NGHIÊN CỨU ẢNH HƯỞNG CỦA KÍCH THƯỚC CỬA SỔ QUAN SÁT ĐẾN ĐỘ CHÍNH XÁC DỰ BÁO TRONG MÔ HÌNH ĐIỀU KHIỂN CHẤP NHẬN LẬP LỊCH

Phạm Trung Đức và Lê Văn Hoà. Tập 129, Số 2A, 2020 106 STUDY THE EFFECT OF THE OBSERVED WINDOW SIZE TO PREDICTION ACCURACY IN SCHEDULING ADMISSION CONTROL MODEL Pham Trung Duc1,2*, Le Van Hoa3 1 Hue Coach Station Join Stock Company, Vietnam 2 Hue University of sciences, Hue University 3 School of Hospitality and Tourism, Hue University Abstract: The problem of prediction which bases on the observations depends on the density of incoming events and the size of the observation window. In fact, if the observation window is small and the density of incoming events is discrete, the accuracy of prediction will be low, but if the observation window is large and the density of incoming events is dense, the accuracy of prediction will increase significantly. In this paper, a survey is based simulation will be performed with different observation window sizes in order to evaluate the efficiency of the incoming burst traffic prediction which serve for scheduling admission control at core node OBS. The simulation’s result shows that, when the incoming burst load is high and the observation window is large, the prediction results will be more accurate. Keywords: OBS Network, QoS, scheduling admission control, prediction base on rate, the size of the window.