Combining Markovian chains and real time agreement protocols to allocate resources in wireless systems

The increase in the number of users and the development of their needs towards multimedia services with quality of service expectations set challenges to mobile radio network operators whose spectral resources are limited. A judicious resource allocation policy will limit call dropping due to handovers and improve bandwidth allocation between users, taking into account factors related to QoS expectations. QoS acceptance is defined by the SLA contract specifying the parameters of the different levels of acceptance. At global level, new call or handover call acceptance is based on a collective and consensual decision between the different participating networks. Such a consensus aims to take a consistent decision to reserve resources according to one of the levels of service predefined by the caller SLA and then to allow the caller to renegotiate the level of service QoS according to traffic load fluctuation, or to negotiate new resources for new applications requested during the session. The consensus algorithm is parametrized in accordance with the network configuration and take account of QoS constraints such as: new call blocking probability, handovers dropping probability, channels utilization rate...These parameters are deduced by applying Markovian chain with two dimensions. It is adopted to model the different service classes which negotiate the network.