135-139

• Xiangyang, Mu
• Taiyi, Zhang
• Yatong, Zhou

Network traffic prediction is important to network planning, performance evaluation andnetwork management directly. A variety of machine learning models such as artificial neuralnetworks (ANN) and support vector machine (SVM) have been applied in traffic prediction. In thispaper, a novel network traffic one-step-ahead prediction technique is proposed based on a state-ofthe-art learning model called minimax probability machine (MPM). In the experiments, the predictiveperformance is tested on two different types of traffic data, Ethernet and MPEG4, at the sametimescale. We find the predictions of MPM match the actual traffics accurately. Furthermore, wecompare the MPM-based prediction technique to the SVM-based techniques. Results show that thepredictive performance of MPM is competitive with SVM.

• Economics and Business
• Computer and information sciences

• Business management

network traffic, minimax probability, support vector machine, prediction

Περιέχει γραφήματα, πίνακες και βιβλιογραφία