Does allowing jockeying as described in the multiteller bank of Sec. 2.6 affect customers’ average delay in queue(s)? To find out, regard the original model (with jockeying) as “configuration 1” and define “configuration 2” to be this same model but without any jockeying allowed; assume five tellers in each case. Use dedicated streams to facilitate CRN—stream 1 for interarrival times and stream 2 for service times. However, it is not exactly obvious how we should generate the service times from stream 2; there are (at least) the following two possibilities:
(a) A customer’s service time is generated when he arrives and is stored with him as an attribute. This corresponds physically to the idea of forcing the “same” customers (in terms of their service demands on the tellers) to arrive to both configurations at the same times.
(b) Generate a service time from stream 2 only when a customer begins service. In this case, the two configurations will not see the same customers, but the ordered sequence of service times begun for both configurations will be identical.
Carry out a simulation experiment to investigate whether (a) or (b) is a better way to implement CRN, in terms of the variance of the estimator of the difference between the expected average delay in queue(s) in the two configurations.
We need at least 10 more requests to produce the solution.
0 / 10 have requested this problem solution
The more requests, the faster the answer.