Impact of Traffic Correlation on the Effectiveness of Multilayer Traffic Engineering

Multi-layer network (MLNW) consisting of packet and optical layer networks has been widely studied as an important next generation network. In the MLNW, the virtual network topology (VNT) formed by a set of optical paths can be dynamically reconfigured by GMPLS protocols. By reconfiguring VNT, called multilayer traffic engineering, network performance factors can be optimized based on the traffic matrices. Thus, the effectiveness of multilayer traffic engineering is considered to be highly dependant on the correlation among traffic demand of different source-destination flows. We propose an analytical model of the dynamics exhibited by traffic matrices and then investigate how the correlation between different pairs of source-destination flows impacts the performance possible with multilayer traffic engineering. We confirm that the effectiveness of periodical multilayer traffic engineering increases with the correlation among source-destination flows