Generating artificial traffic over a local area network using random number generators

Abstract

Effective testing and performance evaluation of a local area computer network requires the ability to generate artificial traffic. This in turn requires algorithms for generating random number sequences. This article evaluates several random number generation algorithms considered for emulating traffic over NBSNET, a local area computer network at the National Bureau of Standards. The algorithms are evaluated in light of certain time, space, and programming constraints dictated by NBS's application. The table-based method, using an additive uniform random number generator for selection from the table, was determined to be a satisfactory method considering NBSNET's constraints, and is being used to generate artificial traffic for continuing local network research experimentation.