Performance analysis of switch access systems

Abstract

We compare simulation and analytical approaches for estimating the blocking of switch access systems. We include the effects of both concentration and limited availability. We allow for two types of calls. Type-1 calls queue if they are initially blocked. In contrast, if a Type-2 call is blocked then with a certain probability r it will enter retry mode and with probability 1-r it will abandon. Once in the retry mode, the call is reattempted at constant time intervals of /spl tau/ seconds. The simulation approach incorporates all the features described above, and moreover, allows for a mix of alternative call holding times. The analytical model is a finite birth-death process which captures the effect of limited availability through scaling factors that modulate the state-dependent arrival and service rates. Although the analytical model does not have the flexibility of the simulation approach, our interest was in determining if it could serve as an approximation to the complexities inherent to the simulation. We compare the simulation and analytical approaches using an example switched access system and conclude that the modeling approach matches the simulation approach very nicely when retries are not allowed. When retries are allowed, the match is not as good but is still satisfactory.