Accuracy evaluation and inaccuracy screening in PC-based delay variation measurement

Quality of Service (QoS) measurement in large-scale QoS networks is required to be both cost effective and well performed. Common PC-based QoS measurement often suffers from measurement accuracy problems, especially for delay distribution edge (99.9 percentile) measurement. We propose a generic measurement accuracy evaluation method applicable for any delay measurement equipment. Applying this evaluation method with a large amount of evaluation data also sheds light on characteristics of the measurement inaccuracy occurrence pattern. Finally, we propose the “measurement accuracy indicator” (MAI) method for improving measurement accuracy with the post-process screening approach. Using this MAI method, measurement inaccuracy was sufficiently reduced.