Real-Time Testing of Automatic Overload Control Systems in a Laboratory Environment

Abstract

In order to investigate overload control strategies for large electronic switching systems (ESS), it is extremely valuable to be able to load test actual switching machines in a controlled laboratory environment. This method is superior when compared to both testing actual machines in service (where service deterioration cannot be tolerated) and simulating test of virtual machines (where sufficient realism may not be attainable). In the past, this form of experimentation has not been feasible for large ESS's because of both technical and economic problems associated with existing methods of generating the high volume of traffic required for overload testing. With the advent of the real-time traffic simulator known as the programmed electronic traffic simulator (PETS), however, it has become practical to conduct real-time tests of on-line overload control systems. PETS has been utilized to perform overload control experiments on an ESS to demonstrate the feasibility of conducting controlledenvironment overload testing on an actual ESS. It is the intent of this paper to describe the methodology used in the simulation experiments conducted and summarize some of the simulation results. In particular, the results of the three overload control systems will be compared with respect to both the total number of calls completed within a specified period of time and the grade of service provided. The sensitivity of system performance to the peakedness factor of the offered traffic will also be presented.