signal reconstruction is significant to communication countermeasure,there we focuse on the algorithm of signal reconstrction.

1,a neural network is build to claify the modulation of signals,which can reach the accuracy of 98% when signal snr is betweeen -15db and 15db.
2,using GAN to regeneration signal similar to the original signal ,like some application in Image Generation,GAN,WGAN,LSGAN were studied.
3,Conditional GAN is applied to generation signal with some different modulation signals mixed,this algorithm can produce signal on the basis of features.

a GAN architecture is build like below,the loss functions includs GAN,WGAN,LSGAN

both the generator and discriminator is based on DNN,whose architectures are shown below

data: generated signal data of different modulation
imag:picture used in readme
log :log info of different modulations signal regeneration algorithm
model ：model meta info which can be recovery when needed
result :generated signal data shown by picture
src : source code

Analog_Clarify.py: clarify signal modulation by traditional methods
Analog_signal_LSGAN.py: regeneration signal data by LSGAN
Signal_Generation_Param_select_2.py: select best params of GAN
Signal_Regeneration_GAN: regeneration signal data by GAN ，which can produce different signal data through different model
modulation.py :clarify signal modulation by neural network