class DDoSDetector:
#DataHandler class object
data_handler = None
#ANN class object
neural_network = None
def __init__(self):
self.data_handler = DataHandler()
self.neural_network = ANN()
def train(self, dataset_index, pcap_index=None):
print("Dataset: "+str(self.data_handler.get_dataset_path(dataset_index)))
if pcap_index!=None:
print("PCAP: "+str(self.data_handler.get_pcap_path(dataset_index, pcap_index)))
packets = []
labels = []
packets = self.data_handler.get_packet_information(dataset_index, pcap_index)
labels = self.data_handler.get_labels(dataset_index, pcap_index)
#turns each packet data from dictionaries into a flat 1d list.
compressed_packets = self.data_handler.compress_packets(packets)
#takes compressed packet data and returns input variables values for neural network
input_data = self.data_handler.generate_input_data(compressed_packets)
#takes input variables and labels, and normalizes them
normalized_input, normalized_output = self.data_handler.normalize_compressed_packets(input_data, labels, dataset_index)
print("Num packets: "+str(len(normalized_input)))
print("Num labels: "+str(len(normalized_output)))
print("These should match")
num_true_labels = sum([ label for label in normalized_output ])
print("Num true labels: {}".format(num_true_labels))
print("Num false labels: {}".format(len(normalized_output) - num_true_labels))
#feeds input data and output data into the neural network
self.neural_network.train_model(normalized_input, normalized_output, dataset_index)
#dataset_index can specify a dataset to predict on, or if None,
# will represent predicting on live packets from "./Live sniffing"
def predict(self, dataset_index=None, pcap_index=None):
if dataset_index == None:
print("Dataset unspecified when calling predict()")
return
#if predicting from a dataset
if dataset_index!=None:
packets = self.data_handler.get_packet_information(dataset_index, pcap_index)
labels = self.data_handler.get_labels(dataset_index, pcap_index)
#turns each packet data from dictionaries into a flat 1d list.
compressed_packets = self.data_handler.compress_packets(packets)
input_data = self.data_handler.generate_input_data(compressed_packets)
normalized_input, normalized_output = self.data_handler.normalize_compressed_packets(input_data, labels, dataset_index)
print("Num packets: "+str(len(normalized_input)))
print("Num labels: "+str(len(normalized_output)))
#feeds input data and output data into the neural network
predicted_labels = self.neural_network.predict(normalized_input)
# self.data_handler.save_prediction(dataset_index, pcap_index)
#predicting live pcap files
def predict_live(self, dataset_index=None):
if dataset_index == None:
print("Dataset unspecified when calling predict()")
return
latest_pcap_path = self.data_handler.get_latest_live_pcap()
if latest_pcap_path=="":
print("There is no pcap file to predict from")
return
print("Latest pcap path: "+str(latest_pcap_path))
#returns normalized input data from the specified pcap path
normalized_input = self.data_handler.get_live_input_data(latest_pcap_path)
print("Num packets: "+str(len(normalized_input)))
latest_packet = [normalized_input[-1]]
#feeds input data and output data into the neural network
predicted_label = self.neural_network.predict(dataset_index, latest_packet)
if len(predicted_label) > 0:
predicted_label = predicted_label[-1][0]
print("Predicted label: "+str(predicted_label))
else:
print("No predictions for live data")
print()
