Pattern recognition in cognitive communication

Abstract

A cognitive communication method based on recognition of continuous signal patterns respecting to digital data streams is proposed in this study. Digital signals to be transferred are first converted to optimally chosen patterns from a pre-defined glossary with respect to bandwidth requirement and medium conditions. Receivers recognize the patterns from the chosen frequency band of the spectrum via neural network. The received patterns are matched to respecting data sequence within the common glossary to recover the information. The purpose of this study is increasing the spectrum efficiency via manageable Signal to Noise Ratio. Pattern based real-time encoding and neural network based decoding technique is used here. It is shown that the transmitted information by multiple sources can be decoded by the receivers in the same time interval, frequency band and location by using this method.