Equilibrium joining strategies in the retrial queue with two classes of customers and delayed vacations (Q6170562)

The paper under review studies an \(M/M/1\) retrial queueing system with non-preemptive priority and delayed vacations. There are two classes of customers, denoted as \(P_1\) and \(P_2\). Assume that \(P_1\) and \(P_2\) customers arrive in the system according to two independent Poisson processes with rates \(\Lambda_1\) and \(\Lambda_2\), respectively. Upon the arrival of a \(P_2\) customer, if the server is available (idle state), the arriving \(P_2\) customer can obtain the service immediately. If the server is unavailable (busy or vacation state), the arriving \(P_2\) customer can join a regular queue with infinite capacity. Upon the arrival of a \(P_1\) customer, if the server is available, the \(P_1\) customer immediately occupies the server and its service begins. If the server is unavailable, the arriving \(P_1\) customer can join an infinite-capacity retrial orbit and retry later. The inter-retrial times of \(P_1\) customers are assumed to follow an exponential distribution with parameter \(\theta\). Additionally, \(P_2\) customers have non-preemptive priority over \(P_1\) customers, which means that if a \(P_2\) customer arrives during the service time of a \(P_1\) customer, then the \(P_2\) customer needs to wait in the line until the server completes the service of the \(P_1\) customer. The service times of \(P_1\) customers (or \(P_2\) customers) are exponentially distributed with parameter \(\mu_1\) (respectively \(\mu_2\)). The service discipline is First-Come-First-Served among each type of customer. If there is no \(P_2\) customer in the system after a service completion, the server will keep an idle period of random length \(K\), which is assumed to be exponentially distributed with parameter \(\alpha\). During the idle period, arriving customers (newly arriving \(P_1\) or \(P_2\) customers or retried successful \(P_1\) customers) are immediately served. If no customers arrive, the server will start a vacation. The vacation times of the server follow an exponential distribution with parameter \(\beta\). Besides, the authors assume that the interarrival times of \(P_i\) customers, service times of \(P_i\) customers (\(i = 1, 2\)), inter-retrial times, the reserved idle times, and vacation times are mutually independent. By constructing a three-dimensional Markov chain, the authors obtain the stability condition of the system and some main performance measures. Then, in dependence of a linear reward-cost structure, they derive customers' two-dimensional equilibrium joining strategies in the fully unobservable case. Due to the complexity of the social welfare function, the authors explore the socially optimal joining strategies through the Particle Swarm Optimization (PSO) algorithm. Finally, they illustrate the impact of system parameters on the two types of joining strategies via numerical experiments.