ABSTRACT

Over the next few years, the basic architecture of the next generation of Internet will be decided, hopefully with a much greater emphasis on security. In a network, security and communication are interdependent. Cryptographic tools will be required to ensure that communication is dependable despite hackers' denial-of-service attacks and other sabotage. On the other hand, proper communications architecture could provide primitives that make powerful cryptographic tools for security, such as secure multiparty computation, e_cient enough to be implemented. Unfortunately, research on reliable communications and computation in a network has traditionally been studied without consideration of security, and vice versa. This research examines a model of communication channels that includes security considerations in a robust way. In this model, protocols can be simultaneously evaluated for the two dual objectives of secrecy and reliability. This model unites previous approaches to these questions from the points of view of cryptography, distributed systems, and error-correction. The model also unites information-theoretic techniques (with security based on the attacker's inability to access certain information) with complexitytheoretic approaches (based on the attacker's inability to solve intractable computational problems). Possible constructions of channels with stronger properties (increased privacy, authentication, or reliability) from those with weaker properties will be explored. For example, is it possible to take an arbitrary channel that gives only slightly more information to the intended receiver than to an attacker and use it to build a highly reliable and almost completely secret channel? Can we use a channel for secret, reliable communication to create a channel emulating a virtual broadcast? What is the relationship between secrecy and anonymity?

Please report errors in award information by writing to: awardsearch@nsf.gov.