ABSTRACT
This chapter focuses on such trade-offs in the context of telecommunications networks and provides a framework for evaluation and presentation of such trade-offs. It presents the (α, β)-fairness concept and the a-fairness, which specify how far a resource allocation is allowed to be from an "ideally fair" allocation, according to any fairness criterion. The chapter describes a framework to produce "efficiency-fairness functions" that allow network operators to first set fairness constraints and then to optimize their efficiency. It formulates an nonlinear programming (NLP) problem, which finds the optimal rate allocation for a general network and any ideally fair rate allocation, under the (α,β)-fairness constraints and the a-fairness constraints. The chapter discusses the two global optimization algorithms, Lipschitz Global Optimizer (LGO) and AGOP, to solving the NLP for a variety of networks. For the examples tested, AGOP seems to be a particularly promising algorithm.
