Comparing Three Algorithms of Implementing ON/OFF Aggregation Model

ON/OFF aggregation model is one of the efficient and accurate models for self-similar network traffic generation. In this paper we propose and compare three algorithms of implementing ON/OFF aggregation model, based on Cavium OCTEON CN3860 network processor, aimed to achieve high-bandwidth and real-time network traffic generation. The model is implemented in a multi-thread approach, in a token-bucket approach and in a discrete-event approach. The Hurst parameter of synthetic traffic is estimated by wavelet estimator, and the validation of the last algorithm is empirically discussed and proved.