Queuing networks with negative customers are wisely used as analytical models of infotelecommunication systems [1]; the effect is that a negative customer removes a waiting customer. Chaplygin considers a multiserver queuing system with unlimited buffer and Markovian service times so that an arriving customer removes a group of waiting customers at the queue head [2].
In this paper, an arriving customer kills one waiting customer at the end of the queue, and uses an efficient matrix algorithm to calculate the stationary characteristics. The underlying Markov process can be used, with some loss of generality, from the recurrent flow to the Markov flow. This reduces the bulky calculations of Chaplygin’s embedded Markov chain approach. Phase-type services and independent flows of positive and negative customers are presented as special cases. The results are useful for programming and numerical calculations.