Methodology for traffic load estimation in WLANs based on real traces

This paper presents a methodology to estimate traffic load parameters in WLAN network consisting of a large number of access points (APs). The methodology takes into account the high variability of data traffic throughout the network that APs have to handle and process. It is based on using a low number of initial data formed from known statistical data about behavior of the network. The methodology exploits the inequality in AP popularity along the wireless network to estimate traffic load parameters. It is validated using real WLAN traces of a popular SNMP data collection of Dartmouth College. The methodology provides the traffic load estimations the coincide with the results of actual load measurements when initial input are extracted from everyday real WLAN traces. It also provides appropriate results when some of the initial data for the model are formed by averaging over a certain arbitrary long time period.