In this paper we derive an expression for the asymptotics of the buffer length distribution of a discrete-time infinite capacity single server queue with of long range dependent discrete-time WG/W processes. This class of arrivd process is defined as follows. At each time slot sequences of back-to-back customers are generated according to a Poisson distribution with parameter A. The length of such a sequence is assumed to asymptotically behave like a Pareto distribution with parameter S, i.e. the probability that a se. quence consists of k customers is given by c / P for k + 00, with c > 0 and < S < 3. Due to the Of distributions, the presented class of WG/m processes has the long range dependence property (i.e. the autocorrelation function decays as a power of the lag time). We show that in this case the asymptotic behavior of the tail probabilities of the stationary distribution of the buffer O C C U P~~C Y is given by ~s . -2~~~~)~l -p ) n2-' for n -+ 00, with P representing the load of the system. This result is obtained using a generating function approach and the Tauberian theorem for power series. Furthermore, an application towards traflic management and simulation results are presented. arrival process, long rsnge dependence, buffer ssymptotics Clearly this particular autocorrelation structure has a major impact on the performance of a queue to which this type of traffic is offered. Measurements, simulations and analytical studies have shown that the Property of exponential decay rate Of the tail Of the queue length distribution observed for Markovian traffic, is not valid any longer when dealing with LRD input traffic. this case, the queue length distribution may have a heavy tail which, when ignored, may lead to important underprovisioning of the required buffer size. Hence, in order to quantitatively investigate the influence Of the LRD property On the queueing we need a traffic model which has the LRD property and we have to investigate the tail of the buffer of a single Server deterministic queue whose input consists of this traffic. This is exactly the goal of this study.In this paper, we consider the discrete-time M/G/w input process. Intuitively, this process consists of sequences of back-toback customers. These sequences are generated at each time slot according to a Poisson distribution. The distribution of the num-I. INTRODUCTION ber of customers in a train has a heavy tail, in order to ensure the LRD property. Back-to-back customers mean that the customers In the past five years, traffic measurement studies on real packet ofa sequence mive in consecutive time mis M / G /~ protime and its input process belonBng to a Of the members Of this networks, Ethernet LAN traffic (see [11)7 cess naturally arises when considering an infinite superposition able Bit Rate video Over the Asynchronous Transfer Mode (see ofidentical on/offsources [3], [4], [5]. Furthermore, the M/(-Jw e.g. i21), etc., have shown that the autocorrelations Of process studied in this paper is in several consistent with the numbe...