An adaptive framework for addressing fairness issues in wireless networks

We present a framework for designing optimal policies addressing resource allocation problems in wireless networks. We consider a general utility function optimization objective. Specific choices of the utility functions lead to policies that satisfy several well-known fairness criteria, e.g. max-min and proportional-fair. Traditional approaches for solving these nonlinear optimization problems in an off-line manner, lead to nonadaptive policies that usually rely on system parameters, which may not be known a priory. Within our framework the development of such policies is based on the adaptive employment of policies that solve linear optimization problems. In several situations the development of policies for these linear problems is fairly simple and depends minimally on system parameters. Subsequently, we apply this method to three specific wireless resource allocation problems. In particular we consider wireless fading channel systems and provide optimal policies for (a) uplink optimal power allocation for constant bit rate connections, (b) uplink optimal average throughput allocation, and (c) downlink optimal scheduling with limited transmission rate capabilities over multiple fading channels.