Secure Host Name Resolution Infrastructure for Overlay Networks

Ayumu KUBOTA  Yutaka MIYAKE  Toshiaki TANAKA  

IEICE TRANSACTIONS on Communications   Vol.E89-B   No.9   pp.2434-2439
Publication Date: 2006/09/01
Online ISSN: 1745-1345
DOI: 10.1093/ietcom/e89-b.9.2434
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Networking Technologies for Overlay Networks)
routing overlays,  domain names,  cryptographically generated identifiers,  PKI,  

Full Text: PDF(189.7KB)>>
Buy this Article

In order to introduce new routing functionality without changing the Internet infrastructure, many routing overlays have been proposed in recent years. Although such overlays allow us to dynamically and flexibly form various types of networks, the current host name resolution mechanism used in the Internet, i.e. DNS, cannot provide us such flexibility in host name referencing because of its delegation-based administration scheme of domain names. And also, it cannot provide us security because of the lack of wide deployment of its security extension, DNSSEC. In this paper, we propose a generic framework for secure and flexible host name resolution infrastructure that can be shared among many routing overlays. In contrast to DNS with which users are forced to use the domain name space managed by IANA/ICANN, our framework separates the name resolution mechanism from the name spaces it handles, which allows users to choose whatever name space they think appropriate for the identity scheme of their overlay-networking community. This realizes decentralized management of domain names and gives users freedom in domain name acquisition. The basic idea to achieve this is to use a cryptographically generated identifier (i.e. a hash of a public key) as a reference to an administrative domain of overlay networking hosts and allow the owner of the domain to securely publish host information using the corresponding private key. We show that a referencing mechanism for such host information can be easily implemented by using distributed hash tables (DHTs), and then show how such "semantic-free" references to domains can be linked to existing identity scheme in order to allow "human-friendly" referencing.