Extreme Maximal Weighted Frequent Itemset Mining for Cognitive Frequency Decision Making

Cognitive Frequency Decision Making (CFDM) is a new application in cognitive radio ad hoc network with limited communication capability, and once solved by our algorithm Extreme Maximal Biclique Searcher (EMBS). In this paper, we extend the CFDM from one subnet to the whole network, and propose Common Frequency Searcher (CFS) to find the solution. CFS uses the result of a novel algorithm Maximal Weighted Frequent Itemset Mining (MWFIM) which is mainly discussed in this paper and also proposed by us to mine all maximal weighted frequent itemsets from transaction database of weighted items. We solve the extended CFDM problem by using the weight of item in a new fashion in which weight is independent of support and by traveling weighted itemset enumeration tree in a depth-first manner. When visiting nodes of the tree, we use two pruning conditions to speed up traveling and reduce computational time. Experimental results show that our algorithm can satisfy the CFDM application in real world at most times.