“This challenge was an opportunity to apply visual analytic design skills without having to build a functional system,” says team lead Diane Staheli, a member of the technical staff in the Cyber Systems and Operations Group. “Without the constraints involved in building a system, we could focus our energy on fully developing the concept and were able to be more creative in the design of a solution.”
The winning design featured a multilayered display that allows operators to call up a variety of information such as resource usage on the network, security breaches, or network performance issues. Icons and color-coded zones give network operations managers immediate cues to problem areas, types of problems, and problem-resolution status.
“Users consume visual information in different ways,” says Staheli. “One group may prefer seeing everything at a glance, others may prefer the ability to choose what to see, and some groups may want the ability to interact with the information. Our design has some of all three elements.”
Staheli attributes the success of the Lincoln Laboratory design to a talented team composed of staff members from multiple technical backgrounds. The team — David Danico, Raul Harnasch, Maureen Hunter, Richard Larkin, Jeremy Mineweaser, David O’Gwynn, Harry Phan, Alexia Schulz, Tamara Yu, Kevin Nam, Andrea Brennen, Michael Snyder, and of course Staheli — included analysts, systems engineers, user interface developers, visualization and information designers, a computer scientist, a physicist, and a network engineer. “A strength of our design was that we developed our requirements from the team’s deep real-world experience with network operations centers.” The project was primarily a volunteer effort with work done outside regular business hours or during lunch meetings.
The field of visual analytics — the use of graphical displays to facilitate understanding and exploitation of data — is growing as more and more analysis tasks require humans to interact with computers. Visual analytics can illuminate relationships between data, suggest avenues for targeted searches, and consolidate disparate information into a cohesive picture. The field draws on expertise from computer science and engineering, cognitive science, interactive and graphic design, and the social sciences to create visualizations that not only illustrate information but also present it in logical ways to allow users to effectively manipulate it. Visual analytics is finding use in a broad range of applications: map-intensive displays for first responders to a forest fire, depictions of airport runway activity, and, as in the VAST mini-challenge, situational awareness visualizations of computer networks.
“So much of understanding the cyber domain is being able to visualize it and make decisions about what it means,” says Jeffrey Gottschalk, associate leader of the Cyber Systems and Operations Group. “Our group has been doing some pioneering work in this area, working closely with network operators and defenders here at Lincoln Laboratory and in the DoD [Department of Defense] to help identify visual analytics that are actually useful. The challenge is not only how to have analytics do something to aid a human, but also how to employ design — the tool needs to be efficient, too. This project was a great way to hone our skills on a problem set relevant for our project work, and get some great feedback on it from other researchers and designers.”
The VAST Challenges are part of the IEEE VIS (Visualization) Conference, which comprises three IEEE conferences — Scientific Visualization, Information Visualization, and VAST. The challenges were incorporated into the IEEE VIS in 2006 and have included scenarios ranging from a bioterrorism attack to arms smuggling and a computer network disruption. Besides the situation awareness challenge, VAST’s first visual design problem, the 2013 challenges were a predictive analysis of the box office success of a new movie and the interpretation of computer network anomalies experienced by a marketing company.
Staheli presented Lincoln Laboratory’s 2013 challenge entry during the VAST Challenge Workshop at the IEEE VIS 2013 held in mid-October in Atlanta, Ga. A two-page summary of the design will be included in the VIS 2013 conference proceedings and the Visual Analytics Benchmark Repository, a resource maintained by the Scientific Evaluation Methods for Visual Analytics Science and Technology project at the Human-Computer Interaction Lab, University of Maryland.