Analysis of prediction-oriented features in dynamic bandwidth reservation schemes

Out of Quality of Service reasons, higher priority shall in cellular networks be assigned to ongoing connections than to new call attempts. This can be accomplished through reserving a portion of the bandwidth assigned to a cell exclusively for high priority connections. A low priority call is only admitted to a cell if the unoccupied bandwidth in the cell following the acceptance of the call exceeds or is equal to the reserved bandwidth. In order to improve the efficiency of the bandwidth reservation scheme, the amount of bandwidth to reserve can be made time-varying and adaptive towards the momentary network conditions. Subscriber movement predictions can be applied to obtain handover arrival rate estimations, which can be used as input to the bandwidth reservation allocation procedure. The subscribers can for instance be grouped according to speed and previous cells. In this paper, a Markovian analysis of such prediction-oriented grouping features will be performed. The probability for a new call attempt to be blocked and the probability for a handover attempt to fail, due to channel occupancies, will be calculated for a bandwidth reservation scheme with and without grouping features. The obtained analytical results will be validated using computer simulations. © 2006 IEEE