Laura Wynter

A survey on networking games in telecommunications

In this survey, we summarize different modeling and solution concepts of networking games, as well as a number of different applications in telecommunications that make use of or can make use of networking games. We identify some of the mathematical challenges and methodologies that are involved in these problems. We include here work that has relevance to networking games in telecommunications from other areas, in particular from transportation planning.

Equilibrium, Games, and Pricing in Transportation and Telecommunication Networks

Network equilibrium models that have traditionally been used for transportation planning have penetrated in recent years to other scientific fields. These models have recently been introduced in the telecommunication networks literature, as well as in the field of game theory. Researchers in the latter fields are not always aware of the very rich literature on equilibrium models outside of their application area. On the other hand, researchers that have used network equilibrium models in transportation may not be aware of new application areas of their tools. The aim of this paper is to present some central research issues and tools in network equilibria and pricing that could bring closer the three mentioned research communities.

Parameter inference of queueing models for IT systems using end-to-end measurements

Performance modeling has become increasingly important in the design, engineering and optimization of information technology (IT) infrastructures and applications. However, modeling work itself is time consuming and requires a good knowledge not only of the system, but also of modeling techniques. One of the biggest challenges in modeling complex IT systems consists in the calibration of model parameters, such as the service requirements of various job classes. We present an approach for solving this problem in the queueing network framework using inference techniques. This is done through a mathematical programming formulation, for which we propose an efficient and robust solution method. The necessary input data are end-to-end measurements which are usually easy to obtain. The robustness of our method means that the inferred model performs well in the presence of noisy data and further, is able to detect and remove outlying data sets. We present numerical experiments using data from real IT practice to demonstrate the promise of our framework and algorithm.

A New Look at the Multiclass Network Equilibrium Problem

The multiclass network equilibrium problem is expressed in general as a nonmonotone, asymmetric, variational inequality problem. We show that in spite of the nonmonotonicity of the cost operator, the problem may actually satisfy a weaker property, induced by the hierarchical nature of the travel cost interactions. This property allows a natural decomposition approach, not otherwise available, that admits provably convergent algorithms. We present one such algorithm, easily implementable using a solver for the single-class network equilibrium problem, together with a convergence proof.

Competitive equilibrium in e-commerce: Pricing and outsourcing

The success of firms engaged in e-commerce depends on their ability to understand and exploit the dynamics of the market. One component of this is the ability to extract maximum profit and minimize costs in the face of the harsh competition that the internet provides. We present a general framework for modeling the competitive equilibrium across two firms, or across a firm and the market as a whole. Within this framework, we study pricing choices and analyze the decision to outsource IT capability. Our framework is novel in that it allows for any number of distributions on usage levels, price-QoS tradeoffs, and price and cost structures.

Incident Duration Prediction with Hybrid Tree-based Quantile Regression

Accurate prediction of incident duration is critical for efficient incident management which aims to minimize the impact of non-recurrent congestion. In this chapter, a hybrid tree-based quantile regression method is proposed for incident duration prediction and quantification of the effects of various incident and traffic characteristics that determine duration. Hybrid tree-based quantile regression incorporates the merits of both quantile regression modeling and tree-structured modeling: robustness to outliers, simple interpretation, flexibility in combining categorical covariates, and capturing nonlinear associations. The predictive models presented here are based on variables associated with incident characteristics as well as the traffic conditions before and after incident occurrence. Compared to previous approaches, the hybrid tree-based quantile regression offers higher predictive accuracy.

Revenue management model for on-demand IT services

This paper presents a model for applying revenue management to on-demand IT services. The multinomial logit model is used to describe customer choice over multiple classes with different service-level agreements (SLAs). A nonlinear programming model is provided to determine the optimal price or service level for each class. Through a numerical analysis, we examine the impacts of system capacity and customer waiting incentives on the service providers profit and pricing strategies.

Capacitated Network Revenue Management through Shadow Pricing

In this paper, we analyze a method that links Lagrange multipliers from a resource allocation problem to the problem of revenue or profit maximization. This technique, first proposed in the transportation science literature by [7] has important implications for telecommunication network pricing. Indeed, the framework provides a generalization of telecommunication resource allocation/shadow price-based schemes such as those of [6] and [9], in that it permits the optimization of the shadow prices themselves, through a computationally simple procedure. We analyze the extent to which revenue can be maximized on a network that uses shadow-price-based prices, and how to deal with cases of unbounded multipliers.

Usage-based versus flat pricing for e-business services with differentiated QoS

Design of competitive e-commerce services, in such a quickly responding market, requires the analyses of prices and price structures. We present a general model of an e-commerce market that allows us to analyze optimal price structures, both flat and usage-based. Based on the price structure of a major Web hosting provider, we consider both single-tier and two-tier (burst-rate) pricing, and our result suggests that the more complex two-tier structure may not be worth the marketing effort, as the firms equilibrium profits will not increase through the use of this structure. An essential feature of our approach is that we model explicitly the spread of price-QoS tradeoffs across the end-user population.

Optimizing Proportionally Fair Prices

We demonstrate the non-uniqueness of proportional fairness prices, and show that the model of Kelly et al. [9] falls into the category of models that can be generalized to allow optimization of the link prices, with a view toward profit maximization. However, through this framework, we also show that the total revenue that can be obtained through proportional fairness pricing on a network is unique.