A Nonlinear Fractal Interpolation Method for High Sampling Ratio Network Traffic Reconstruction
Kun Guo, Zeshen Chen, Yingting Luo · 2025
Directly measuring large-scale network traffic incurs substantial overhead and is impractical in many cases, so network traffic is sampled in many practical situations. This paper investigates the reconstruction of origin-destination flows under fixed sampling ratios. A special form of nonlinear fractal interpolation is used to reconstruct network traffic. Based on the idea of minimizing symmetric differences, we propose a parameter estimation method for this form of nonlinear fractal interpolation to guarantee that its interpolation results are constrained within a parallelogram. Simulation results demonstrate that our method yields more robust reconstruction results than some existing methods and those with fixed vertical scaling factors.