Modeling and Generation of Realistic Network Activity

The growing quantity of wireless network activity generated every second of every day creates challenges for network operators, such as detecting anomalies and providing sufficient capacity. This same network activity also creates opportunities for Smart and Connected Systems (SCSs) to adapt to changing population dynamics, detect and proactively adapt to unexpected events such as public safety threats, traffic jams, or adverse weather events, for example. The GHOST project is researching the challenges of modeling, analyzing, and generating patterns of network activity. The GHOST project has demonstrated that Nonnegative Matrix Factorization (NMF) provides a robust mechanism for modeling network activity patterns that can be used to generate realistic network activity. The GHOST team has further demonstrated the capability for injecting programmed activity patterns into a live, functioning wireless network.