This work is supported by the National Science Foundation under NeTS Grant CNS-1018464.
Maintaining the secrecy of information has become a major research topic in modern communication networks, and the propagation properties of the wireless medium make such secrecy particularly challenging in wireless networks. There are two basic approaches to secrecy: (1) cryptographic secrecy, where the eavesdropper is assumed to obtain the transmitted signal without distortion but lacks the key required to efficiently process that signal to extract the message, and (2) information-theoretic secrecy, where an advantage on the wireless channel is exploited to transmit a signal from which the adversary can never extract the information. Information-theoretic secrecy, which is provably everlasting regardless of the adversary's capabilities, is obviously desirable, but it can be difficult to guarantee the required advantage on the wireless channel.
In this project, we look broadly at secrecy in wireless networks, including not only cryptographic and information- theoretic approaches but also hybrids of the two, to consider the initialization and operation of secure wireless networks. Whereas the provisioning of secrecy in single-hop wireless networks has been extensively studied in the past decade in both the cryptographic and information-theoretic community, the consideration of secrecy in multi-hop wireless networks is in its infancy. The goal of this project is provide fundamental solutions to the challenges presented by secure multi-hop routing in wireless networks.
The main focus of this project is to consider secrecy scaling in large wireless networks. The scaling of throughput in large wireless network has been studied in earnest since the Gupta/Kumar work of 2000. Now, attention is turning to what secure throughput can be maintained in such networks, with the metric being the number of eavesdroppers that can be present in the network while information transmitted at some rate is secure from those eavesdroppers. Work prior to ours has generally considered the case where eavesdropper locations are known, which is not only problematic from an implementation point of view, but also dodges the important problem of trying to obtain the required advantage on the wireless channel to obtain information-theoretic security when an eavesdropper may be close to a source.
As a building block for large multi-hop networks, our consideration began with security in large two-hop networks without knowledge of eavesdropper location information. In particular, we considered how cooperative jamming by friendly nodes can impact the security of the network and compared it to a straightforward approach based on multi-user diversity. We then embedded cooperative jamming techniques for protecting single links into a large multi-hop network and considered network scaling results.
Major papers (if a conference version has been superseded by a journal version, only the journal version is noted here):
General Scaling During the course of the project, other network scaling results have arisen:
Securing a Single Link A key goal of the project was to consider the security problem broadly, incorporating cryptographic and information-theoretic techniques. One thrust has been to use an information-theoretic approach to bootstrap cryptographic security:
C. Capar, D. Goeckel, K. Paterson, E. Quaglia, D. Towsley, and M. Zafer, "Signal-Flow-Based Analysis of Wireless Security Protocols", submitted to the Information and Computation: Special Issue on Information Security as a Resource .
And another interesting problem, arising mainly during the second year of the project, is not only secure a communication but also to hide its existence:
B. Bash, D. Goeckel, and D. Towsley, "Square Root Law for Communication with Low Probability of Detection on AWGN Channels", IEEE International Symposium on Information Theory (ISIT), 2012.
Neighbor Discovery An original goal of the project was to consider network initialization as well as security. However, at the suggestion of peer reviewers, this has been de-emphasized to allow more of a focus on security.
S. Vasudevan, M. Adler, D. Goeckel, D. Towsley, "Efficient Algorithms for Neighbor Discovery in Wireless Networks", IEEE/ACM Transactions on Networking , to appear.