Modified Automatic Digital Modulation Recognizer for Software Defined Radio

Abstract

From Communication intelligence (COMINT) to the seamless working of 3G/4G networks, Automatic Modulation Recognition (AMR) has become prominent parts of the communication system. For Software-Defined Radio (SDR) or Cognitive Radio (CR), where seamless communication is targeted with no or minimum human intervention. Due to this, it is inevitable to have an AMR. AMR has to recognize any modulation technique both analog and digital. Most of the communications of the present era is digital, a subsection of Automatic Digital Modulation Recognition (ADMR) is derived in spite of the fact that AMR by default has to identify and state, the type of modulation, irrespective of analog or digital type of modulations used. This article presents feature-based digital modulation recognition. ASK2, ASK4, PSK2, PSK4, FSK2, and FSK4 are considered as the contenders for recognition. For classification of the incoming signal, decision tree-based approach is used. An analytic signal is built using the intercepted signal. Spectral features extracted from the analytic signal. Features are being used as a decision-making parameters. Thresholds fixation for each of parameter to make a decision is fixed by a large set of symbols as a single segment say 65,536 symbols per segment. The accuracy of 100% has been achieved even when there are symbols over 512 per segment or frame.