You are here

Online Thread Detection in Social Networks Using Accelerators

Award Information
Agency: Department of Energy
Branch: N/A
Contract: DE-SC0013181
Agency Tracking Number: 215608
Amount: $149,500.00
Phase: Phase I
Program: SBIR
Solicitation Topic Code: 03a
Solicitation Number: DE-FOA-0001164
Solicitation Year: 2015
Award Year: 2015
Award Start Date (Proposal Award Date): 2015-02-17
Award End Date (Contract End Date): 2015-11-16
Small Business Information
3405 Piedmont Road NE Suite 100
Atlanta, GA 30305-1741
United States
DUNS: 827568226
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Oded Green
 (800) 570-1941
Business Contact
 Oded Green
Title: Dr.
Phone: (800) 570-1941
Research Institution

Problem Statement: Real-time analysis and monitoring of networks is crucial for threat detection. Networks are ubiquitous in todays world and can appear in many formats and to name a few we have: communication networks, social networks, and power networks. In many instances it is important to know who the key players are in the network these can be people, computers, or relay stations. One of the most widely used metrics for finding such players is between-centrality.

Solution Statement: Betweenness centrality is computationally demanding. In recent times several approaches have been created that allow for tracking key-players in the network as it evolves (in real-time). These approaches have limited parallel scalability which limit the capability to analyze large(r) networks. Our work will focus on taking these new approaches and improving on the scalability by applying new load-balancing techniques to that analytic that will allow for faster and better real-time monitoring of the system.
Phase I Goals: In Phase I we will apply several workload estimation schemas that will try to estimate at real time the expected amount of work that is required for updating the analytic due to the network-increment. In practice, we will need to estimated work for the two main stages of the algorithm. Once the work has been estimated, we will need to partition the work in a way that each computational resource (aka thread, core, or processor) will get an equal amount of work. Thus, our work is split into main phases: 1) estimate the work and 2) load-balance the work. Achieving these two goals will get us significantly closer to real-time monitoring of these networks.
Commercial Applications and Other Benefits: There are multiple industries that would benefit from having fast real-time monitoring of networks. This includes communication companies that want to find key routers in their network to avoid a single point of failure, for recommendation systems to find key players that would help market their product, or for security applications it might direct you to the leader of an organization. Congressional Summary: This project would improve real-time capabilities for tracking key players in dynamically changing networks. The newly proposed research would allow faster tracking of important players in larger networks.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government