Modeling Supervisory Control and Team Performance in the Air Defense Warfare Domain with Queueing Theory. Part II

摘要

In our previous research, we hypothesize that the performance of a supervisory control operator that must process tasks recommended by a system task manager is analogous to the performance of a vacationing server, M/Er/1 queue. Thus, we assume that the arrival of tasks is Markovian and that service consists of r- stages of processing each of which is exponentially distributed. In addition, we assume that when there are no tasks in the queue to process, the operator takes a vacation, i.e., goes off and performs other duties. The model assumed vacation time was exponentially distributed. We derived the queueing statistics for this system. These statistics include (1) the average number of customers (tasks) in the queue, (2) the average time a task spends in the queue, and (3) the average waiting time in the queue. We extended this model to a two-class priority M/Er/1 vacationing server system. The results of these predictions were compared to actual operator performance. Our current research generalizes the arrival processes. That is, instead of assuming that the arrival of tasks follows a Poisson process, we assume a Markov-Modulated Poisson Process (MMPP). The MMPP allows for a change in the rate in which tasks arrive to the system. Thus, rush hour effects and the ebb and flow of task arrivals may be taken into account by the new model. In the Command and Control environment, it is particularly important to estimate the rush hour effect on time critical events. A new set of queueing statistics was generated for a two- class MMPP/M/1 vacationing server system. This allowed us to compare the model to operator performance on the test scenario over extended periods of time.