Optimal channel assignment in cellular networks with non-homogeneous demands

This paper deals with the channel assignment problem (CAP) in a hexagonal cellular network with 2-band buffering, where the interference does not extend beyond two cells. Here, we first present new lower bounds on the required bandwidth for a cellular network with homogeneous demand ω, for various relative values of s₀, s₁, and s₂, the minimum frequency separations required to avoid interference for calls in the same cell, or in cells at distances one and two respectively, where s₀≥s₁≥s₂. Next we present some novel strategies for assigning channels to the cells of the entire cellular network with a homogeneous demand w using a genetic algorithm. We then show how these strategies for homogeneous demand can be extended to the cases with nonhomogeneous demand vector W=(w_i), where w_i represents the channel requirement for cell i. Most interestingly, it shows that in some cases depending on s₀, s₁ and s₂ the required bandwidth is mainly determined by the maximum demand w_max in W. It shows that in terms of bandwidth requirement the case is almost equivalent with cellular networks having homogeneous demand w_max. The technique also results optimal bandwidth solutions for all the eight well-known benchmark instances including the most difficult two.