Predictive resource allocation in cellular networks using Kalman filters

Controlling handoff drops in a cellular network is a very important QoS issue. In order to keep the handoff-dropping probability as low as possible, we have designed a new predictive scheme for radio resource allocation. This scheme consists in dimensioning the amount of resources to reserve for handoffs in each cell. The solution proposed is based on a reservation mechanism with a threshold estimated according to the usage history of the resources in the cell. This estimation is done using the Kalman filter, a powerful mathematical tool. Evaluation of the proposed scheme demonstrates a real improvement in the handoff-dropping probability compared to the static reservation scheme.