Catálogo de publicaciones - libros

We consider the problems of explaining and forecasting the penetration and the traffic in cellular mobile networks. To this end, we create two regression models, one to predict the penetration from service charges and network effects and another one to predict the traffic from service charges and diffusion and adoption effects. The results of the models can also be combined to compute the likely evolutions of essential characteristics such as Minutes of Use (MoU), Average Revenue per User (ARPU) and total revenue. Applying the models to 26 markets throughout the world we show that they perform very well. Noting the significant qualitative differences between these markets, we conclude that the model has some universality in that the results are comparable for all of them.

This paper considers the packet recovery performance of forward error correction (FEC) for a single-server queueing system with a finite buffer fed by two independent input processes; one is a general renewal input process in which the interarrival times of packets are independent and identically distributed according to a general distribution, and the other is a Poisson arrival process. We analyze the packet- and block-level loss probabilities, and investigate the recovery performance of FEC at block level. Numerical examples show that the block-loss probability is significantly improved by FEC when the system can only accommodate a small number of packets. It is also shown that FEC recovery is more effective than packet buffering when the traffic intensity is large.

A novel multi-server queueing model with finite buffer and batch arrival of customers is considered. In contrast to the standard batch arrival when a whole batch arrives into the system at one epoch, we assume that the customers of a batch arrive one by one in exponentially distributed times. Service time is exponentially distributed. Flow of batches is the stationary Poisson arrival process. Batch size distribution is geometric. The number of batches, which can be admitted into the system simultaneously, is subject of control. The problem of maximizing the throughput of the system under the fixed value of the admissible probability of losing the arbitrary customer from admitted batch is considered. Analysis of the joint distribution of the number of batches and customers in the system and sojourn time distribution is implemented by means of the matrix technique and method of catastrophes.

In this paper, we present an exact performance analysis of a tandem network with arbitrary number of multiplexers. Each multiplexer is fed by the output of the preceding multiplexer as well as the traffic generated by a number of independent binary Markovian sources. We model the network as a discrete-time queueing system. After determining the unknown boundary function, the probability generating function (PGF) of the distribution of queue length and number of sources for each multiplexer is derived; from these results expressions for the mean and variance of queue length and packet delay have been determined.

Some systems discard messages that wait in the queue longer than a certain threshold. Identifying such messages and discarding them takes the processor a fixed amount of time. That time is considerably smaller than an average service time, but greater than zero, therefore the models for queuing systems with reneging do not apply. This paper provides a model for such queues. It describes a method of deriving exact formulas for the distribution of waiting times, and proposes simple, yet accurate approximations. The paper then analyzes the model and discusses the properties of such systems.

A multirate loss system with complete sharing is investigated, in which multiple classes of customers arrive as a state dependent Poisson processes. This arrival process includes the Bernoulli-Poisson-Pascal (BPP) and the batched Poisson process with geometric distributed batch sizes. Asymptotic uniform approximations to the blocking probabilities are derived, when the capacity and a parameter of the arrival processes are commensurately large. The results are obtained with the saddle-point method of integration and the approximation uniformly holds across all traffic regimes, where the blocking probabilities may vary by several order of magnitude. Moreover, a numerically stable representation of the approximation is given, which gives accurate results also for the critical traffic region. Numerical results show that while prior asymptotic approximations are quite accurate, the new approximations are very accurate.

We investigate the effect that power control and rate adaptation could have on the performance of multi-hop access networks. Therefore, we formulate an optimization problem to find optimum routes in a multi-hop access network. These optimum routes are found using global knowledge and thus have only limited practical relevance. However, they can well be used to calculate theoretical upper limits on the performance, which no routing algorithm could exceed. We calculate the performance limits with and without power control and rate adaptation and compare them with each other to judge whether these mechanisms are worth the effort to be deployed in wireless multi-hop access networks.

We consider a network with a fixed number of links whose transmitters are saturated and access a channel using a random back-off algorithm. Some of the links may be hidden in the sense that they do not interfere with all other links but rather with a subset of the links. Using mean field techniques, we analyze a variety of random back-off algorithms by explicit calculating the throughput of the links in such networks. We apply our results to analyze the performance of the exponential back-off algorithm in networks with partial interaction. The results are striking and confirm experimental results. Hidden transmitters that fail to sense collisions with other links unfairly grab too much bandwidth at the expense of transmitters that comply with the back-off rules. We believe the model can be used to develop new algorithms realizing an adequate trade-off between fairness and efficiency.

To support battery powered mobile broadband wireless access devices efficiently, IEEE 802.16e defines a sleep mode operation for conserving the power of mobile terminals. In this paper we propose a theoretical Phase-type (PH) based Markov chain model to analyze the performance of IEEE 802.16e sleep mode operation. The model describes the behavior of the mobile stations working in sleep mode. In particular, the service process is designed as a discrete PH model. By means of the mathematic tools of PH theory we then derive the closed-form expressions of the mean sojourn time of packets and power consumption. Comparison with simulation results shows that the model provides an accurate prediction of the system performance. Furthermore, we propose a simple utility function to quantify the efficiency of sleep mode operation which takes the joint effect of sojourn time and power saving into account. This function allows mobile stations to decide when to enable sleep mode operation for power saving.

We develop a queueing model characterizing explicitly the impact of interference on end-to-end performance measures such as [4] throughput in ad hoc networks, emphasizing the performance trade-off between single-path and multi-path routing. It may seem attractive to employ multi-path routing, but as all nodes share a single channel, efficiency may drop due to increased interference levels thus yielding single-path performance for some topologies. We formulate a nonlinear programming problem to optimize network performance. Next, we focus on network capacity and show that for this objective the optimum could be found by solving an exponential number of linear programmes. We propose a greedy algorithm that efficiently searches these programmes to approximate the optimal solution. Numerical results for small topologies provide structural insight in optimal path selection and demonstrate the excellent performance of the proposed algorithm. Besides, larger networks and more advanced scenarios with multiple source-destination pairs and different radio ranges are analyzed.