Source Line Load Equalization Methods for the Source Aggregation Algorithm (SAA)

The source aggregation algorithm (SAA) is a well known method for reproduction of high quality self-similar data traces, often matching some predefined packet size distribution (PSD). However, lack of target link load boundaries and estimation errors in excess of 100% for short traces make it a method of choice only for 10 packets or longer traces. A small sources population, preferable for rapid link load convergence, results in quality deterioration of the generated trace. In this paper we aim to present the first, to the best of our knowledge, link load control mechanism for the SAA, based on modification of Pareto distribution parameters, allowing for reduction in line load estimation error to 3% for a sample population in excess of 10 packets with 100 individual data sources. In total, 4 individual methods are examined and their advantages and shortcomings are identified.