DDoS Attack Identification
RL-agent then tries to maximize the expected cumulative reward . In this learning setup , the selfsupervised Q-learning loss is defined as a cross-entropy loss . The cross-entropy loss is used to rank the sequence of events to a binarized indicator of the occurrence or absence of the anomalous event . This cross-entropy loss measures the performance of a classification model where the output may be a probability value between 0 and 1 . For example , an anomaly may be a 1 , and a non-anomaly may be a 0 . Then , via a neural network , may now generate probability values between 0 and 1 , to match the training data . If a value is 0.02 at a place where it is 1 , it may give a high loss value . A perfect model may have the cross-entropy loss to be 0 , that is , e . g ., that the 1 may be predicted as 1 . The self-supervised reinforcement learning module may then learn patterns in the data that may potentially give rise to local anomalies .
For the actor-critic variation , the self-supervised head is the “ actor ”, and the Q-learning module is the “ critic ”. The self-supervised learning network may be used to determine which factors to give more weightage to in the predictive model . Factors or features are combinations of variables that may give rise to an anomalous event , e . g ., combinations of KPIs , such as interference , load , atmospheric conditions , etc . The training loss of the approach versus the epochs is shown in Figure 3-4 .
Figure 3-4 : Decreasing loss of the self-supervised actor-critic method for anomaly detection .
Based on the above modeling , experimental results from this real-world dataset have been shown to accurately predict noise rise 60 % of the time , before it is strong enough to deny service . This ensures that we can instantiate other anomaly detection , causality , and countermeasures before the IoT DDOS has achieved full effect . To the best of our knowledge , this is the first work where DDOS anomalies are predicted and countered before they occur .
62 March 2022