Busy/Idle Duration Models of Video and Audio WLAN Traffics and Their Prediction Performance

Recently, efficient spectrum usage has become a critical issue, especially for some extreme wireless environments with huge access from massive devices and peoples such as hospitals, railway stations and airports. Cognitive radio (CR) is expected to solve such issue by predicting of channel status from the current statistics information of spectrum usage. This paper will investigate the distribution model of continuous busy/idle duration of two major and widely used wireless services: video service; and audio service, and will show their prediction performances using a simple auto-regressive (AR) based predictor. The results shows that both busy and idle duration data can be fitted using some simple distribution functions. In addition, the AR predictor can provide efficient prediction results especially for idle duration prediction.