Prediction-based techniques for hand-off prioritization in channel assignment schemes

We investigate prediction-based techniques for prioritizing hand-offs in a general class of channel assignment schemes. We generalize the concept of guard channels to that of guard or reject states. Hand-off prioritization is implemented by imposing different call acceptance criteria for hand-off call requests and new call requests. Hand-off calls are always accepted if a channel is available whereas new calls are accepted or rejected based on predictions about the future states of the system. We consider two criteria to classify the system state: maximum likelihood state (MLS) criterion and cost function criterion (CFC). Both methods accept or reject new calls based on the statistics about the mobility and death processes of the users and are able to adapt to non-uniform traffic in the cells. They are both parameterized by the look-ahead time /spl tau/, which can be used to adjust the hand-off failure and new call blocking probabilities. We also propose a simplified prioritization scheme which classifies states based on the total number of users in the system.