High Bandwidth Real-Time Network Traffic Generation with Self-Similarity

Network traffic generation has always been a hot spot in network measurement and simulation research. However, traditional methods of generating network traffic are either insufficient in fulfilling bandwidth requirement or matching the characteristics of a real network. In this paper we propose a platform of generating high bandwidth real-time network traffic with self-similarity characteristic. The high bandwidth real-time generating feature guarantees the feasibility and self-similarity characteristic makes the platform more reliable. In the platform, we choose Cavium OCTEON CN3860 as the core and implemented ON/OFF aggregation model in a discrete-event way. The Hurst parameter of synthetic traffic is estimated both by R/S statistics method and wavelet estimator. Although there are differences between two estimators in various kinds of conditions, the validation of the platform is empirically discussed and proved.