A Kalman filter-based prediction system for better network context-awareness

This article investigates the use of Kalman filters at strategic network locations to allow predictions of future network congestion. The premise is that intelligent agents can use such predictions to form context-aware, cognitive processes for managing communication in mobile networks. Network management is improved through the use of context-awareness, which is provided through rough long or mid-term plans of operation and short-term predictions of network state and congestion levels. Research into incorporating an intelligent awareness of the network state enables a middleware platform to better react to current conditions. Simulations illustrate the advantages of this techniques when compared to traditional mobile network protocols, where the general assumption is that nothing is known about the mobility or communication patterns of the mobile entities and the network is often treated as an opaque black box. Our approach shows promise for improved network management.