Bounds on the call blocking for Dynamic Channel Allocation algorithms in cellular networks are presented. These bounds are tighter than the well known Erlang bound and can be derived by introducing a system of rewards for the calls arriving to various cells within the network.

This approach enables the examanation of the maximum packing scheme introduced by Everitt and McFadeyn. We show that it is close to optimal for conventional DCA schemes (those which respect a common reuse constraint). However results from an exact analysis for a maximum packing scheme with varying reuse constraints do not produce results which lie close to the corresponding bound. Indeed an asymptotic analysis shows that maximum packing applied to networks with varying reuse constraints can be neither fair nor optimal.