A \(c\)-server Poisson queue with customer impatience due to a slow-phase service (Q2073500)

Summary: In this paper, we investigate a queue with \(c\)-servers in a Markovian environment (ME) under two phases \((S\)-slow and \(F\)-fast). We assume that the sojourn times follow an exponential distribution at states \(S\) and \(F\) with parameters \(\nu\) and \(\eta\) respectively. Each customer chooses a random relative deadline duration (RDD), which follows an exponential law with parameter \(\alpha\) under \(S\)-phase but abandons the system as soon as its RDD expires and never returns. When the environment remains under phase \(j(=S,F)\), customers arrive according to a Poisson process with rate \(\lambda_j\), and are served according to an exponential distribution with rate \(\mu_j\) where \(\mu_S<\mu_F\). We formulate the queue length process \(Y_{(j,n)}(t)\), representing the number of customers \(n(t)\), waiting in the phase \(j\) at time \(t\) as level dependent quasi birth-death (LDQBD) process in a 2-dimensional space. We apply matrix analytic methods on computational procedures and iterative methods to obtain scalar types of explicit expressions for the stationary probability distributions of \(Y_{(j,n)}(t)\) as \(t\) tends to infinity. We present comparable charts to highlight the variations between the steady state measures of an \(M(\lambda_0)/M(\mu_0+\alpha)/c\) queue with customer impatience (CI) and of another \(M(\lambda)/M(\mu)/c\) facility without CI.