This paper develops a statistical framework to optimize the impact of a large number of simultaneous reactive handovers on the access network performance. In mobile communication the huge load of synchronous handovers can cause serious QoS degradation. To circumvent this effect we develop optimal stochastic timer distributions to spread the handover requests in time resulting in a much more balanced network load. To investigate the effect of the random delays on the handover mechanisms we have created a queueing model where the queues are organised into a tree topology. In this model single queues represent the performance of various system components, while the load which is caused by a handover request is modeled by a single packet. This packet is fed into one of the queues on the lowest level and then traverses the tree up to the root node. The QoS parameters of the handover are then calculated by evaluating the resulting packet loss and average delay in all the queues along the packet's route. The main challenge is to optimize the random delay mechanism (and to develop the optimal distribution of the corresponding stochastic timer), in order to maximize the QoS. In this way, a more robust reactive handover mechanisms can be created which is essential in low quality radio environment.
展开▼
