Low probability of detection (LPD) communication is the transmission of a message in the presence of a watchful adversary without that adversary even being aware that communication has taken place.  Most research in security considers the protection of message content from an adversary through cryptographic approaches.  However, there are numerous applications where even the adversary's detection of the presence of a message, particularly if that message is encrypted, is problematic for the sender.  For example: (1) encrypted communication can indicate the presence and/or scale of military operations;  (2) an authoritarian government could shut down encrypted communication if it is detected between unauthorized parties during a time of social unrest; (3) a government might track user activities based on communications they originate.  Hence, the topic of LPD communications, also now more commonly referred to as "covert communications," is of emerging interest.

Our work shortly before grant inception set down the square root law (SRL) for the fundamental limits of covert radio communications.  The SRL states that, in n time slots, sqrt(n) bits can be sent reliably and covertly; conversely, sending more than sqrt(n) bits results in either detection of the communication by the adversary or erroneous reception at the desired receiver.  There is a basic tension that leads to this result:  covert communication requires a low transmission power so as to hide in the background noise, and this limited power makes reliable transmission challenging.

This project has developed the nascent field of covert communications through a collaboration between the University of Massachusetts at Amherst and Raytheon-BBN Technologies.  The outcomes in Intellectual Merit have been developed in three areas:

1.  Uncertainty at the adversary:  Critical to the adversary's ability to detect the presence of a message in a noisy environment is the adversary's knowledge of: (1) the message parameters; (2) the characteristics of the background environment.  We have shown that uncertainty at the adversary about the timing of a potential message, even if the desired recipient also has such uncertainty, improves the rate of covert communication.  We have also shown that generating uncertainty at a powerful adversary about the characteristics of the background environment is challenging, but we have further shown how to cause such uncertainty via the use of electronic jamming to greatly improve the rate of covert communication.

2.  Quantum mechanical considerations at optical frequencies:  Our early work under which the SRL was developed considered models appropriate for radio communications.   However, at the frequencies employed for optical communication, which is often desirable for secure communications (e.g. for high rate or the ability to point the transmitter), the consideration of the most powerful adversary must include quantum mechanical aspects.  We have performed the extension of our work to this scenario by developing theoretical bounds and validating the results with a proof-of-concept experiment that provided the first experimental validation of the SRL.

3.  Timing channels:  The randomness in channels appropriate for Internet communications comes from uncertainty in packet timings rather than background noise.  We have considered how such randomness can hide covert communications in the presence of an adversary looking for any aberrations that might indicate the presence of covert communications.  Here, we have demonstrated another SRL:  in n overt packets, a user can covertly insert sqrt(n) packets.  In addition, if the covert user is allowed to alter packet timings, we have demonstrated a method by which the covert user can use such to greatly improve the rate of covert communication.

The Broader Impacts of the project have been significant.  First, our foundational work in this area has been widely disseminated and ignited interest in covert communications in the information theory community, with a number of other research groups now working in the area.  This led to an informal workshop on covert communications at the 2016 IEEE International Symposium on Information Theory.  In addition, we have given a number of invited talks that have either focused solely on covert communications or included results from this grant.  The project has also had a significant impact on education and research training.  Four PhD students and one undergraduate student have made significant contributions to the project; this includes two female PhD students, one of whom is from a traditionally underrepresented group.  Boulat Bash completed his PhD under the support of this grant and then joined Raytheon BBN, where he leads a follow-on project from the Defense Advanced Research Projects Administration (DARPA): Quantum-secured Imperceptible and unExploitable communication and sensing Technologies (QUIET), which looks further at covert communications in the presence of an all-powerful quantum-capable adversary.  Finally, our work under this grant received recent recognition from the National Security Agency (NSA): B. Bash, A. Gheorghe, M. Patel, J. Habif, D. Goeckel, D. Towsley, and S. Guha, "Quantum-secure Covert Communication on Bosonic Channels,'' Nature Communications, October 19, 2015, won Honorable Mention in the Fourth Annual NSA Best Scientific Cybersecurity Paper Competition.

Last Modified: 11/25/2016
Modified by: Dennis L Goeckel