Deriving link load with edge-based measurement in high-speed IP network

One vital aspect of traffic engineering is traffic measurement that is beneficial for monitoring network state and controlling network traffic. Obtaining a volume of load on each internal link of the network is crucial for network management and control. Though collecting can be available on each link, such as applying traditional SNMP scheme, the approach would be expensive because it may cause heavy processing load and sharply degrade the throughput of the core routers in high-speed IP backbone. In addition, collection of the huge volume of measurement data will also cause great overhead within backbone network. Then monitoring merely at the edge of the network and estimating traffic in the core provides a good alternative way for overcoming such functionality limitations. In this paper, we explore a scheme to estimate the volume of traffic on internal links of network with edge-based measurements. In contrast to the expensive method of collecting routing data from core routers, we propose a route probing method based on hash sampling technique and IP Measurement Protocol between node-pairs. Based on statistics theory, we prove that our approach is effective and present the algorithm. We also show performance simulation results that indicate potentials of our approach.