Wyniki wyszukiwania - BazTech

1

Zastosowanie transformacji falkowej i sieci neuronowej w rozpoznawaniu typu modulacji sygnału

Walenczykowska M., Kawalec A.

Elektronika : konstrukcje, technologie, zastosowania

|

2012

|

Vol. 53, nr 9

120-123

PL

W artykule omówiono sposób rozpoznawania rodzaju modulacji przy użyciu transformacji falkowej (CWT) oraz sieci neuronowej. Przeanalizowano sygnały M-PSK, M-FSK, M-QAM, M-ASK oraz MSK. W celu wyodrębnienia cech wykorzystano wartość średnią i odchylenie standardowe |CWT| sygnałów nieunormowanych oraz unormowanych. Natomiast skuteczność sieci przetestowano dla sygnałów o stosunku sygnału do szumu równym kolejno 20, 10, 6 oraz 3 dB. Poprawność klasyfikacji wynosiła od 100% (dla S/N = 20 dB) do 81% (dla S/N = 3 dB).

EN

The article investigates possibility of modulation type recognition using wavelet transform (CWT) and neural network. There are five types signal modulation analised: M-PSK, M-FSK, M-QAM, M-ASK and MSK with and without normalization. The mean value and standard deviation of CWT are used as a signal features. Two layer neural network with backpropagation algorithm for training is proposed as classifier. Effectiveness of classifier is 100% for 20 dB signal to noise ratio to 81% for 3 dB signal to noise ratio.

2

Adaptive procedure for automatic modulation recognition

Liedtke F.

Journal of Telecommunications and Information Technology

|

2004

|

nr 4

91-97

EN

An adaptive procedure for automatic modulation recognition is described. With it the automatic modulation classification and recognition of radio communication signals with a priori unknown parameters is possible effectively. The results of modulation recognition are important in the context of radio monitoring or electronic support measurements. The special features of the procedure are the possibility to adapt it dynamically to nearly all modulation types, and the capability to recognize continuous phase modulation (CPM) signals like Gaussian minimum-shift keying (GMSK) too. A time synchronization to the symbol rate is not necessary.

3

Recognition of narrowband radio signals using autoregressive models and pattern comparison approach

Łopatka J.

Journal of Telecommunications and Information Technology

|

2002

|

nr 1

56-59

EN

This paper presents an improved spectral recognition method for digitally modulated radio signals. It is based on a signal autoregressive (AR) model. Model poles are used as signal features for neural network based on recognition process. To reduce an influence of the signal noise and distortions introduced by the digital receiver, a nonlinear Z plane transformation is proposed.

4

Fuzzy logic classifier for radio signals recognition

Łopatka J., Pędzisz M.

Journal of Telecommunications and Information Technology

|

2001

|

nr 2

31-34

EN

This paper presents a new digital modulation recognition algorithm for classifying baseband signals in the presence of additive white Gaussian noise. Elaborated classification technique uses various statistical moments of the signal amplitude, phase and frequency applied to the fuzzy classifier. Classification results are given and it is found that the technique performs well at low SNR. The benefits of this technique are that it is simple to implement, has generalization property and requires no apriori knowledge of the SNR, carrier phase or baud rate of the signal for classification.