Coding for Resilient Communication
Abstract
From a technological standpoint, battlefield communications is distinctive in a number of ways. Arguably the most important, and least understood, of these is the presence of an adversary with the capacity and the motivation to actively disrupt the communication. This form of disruption is fundamentally different from the noise that naturally occurs in communication systems. On the one hand, noise follows statistical norms such as ergodicity, while the adversary is free inject signals that are completely unpredictable and maximally damaging to communication. On the other hand, while noise is omnipresent, it is reasonable to assume that an adversary is localized. Thus the standard design methodologies for combating noise are not applicable. Recent work of the PI has shown that it is possible to design coding schemes that leverage the localized nature of the adversary to minimize its impact. These coding schemes are quite different from those employed to combat noise, demonstrating that pursuing such schemes can lead to fundamental scientific advances. This recent work is theoretical in nature, however, and there is a substantial divide between the coding schemes that it provides and practical, battlefield communication. This project will bridge this divide by extending the prior work in various ways. During the first 12-month period, we shall consider a richer class of sources that can model practical multimedia sources such a audio and video. During the second 12-month period, we shall consider other weaker, but more realistic, adversarial models, including adversaries that are non-omnisicient in certain ways and adversaries that can only scramble packets. During the final 12-month period, we shall consider bi-directional and interactive communication models. In all cases, we will develop both code designs and fundamental, information-theoretic limits. In parallel, during the second and final 12- month periods, we shall develop a socket-level testbed that demonstrates, in the context of video streaming, the coding schemes developed over the course of the project. The project overall, and the testbed development in particular, will be conducted in collaboration with engineers at MITRE, with whom the PI has been actively working on this topic.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 06, 2019
- Source ID
- W911NF1810426
Entities
People
- Aaron B. Wagner
Organizations
- Army Contracting Command
- Cornell University
- United States Army