ADELPHI, Md. -- An Army researcher coauthored a book that covers ground-breaking research for the enhancement of warfighter communication networks.
Cambridge University Press published the book, Network Tomography: Identifiability, Measurement Design, and Network State Inference, which is the result of collaboration between researchers at the U.S. Army Combat Capabilities Development Command, known as DEVCOM, Army Research Laboratory; IBM U.S.; and the University of Massachusetts-Amherst and Imperial College, London, under the auspices of the U.S.-U.K. Network and Information Sciences, International Technology Alliance, or NIS ITA.
“As tactical wireless networks get more complex, autonomous network control will become increasingly important,” said Dr. Ananthram Swami, the Army’s senior research scientist for network science. “This requires knowledge of the internal state of the network, such as the status and properties of links and nodes. The techniques described in this book provide a set of tools that address some of the issues in network state inference, extending beyond communication networks.”
The book covers the most prominent and up-to-date results in the field of network performance tomography.
Network tomography refers to the use of inference techniques to reconstruct detailed views of the network’s internal state from external measurements taken between a selected subset of peripheral nodes referred to as monitors. It advocates that it is possible to map the path data takes by examining information from edge nodes, the computers in which the data are originated and from which they are requested.
In contrast to the conventional approach of direct measurement, as adopted in most commercial network monitoring systems, network tomography works on end-to-end measurements taken between monitors, and thus avoids the needs to rely on monitoring agents or protocol support at internal nodes, Swami said. This is particularly useful in monitoring closed network systems, for example peer autonomous systems in the Internet, underlay networks in an overlay network, optical components in a hybrid optical/electrical network, and partner networks in a military coalition network.
Crucial to such monitoring is reducing the cost of monitoring by minimizing the number of monitors, placing them optimally, and designing the measurement process.
As a scientific discipline, network tomography is fairly new with about 20 years of history and thus few books have been published on it, Swami said. Existing books either do not cover the fundamental principles or are not comprehensive.
“This book fills this gap by providing a comprehensive overview of network tomography with detailed coverage of the state of art in its fundamental principles and algorithms,” Swami said. “The book covers all major components of network tomography, from the construction of a measurement system to the computation of the network state of interest.”
As an added bonus, the book provides hints and tips for network administrators on best practices and guidelines in applying the presented solutions where applicable.
“Our research has led to scalable solutions for optimal monitor placement, measurement designs and inference in the deterministic case,” Swami said. “For the stochastic case, we have examined issues of probing experiment design via Fisher information, and developed optimal design approaches to ensure fast convergence of estimates to the ground truth network state. Theoretical results have been validated via extensive simulations based on synthetic topologies as well as with real-world topologies.”
Moving forward, the researchers are looking at extending this work to infer node properties and other link properties, which would be useful for resource allocation for distributed analytics in tactical networks.
“In an adversarial environment, all measurements will not be trustworthy; developing approaches to characterize and cope with such vulnerabilities is crucial,” Swami said.
Another next step will be combining network tomography with topology inference.
“Combining topology and state inference, known as network topology tomography, provides means to construct a logical view of a closed network that otherwise appears as a complete black box,” said Professor Ting He, formerly at IBM U.S. and currently at Pennsylvania State University. “The logical view is consistent with the underlying network in terms of performances on the measured paths, but may have a different structure. How to bring them closer remains a challenge in many scenarios.”
Another step is the extension of network tomography developed for “classical” networks to “quantum” networks.
“The next ten years will see a rapid development and growth of quantum networks,” said Professor Don Towsley, University of Massachusetts, Amherst. “Management and control of these networks will benefit tremendously from the development and deployment of network tomography. However, adapting current techniques to the quantum environment will pose an exciting challenge.”
While the focus of the book is on communication networks, the techniques therein should be applicable to broader classes of networks, Swami said.
As the Army’s national research laboratory, ARL is operationalizing science to achieve transformational overmatch. Through collaboration across the command’s core technical competencies, DEVCOM leads in the discovery, development and delivery of the technology-based capabilities required to make Soldiers more successful at winning the nation’s wars and come home safely. DEVCOM Army Research Laboratory is an element of the U.S. Army Combat Capabilities Development Command. DEVCOM is a major subordinate command of the Army Futures Command.