Establishing Network Reputation via Mechanism Design

Abstract

In any system of networks, such as the Internet, a network must take some measure of security into account when deciding whether to allow incoming traffic, and how to configure various filters when making routing decisions. Existing methods tend to rely on the quality of specific hosts in making such decisions, resulting in mostly reactive security policies. In this study we investigate the notion of reputation of a network, and focus on constructing mechanisms that incentivizes the participation of networks to provide information about themselves as well as others. Such information is collected by a centralized reputation agent, who then computes a reputation index for each network. We use a simple mechanism to demonstrate that not only a network has the incentive to provide information about itself (even though it is in general not true), but also that this information can help decrease the estimation error.

The work is partially supported by the NSF under grant CIF-0910765 and CNS-121768, and the U.S. Department of Commerce, National Institute of Standards and Technology (NIST) Technology Innovation Program (TIP) under Cooperative Agreement Number 70NANB9H9008.