We consider the problem of minimizing the span of frequencies required to s
atisfy a certain demand in a cellular network under certain interference co
nstraints. A new iterative algorithm exploiting the special nature of such
systems is presented. The general procedure has the ability to react to var
iations of the traffic demand as more and more channels are being assigned
to cell requirements. Allocations of channels to cells are made,vith a meth
od that borrows insight from the theory of convex maximization. This method
is, however, equivalent to simple and fast heuristics when selecting prope
r values for its parameters. Our technique yields quite encouraging results
, showing that it represents an efficient alternative to attacking this typ
e of problems.