THROWDOWN MACHINE FOR TRAFFIC STUDIES 293 



too expensive or indeed impossible. Then, too, the system may be pro- 

 posed only and not yet exist. 



One solution is to devise a method of simulating the performance of 

 the actual or proposed system which through the use a suitable time scale 

 will permit the necessary information to be obtained. The simulation 

 may be done entirely on paper by recording each state of the system and 

 modifying this state mth each bit of input information according to the 

 system plan, as though a log were being kept of the performance of the 

 actual system. 



Since the general problem involves large quantities of input data 

 which are statistical in nature, all possible variations cannot be studied. 

 A sufficient number of typical situations must be tried to obtain statisti- 

 cally reliable results. These methods have been extensively used in tele- 

 phone traffic studies and are called ^'throwdown" studies. The name 

 stems from the use of dice in the early study of telephone traffic problems. 

 Each die is designated to represent a particular independent event and 

 the faces of this die are designated according to the probability of the 

 event taking place. By repeatedly "throwing down" a number of such 

 dice and observing the results, the probability of a particular combi- 

 nation of events taking place can be estimated. Other similar methods 

 based on selections from lists of random numbers have been used in 

 telephone traffic studies for a number of years. Recently, mathematicians, 

 using digital computers, have employed similar statistical methods in 

 problems relating to the diffusion of gases, electron ballistics and the 

 solution of certain types of differential equations. They have called this 

 the "Monte Carlo" method. 



Various mechanical aids can be used in running a throwdown study. 

 This paper wdll discuss the techniques of throwdown studies and will 

 describe a semi-automatic throwdown machine which was constructed 

 for studies of the neAV No. 5 crossbar switching system used in local 

 telephone central offices. A general view of this machine is shown in 

 Fig. 1. It is a system of electrical switching circuits, signal lamps and 

 mechanical devices which simulates a large telephone switching system 

 and its associated subscribers. The machine is controlled by a team of 

 four operators. Artificially generated telephone traffic is processed by 

 this machine in a manner analogous to the action of the actual system. 

 Detailed records are made of the progress of each call and the traffic 

 situations encountered. After a sufficient number of calls have been 

 processed, the recorded information can be analyzed by statistical meth- 

 ods to obtain desired information. The action of the system is simulated 

 in sufficient detail to insure that results are representative of actual 



