Network dimensioning with maximum revenue efficiency for the fairness index

Abstract

Network dimensioning is a specific kind of the resource allocation problem. One of the tasks in the network optimization is to maximize the total ow on given pairs of nodes (so-called demands or paths between source and target). The task can be more complicated when different revenue/ profit gained from each unit of traffic stream allocated on each demand is taken into account. When the total revenue is maximized the problem of starvation of less attractive paths can appear. Therefore, it is important to include some fairness criteria to preserve connections between all the demands on a given degree of quality, also for the least attractive paths. In this paper, a new bicriteria ratio optimization method which takes into account both, the revenue and the fairness is proposed. Mathematical model is built in a form of linear programming. The solutions are analyzed with some statistical measures to evaluate their quality, with respect to fairness and efficiency. In particular, the Gini's coefficient is used for this purpose.