that is currently waiting in the RFC Editor queue for publication. It is further identifying real, open issues with existing congestion control techniques in close cooperation with IETF WGs. Some of the open issues are heterogeneity, stability, and fairness. The ICCRG is trying to create a vision for moving beyond the current limitations of TCP-friendly concepts.

Feedback from the implementer’s perspective has been sought, and a report has been published related to the evaluation of CUBIC and H-TCP as experimental TCP congestion control mechanisms. The RG is working on a vision for enabling capacity sharing within the network infrastructure rather than simply at the end hosts.

Research topics discussed at IETF-74 included:

  • MulTFRC congestion control with tunable aggression (N-TCP-friendliness)
  • Explicit feedback on access links: using mechanisms like Explicit Control Protocol on access link bottlenecks even if end-to-end deployment isn’t possible

For more information about the Internet Research Task Force.