A new Markovian approach for dynamic bandwidth allocation in wireless networks

In wireless networking, users mobility has a heavy impact on QoS parameters and the existing architectures for real-time services with motionless hosts become inadequate for QoS management. In particular Doppler shift and multipath fading introduce some degradations in wireless transmissions. For these reasons, the management of real-time flows and handover events (in terms of bandwidth guarantee and service continuity) when the system is dealing with Mobility Independent Predictive (MIP) services is mandatory, if QoS constraints need to be satisfied. In order to offer a soft QoS increasing the total system utilization, a bandwidth reallocation algorithm has been considered for dynamic resource adaptation, based on wireless channel modeling, in order to take into account the degradations related to the channel state conditions. The main idea of this proposal is the utilization of a pre-reservation phase in the admission control through a Markovian approach, in order to predict the amount of bandwidth needed by a mobile host during its movements among active and passive cells. The performance evaluation of the proposed idea has been made in terms of total admitted/dropped MIP flows, assigned bandwidth and system utilization.