ABSTRACT
We develop models for the collaborative provision of advanced data services around a service platform operated by a service aggregator. This situation is modelled as one leader/multiple followers Stackelberg game. The leader (service aggregator) decides the distribution of the revenue that maximizes his profit, while followers (service providers) select the allocation of their provision capacities by striking the balance between profit and risk while choosing the appropriate Nash equilibrium. The origin of risk is random demand, we consider both risk-neutral and risk-averse agents. Two approaches are pursued: one in the spirit of linear stochastic programming with uncertainty being described by a finite number of scenarios, leading to mixed integer linear programming problems of special structure. The second approach exploits the affinity with newsvendor problems, obtains explicit characterization of Nash equilibria and draws upon connection with portfolio theory. The models applied to the case study originated from the design of a platform for mobile data services.
