Maximizing Profit of Network InP by Cross-Priority Traffic Engineering
Yi Xu, Xu Du · 2018
Traffic engineering (TE) plays an important role in determining the network performance and reliability, which has been studied thoroughly in past decades. However, due to the advances in cloud and mobile computing, the confliction of bandwidth provision and consumption in network becomes more and more complicated. A major challenge of the Infrastructure Provider (InP) is how to guarantee the high-priority user demands, select valuable and appropriate low-priority user demands, and keep maximum profit simultaneously. This kind of optimal problems related with cross- priority TE was studied in this paper. We first analyzed the relationship between network resources, multi-priority user demand and provider's revenue. And then constructed an event- driven SDN-like control system that could support multiple priority TE. Based on the parameters of network topology, SLA and system measurement, an optimal model for InP's profit was given. By building a genetic-based algorithm called global resource allocating (GRA) algorithm, we exploited the near optimal profit that satisfied partial low-priority demands after satisfying all high- priority demands. Moreover, a dual GRA (DGRA) algorithm which can find out optimal mixture deployment patterns was also studied. With numerical results, we show that both GRA and DGRA could always obtain better result compared to a two-step greedy.