Networks and queues: Probabilistic methods (Q1587455)

The book presents a modern approach to networks and queueing theory based on probability theory which has the origin since Erlang in 1917 in telecommunication problems as well as nowadays in information networks. The aim of the monograph is to show new probabilistic methods applied to queueing theory which studies the stationarity, equilibrium conditions, asymptotic methods using theory of martingales, Markov processes, and ergodic theory. The book consists of 12 chapters and 4 appendices. The first chapter is dedicated to the elements of point processes which describe the stream of customer arrivals. Here the Laplace transform of a point process is explained and used to Poisson processes and their construction, and renewal processes. The next two chapters are devoted to study queueing theory of GI/GI/1 types and the processes connected with them: arrival distribution of customers, using random walks. The main probabilistic tool is the Wiener-Hopf factorization which enables to describe the stationary distribution of the waiting time which is applied to GI/GI/1 queues. Then the author describes a one-channel queue where the service time is exponential or the interarrival time is exponential. In addition, the one-server queue is studied under heavy conditions. The reversibility and equilibrium conditions of networks are the leitmotiv of the fourth chapter. Here the reversibility of Markov processes is explained, and Jackson and Gordon-Newel networks are introduced. The M/M/1 queue is exhibited in the fifth chapter in more detail with accent to the convergence velocity to the equilibrium. Renormalized associated processes are introduced and large deviations are studied. The \(\text{M/M/}\infty\) queue is studied in the sixth chapter where the main accent is dedicated to Ornstein-Uhlenbeck processes. Queueing systems of M/GI/1 type with FIFO and FILO disciplines are exhibited in the next chapter using inserted Markov chains, and the problem of ergodicity is exhibited in the eighth chapter to know when the process converges to its equilibrium. More information on ergodic theory can be found in the tenth chapter. The ninth chapter introduces a renormalization technique for the study of Markov processes in more detail. The last two chapters deal with stationary point processes and queueing systems of type G/G/1 and G/G/2. The book is accompanied with four appendices where Poisson distribution, martingales, Markov processes and distribution convergence are explained. The book is very nice and it is welcome into the family of probability books on queueing systems.