Tail asymptotics for M/G/1 type queueing processes with subexponential increments (Q1975030)

A bivariate regenerative Markov modulated process \(\{I_n, L_n\}\) is described, where \(\{I_n\}\) is the phase process, and \(\{L_n\}\) is the level process with a subexponential distribution of increments. The structure of the transition matrix of the process is considered in detail. The main assumption is that elements of the transition matrix have the same subexponential tail (up to a constant) when the increment of the level process increases. The asymptotic tails of the cycle maximum \(M\) (during a regeneration cycle) and stationary level process \(L\) are derived. It is shown that the tail of \(L\) is effectively heavier than that of \(M\). The theory of renewal and (semi)-regenerative processes and some properties of random walks are widely applied.