skypitcher/project_dcnnmld

Offical repo for the paper titled "A MIMO Detector with Deep Learning in the Presence of Correlated Interference"

Project DCNN-MLD

Offical repo for the paper titled "A MIMO Detector with Deep Learning in the Presence of Correlated Interference"

Dependencies

1. pip install tensorflow-gpu==1.12 (you have to install cuda by your hand)
2. pip install matplotlib
3. pip install mpmath

How to run benchmark?

1. Establish all the dependencies
2. Keep calm and run python test.py

How to train model?

1. Generate training set and valid set
2. train your model there is a example code snippet for you

def generate_training_set_and_valid_set(rho, sir_db):
    print("Generating data sets, rho={:.1f} sir={}".format(rho, sir_db))
    training_set = DataSet(flag=1, rho=rho, sir=sir_db)  # flag == 1 : training set
    training_set.produce_all()

    valid_set = DataSet(flag=1, rho=rho, sir=sir_db)  # flag == 2 : valid set
    valid_set.produce_all()
    
def train_model(rho, sir_db, is_improved=True):
    model = DCNNMLD(rho, sir_db, is_improved=is_improved)
    model.train()
    
def gennerate_data_and_then_train_model(rho, sir_db, is_improved=True):
    generate_training_set_and_valid_set(rho, sir_db)
    train_model(rho, sir_db, is_improved)

License

Anti 996 License Version 1.0