Open AccessIdentification of Analogue Modulated Signal via Artificial Intelligence
Author(s) -
Mohammed S. Khalaf,
Aeizaal Azman A. Wahab,
S. S. N. Alhady,
Haliza Mat Husin,
W. A. F. W. Othman
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1818/1/012003
Subject(s) - classifier (uml) , computer science , artificial intelligence , demodulation , pattern recognition (psychology) , feature extraction , signal processing , speech recognition , telecommunications , channel (broadcasting) , radar
Referring to a huge usage of the analogue modulated signal in both civilian and military applications, and impotency of identification it, the importance is came from many reasons also to how to get an algorithm to recognize an analogue modulated signal via AI for high performance of identification and simple to use. The identification is a middle-step between recognizing and signal demodulation. Selecting the most suitable variables are for input modulated signal and with (AI) assistant to identify some of analogue modulated signals such as (AM, DSB, SSB, and FM). Using simulation with Gaussian noise (0 dB to 10 dB) and AI. Classification system consisting from 3 parts: 1-Pre-processing stage which is talking about Feature keys extraction did. The element keys extraction did so as to get input include keys for the AAMR classifier, 2-Training stage which talked about the preparation, and Analog programmed adjustment acknowledgment (AAMR) classifier advancement., and 3-Testing stage which talked on the classifier performance evaluated using 25% of all generated data as a test data. Using two algorithms (SCG and CONJGRAD), In the SCG, the possibility of each single key feature for each single modulation scheme with variance SNR results got. A couple of methods compared and the results show that the performance of the AI based recognition system is superior to the decision classifiers based on way with less SNR values. Both of the suggested identification systems are capable to discriminate (AM, DSB, SSB, and FM) at SNR>5 dB with a success rate 100%.