Cellular systems have hitherto been modeled mostly by graphs for the purpose of channel assignment. However, hypergraph modeling of cellular systems offers a significant advantage over graph modeling in terms of the total traffic carried by the system. For example, we show that a 37-cell system when modeled by a hypergraph carries around 30% more traffic than when modeled by a graph. We study the performance of channelized cellular systems modeled by hypergraphs in comparison with those modeled by graphs. For this purpose, we have evaluated the capacities of these cellular networks defined [3]. Evaluation of the capacity necessitates generation of maximal independent sets of hypergraphs. We describe some new algorithms that we have developed for this purpose.