Efficient Dynamic Spectrum Allocation in Self-Organized Cognitive Radio Networks

The cognitive radio network (CRN) is one of the key technologies that will enable the future 5G wireless communication networks. This work proposes the integration of temporal correlation principles such as the Least Recently Used policy with two stages of active learning using self-organizing (SO) map for dynamic spectrum allocation (DSA) in CRN. Similar to web-caching in computer memory where the locality of reference plays an important role, in CRNs each CR unit stores an array of channel weights which is updated by the notion of locality of reference. The least recently used sub-band is given more priority in the weight array and is assigned to the CRs for communication contrary to caching, where the least recently requested document is replaced by the current request. The SO-distributed DSA coupled with temporal correlation ordering is a novel innovative technique thought of for CRN. Various simulation experiments validate the performance of the proposed approach.