Support vector regression for link load prediction

From weather to networks, forecasting techniques constitute an interesting challenge: rather than giving a faithful description of the current reality, as a looking glass would do, researchers seek crystal-ball models to speculate on the future. This work is the first to explore the use of support vector machines (SVM) for the purpose of link load forecast. SVMs work well in many learning situations, because they generalize to unseen data, and are amenable to continuous and adaptive on-line learning, an extremely desirable property in network environments. Motivated by the encouraging results recently gathered by means of SVM on other networking applications, our aim is to enlighten whether SVM is also successful for the prediction of network links load at short time scales. We consider the problem of link load forecast based only on its past measurements, which is referred to as "embedded process" regression in the SVM lingo, and adopt a hands-on approach to evaluate SVM performance. Our finding is that while SVM robustness is more than satisfactory, accuracy results are just close to be tempting, but not enough to convince. Based on the result of our experimental campaign, we then speculate on what directions can be undertaken to ameliorate the performance of SVM in this context.