Deep learning based modulation recognition in wireless channel | |
| Author | Poonyavee Tabyam |
| Call Number | AIT Thesis no.TC-19-02 |
| Subject(s) | Machine learning Pattern recognition systems Modulation (Electronics) Neural networks (Computer science) |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Telecommunications |
| Publisher | Asian Institute of Technology |
| Abstract | In this thesis, we investigate the effect of data representation and pre-processing technique on deep neural network models for modulation classification task. For the overall system of interest, RadioML2016.10a dataset of (West & O’Shea,2016), which is generated by GNU’s Not Unix (GNU) radio and contains 11 types of modulations, is used as benchmark input data. Convolutional neural network (CNN), convolution long short-term deep neural network (CLDNN) architectures of (Liu et al.,2017) and long short term memory (LSTM) network architecture of (Rajendran et al.,2018) are trained and tested with several data representation, namely in-phase and quadrature (IQ), fast Fourier transform (FFT), amplitude and phase data, and pre-processing technique, namely normalization and feature extraction. We have found that the highest classification accuracy of all tested neural network model is obtained when using normalized amplitude and phase data as an input. Four convolutional layer CNN, CLDNN and LSTM have achieved approximately 90% accuracy at high signalto-noise ratio (SNR) when using min-max and L2 normalized amplitude and phase data as an input. Moreover, LSTM has significantly lower graphics processing unit (GPU) memory consumption while four convolutional layer CNN has significantly lower computation time among all tested neural network architecture. |
| Year | 2019 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Telecommunications (TC) |
| Chairperson(s) | Attaphongse Taparugssanagorn; |
| Examination Committee(s) | Teerapat Sanguankotchakorn;Poompat Saengudomlert; |
| Scholarship Donor(s) | Royal Thai Government;AIT Fellowship; |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2019 |