Real Time Network Load Estimation in Packet Switched Networks

ABSTRACT In this paper we consider the problem of real-time traffic load estimation in a packet-switched network environment. In emerging highspeed packet-switched networks, many control functions such as bandwidth management, dynamic route selection and rate-based traffic control depend on real-time characterization of traffic load in the network. In order to perform these functions properly, the traffic parameters have to be continuously monitored and an accurate estimate of traffic parameters must be obtained. Since the traffic parameters are generally time varying, the filtering mechanism must distinguish between a real change in traffic level and a temporary fluctuation of the measured data. Also, the estimator must have the capability to react quickly and trace fast changes in traffic level. To achieve this objective, we propose a time-varying adaptive bi-modal filtering technique which on one hand is very responsive to fast variations of traffic level, and on the other hand, less sensitive to measurement fluctuations. This objective is also desirable in order to limit the number of unnecessary updates and reports for various network control functions.