Catálogo de publicaciones - libros

The talk will consist of two parts. In the first part of the talk, we will examine how transit and customer prices and quality of service are set in a network consisting of multiple ISPs. Some ISPs may face an identical set of circumstances in terms of potential customer pool and running costs. We will examine the existence of equilibrium strategies in this situation and show how positive profit can be achieved using threat strategies with multiple qualities of service. It will be shown that if the number of ISPs competing for the same customers is large then it can lead to price wars. ISPs that are not co-located may not directly compete for users, but may be nevertheless involved in a non-cooperative game of setting access and transit prices for each other. They are linked economically through a sequence of providers forming a hierarchy, and we study their interaction by considering a multi-stage game. We will also consider the economics of private exchange points and show that they could become far more wide spread then they currently are. This is joint work with Srinivas Shakkottai (UIUC) (see [1]).

In this talk we provide an overview of our work on autonomic service management systems. The purpose of these systems is to ensure delivery of a multiplicity of network-based applications and content, according to service metrics and in the face of variable demand and equipment failure. We begin by describing our vision of the application-oriented infrastructure that must emerge from the convergence of computing and communications technologies. We then describe an autonomic service architecture that builds on this highly flexible and controllable infrastructure to support the delivery of applications at very low cost and to enable the introduction of new services at very rapid pace.

We describe approaches for creating resource abstractions that accommodate communications and computing capabilities so that service management can be carried out by a single common control framework. We also describe how an application can be supported by the allocation of virtual resources that are managed by autonomic managers that ensure that service metrics are met even in the presence of changing demand. We then present recent results for autonomic resource management algorithms for: 1. the management of network flows; 2. the management of application flows; 3. the management of radio resources in cognitive radio settings.

What are the big movements in networking that researchers should heed? A standout is the global spread of communities of interest (the networking analogue of the flat world) and their need for “dynamic virtual networks” that support rich applications requiring resources from several domains. The imperative for inter-networking, i.e., the enablement of coordinated sharing of resources across multiple domains, is certain. This challenge has many facets, ranging from the organizational, e.g., different, possibly competing, owners to the technical, e.g., different technologies. Yet another key characteristic of the emerging networking environment is that the service provider is required to handle ever-increasing uncertainty in demand, both in volume and time. On the other hand there are new instruments available to handle the challenge. Thus, inter-networking and uncertainty management are important challenges of emerging networking that deserve attention from the research community.

We describe research that touch on both topics. First, we consider a model of data-optical inter-networking, where routes connecting end-points in data domains are concatenation of segments in the data and optical domains. The optical domain in effect acts as a carrier’s carrier for multiple data domains. The challenge to inter-networking stems from the limited view that the data and optical domains have of each other. Coordination has to be enabled through parsimonious and qualitatively restrictive information exchange across domains. Yet the overall optimization objective, which is to maximize end-to-end carried traffic with minimum lightpath provisioning cost, enmeshes data and optical domains. This example of inter-networking also involves two technologies. A mathematical reflection of the latter fact is the integrality of some of the decision variables due to wavelengths being the bandwidth unit in optical transmission. Through an application of Generalized Bender’s Decomposition the problem of optimizing provisioning and routing is decomposed into sub-problems, which are solved by the different domains and the results exchanged in iterations that provably converge to the global optimum.

In turning to uncertainty management we begin by presenting a framework for stochastic traffic management. Traffic demands are uncertain and given by probability distributions. While there are alternative perspectives (and metrics) to resource usage, such as social welfare and network revenue, we adopt the latter, which is aligned with the service provider’s interests. Uncertainty introduces the risk of misallocation of resources. What is the right measure of risk in networking? We examine various definitions of risk, some taken from modern portfolio theory, and suggest a balanced solution. Next we consider the optimization of an objective which is a risk-adjusted measure of network revenue. We obtain conditions under which the optimization problem is an instance of convex programming. Studies of the properties of the solution show that it asymptotically meets the stochastic efficiency criterion. Service providers’ risk mitigation policies are suggested. For instance, by selecting the appropriate mix of long-term contracts and opportunistic servicing of random demand, the service provider can optimize its risk-adjusted revenue. The “efficient frontier”, which is the set of Pareto optimal pairs of mean revenue and revenue risk, is useful to the service provider in selecting its operating point.

The emerging trends in the market place are driving towards convergence at device, network, and application levels. Today consumers are looking for a single device capable of fulfilling their daily communication, computing, and personal productivity application needs. Businesses are working tirelessly to offer enhanced quality of service and application portability to consumers irrespective of their locality, mobility, and system preferences. These user preferences squarely challenge existence of stovepipe platforms, islands of interoperable systems, centralized command and control structure, and fragmented connectivity landscape. The new paradigm is all about being technology, platform, operating system, and network agnostic. There are clear indications, which confirms this transformation - e.g. connectivity migrating form circuit switched to packet switched, system resources and application transitioning from centralized to distributed (e.g. thin client, P2P), applications becoming platform and operating systems agnostic (e.g. Java applets), fixed and mobile convergence (IP centric - IMS), and aggregation of voice, data, and video (triple play).

This paradigm shift brings challenges and opportunities to system developers and raises the threshold of entry. To become a successful player in this new world one needs to understand the full value chain of the new ecosystem. The existing base of deployed system, equipments, and hierarchies can’t be overhauled overnight due to business, political, and regulatory reasons and one need to work in the confine of these constraints. Coexistence and collaboration across the heterogeneous systems and platforms need to be evaluated and commonality across must be exploited. We have to take advantage of the recent advancements in VLSI, system integration, and signal processing technologies. The mobile industry has been in the forefront and enabler of this transformation. They have taken full advantage of Moore’s law in integrating myriad of complementary applications such as WLAN, BT, 3D gaming, GPS, camera, audio/video streaming, and personal productivity gadgets to name a few. And this all has been done under the constraints of the device size and battery power. It is no coincidence that today’s mobile device deliver the capabilities of PC of yester years.

There has been a lot of progress already made towards the global convergence but more needs to be done. If current trend continues then there is a safe bet to say that today’s mobile evolution would be able to reduce the “digital divide” for the human kind globally.

The required capacity for an IP TV distribution network decreases if the right TV channels are multicast, while the others rely on unicast. In [1] we proposed mathematical models to determine this required capacity in two network scenarios: the “static” and “dynamic” scenarios. As the exact calculations turned out to be too tedious, we developed approximative models, assuming the variables have a Gaussian distribution and proved that the approximations yield results close to the exact results. In this paper, we build further on this approximation. First, we determine the optimal parameters in the static scenario. Second, we compare the static case resource demand function (with optimal settings) to the dynamic case function and we prove the superiority of the dynamic dimensioning approach. Then, the applicability regions of both scenarios (where gain is achieved) are explored. Finally, we illustrate the methodology by calculating the required capacity in some realistic examples.

Large-scale Video-on-Demand (VoD) systems with high storage and high bandwidth requirements need a substantial amount of resources to store, distribute and transport all of the content and deliver it to the clients. We define an extension to the as determining the number and model of VoD servers to install at each potential replica location to minimize deployment costs for a given set of distributed demand and available VoD server models. We propose three novel heuristics that generate near-optimal solutions and show that the number of replica sites for networks where the load is unevenly distributed is low (35 − 45% of potential locations), but that the hit ratios at deployed replicas are high (> 85%).

IPTV over DSL access networks is becoming a reality. This paper studies the impact of the presence of a video-aware traffic management module in the access network on the performance of the video dejittering mechanism at the receiver. The presented numerical results illustrate that with the traffic management module, the packet loss probability at the dejittering buffer decreases. Since the traffic management module will only drop the least important enhancement layer packets of the video in case of congestion, especially the important base layer packets of the video gain from the use of the traffic management module. All results in this paper are obtained using a mathematical model.

Semi-Markov processes (SMPs) are widely used to model various types of data traffic in communication networks. Also, efficient and reliable analysis techniques are available. In this paper, we consider several present methods of deriving the parameters of a discrete-time semi-Markov process from given H.264 video traces in order to model the original traffic adequately. We take the distribution of frame sizes and the autocorrelation of both the original trace and the resulting SMP model into account as key quality indicators. We propose a new evolutionary optimization approach using genetic programming, which is able to significantly improve the accuracy of semi-Markov models of video traces and, at the same time, requires a smaller number of states.

This paper presents the results of a simulation study that assessed SIP based VOIP networks under overload. This work addresses the issue of network level congestion controls in SIP based telephony networks. The simulation network consists of Media Gateways and Call Controllers, each with internal overload detection and control mechanisms. The simulation includes SIP timers tuned to operate gracefully with PSTN interfacing protocols. The traffic model is SIP based VoIP calls not involving application server or media server interactions. This work demonstrates that combining external overload controls with internal overload controls reduces blocking and increases goodput across the overload levels that ranged from 1 to 4 times the engineered load.

The paper considers a problem of routing, protection and load balancing in the IP/MPLS network. A network design problem combining all these aspects is presented. Proportionally fair distribution of residual bandwidths on links is used for load balancing, and protection is achieved with failure–dependent backup paths. The efficiency of the proposed approach is tested by combining optimization and simulation tools. Numerical experiments show that using the proposed load balancing and protection mechanisms decreases the number of disrupted LSPs in case of failures, as compared to other recovery options considered.