OMEGA: Online Management, Experimentation, and GAme of Large-Scale Networks
Online management of a running large-scale network poses many challenges, which have attracted significant research. As critical applications, such as high-definition TV (IPTV) and financial markets, are converging onto the Internet infrastructure, effective response to large-scale network dynamics like failures and demand spikes is gaining more importance. Link or node failures are not rare events for a large-scale network of thousands of devices. Major portion of the time for handling such network dynamics is determining how to respond, mostly performed manually in the current practice. Seeking the optimal response is often impractical, but even settling on a “good” response is very hard as well. Emergence of various networking technologies like 3G wireless and mesh networking is further complicating these management tasks. In most cases, getting the large-scale network to work is the typical target. Experienced human administrators are typically the ones who can quickly find a close-to-optimum response. However, as the networks are getting larger and more diverse, managing and attaining effective responses for an online operational network necessitates meta-tools to swiftly learn and characterize the network. This project responds to this fundamental need by developing tools to achieve automated ways of managing a running network.
The project develops tools for automated management of a running network by framing heuristic optimization, empirical learning, experimental design, and network management with a “game” interface. The project will develop an online management and experimentation system for large-scale networks in a game-like environment for trainee administrators to play with and explore what-if scenarios, without having to risk the network operation. The project will also develop algorithms for empirical characterization of network dynamics, and tools for quick and close-to-optimal configuration of numerous network parameters in response to failures or customer traffic trends. Such a framework will automate the process of configuring a large-scale network, and thus reduce the dependency of ISPs to human network operators.
The project integrates behavioral scientific concepts into the practice of operational network management. The automated management using online optimization may establish a foundation for managing multi-owner systems, e.g., power grid, transportation, and water infrastructure networks. The project’s heuristic optimization and experiment design methods as well as the game-based approach to operator training are applicable to training in safety and mission critical industries where mistakes of ill-trained administrators are intolerable, e.g., airline pilot and nuclear reactor administrator training.
· Prasun K. Dey (University of Central Florida), prasun@Knights.ucf.edu
· Mustafa Solmaz (University of Nevada - Reno), email@example.com
· Ahmet Soran – graduated with a Ph.D. in 2017
· Engin Arslan – now faculty at UNR
This project is supported by National Science Foundation award 1321069.
Last updated on October 2, 2017