Game Theory and Reinforcement Learning for inter-provider SLA negotiation

The emergence of high-demanding applications over the Internet leads to reconsider economic agreements among Network Service Providers (NSPs) in order to guarantee Quality of Service (QoS) across various networks (e.g. end-to-end delay, bandwidth, etc.). Service Level Agreements (SLA) sustaining both QoS and business commitments have to be negotiated among NSPs. Meanwhile, customer utility have to be taken into account in order to make acceptable offers. Hence, we propose that at a high level, network providers regroups into alliances [1]. Such alliances can be administrated by a central neutral third-party to reduce common costs or in a distributed manner to compel with NSPs' requirements on confidentiality and independence.