Revised architecture for automatic modulation recognition

摘要

Cognition and adaptability is a prominent aspect of Communication Intelligence (COMINT). Automatic Modulation Recognition (AMR) is one such aspect. AMR focuses on identifying modulation technique used at a particular carrier frequency which is got by spectral analysis (SA). The article deals with modulation recognition using feature based approach. This article addresses recognition of modulations both analog and digital sets. Analog modulations considered are Amplitude Modulation (AM), Frequency Modulation (FM), Phase Modulation (PM), Single Side Band Modulation (SSB) and Double Side Band Suppressed Carrier (DSB-SC). Digital modulations considered include Shift Keyings namely Amplitude Shift Keying (ASK), Frequency Shift Keying (FSK), Phase Shift Keying (PSK) with two and four symbols addressed as ASK2, ASK4, FSK2 , FSK4, PSK2, PSK4, respectively. Apart from this Quadrature Amplitude Modulation (QAM) with 8 symbols (QAM8) and Vestigial Side Band Modulation (VSB) stand as contenders for recognition. Its presumed that knowledge of carrier frequency,data rate and bandwidth of information is available with the recognizer; which might be extracted from SA. Analytic signal is built out of the intercepted signal frames and then Instantaneous Amplitude and Phase (IAP) is extracted from the signal frames. Statistical features are extracted from IAP. Intercepted signal frames are constructed using large sampling time, in other way considering large number of symbols in a frame which are used for threshold fixation for decision making. Finally a decision tree is designed and accuracy of the same is tested across varying modulations and noise levels.