| Due to the development of network technology,network security problem is becoming more and more serious.Distributed Denial of Service(DDoS)attack is a common attack method used by hackers at present.It is difficult to prevent by exhausting the resources of the target host and denying legitimate users normal access to the network system resources,which poses a great threat to organizations and companies.Once attacked,it is often accompanied by huge losses.So now also developed a lot of detection methods for DDoS attack.At present,most attack detection methods are based on statistics.Even though machine learning and deep learning have been widely used in the detection of DDoS attacks,they also rely on a large number of feature projects and rely too much on the input features.The quality of the classification results depends on the selected features.Therefore,based on the original traffic data,a DDoS attack detection method based on gray level co-occurrence matrix and improved genetic algorithm to optimize CNN structure was proposed.In the first method,the gray level co-occurrence matrix is used in the data processing stage to reconstruct the data and extract the spatio-temporal characteristics of the data,and the binary matrix is generated by combining with the original data.These matrices together serve as inputs to the neural network;Secondly,the improved genetic algorithm is used to automatically learn to determine the optimal network structure of the convolutional neural network,and the model is trained by the determined network structure.The main work of this paper is as follows:1)A data reconstruction method based on gray level co-occurrence matrix is proposed.The part of the original traffic data is processed in two directions at the same time.One is to generate the decimal matrix according to the bytes of the data,and then generate the corresponding horizontal and vertical grayscale co-occurrence matrix,and carry out normalization processing and filling processing.The second is to convert the data into binary numbers to form a bit matrix consisting of zeros and ones.These two processing directions will eventually form a matrix of the same size,which will be used as the input part of the subsequent detection.The effective information of the original data is retained,and the spatial and temporal characteristics of the data are further extracted,thus avoiding a lot of feature engineering.2)A convolutional neural network structure optimization method based on improved genetic algorithm is proposed.By increasing the convolutional layer information coding and the full connection layer information coding of the network,the coding mechanism is improved,and the genetic operation of the population is improved,so that the genetic operation is applicable to the coding method,improving the quality of the whole population and getting a better optimal network structure.Finally,through the experimental comparison data,it can be seen that the detection method proposed in this paper has a good detection effect.3)According to the detection method proposed in this paper,a DDoS attack detection system is designed and implemented,the overall framework of the system and the functional modules of the system are constructed,the functions of each functional module and the docking method between each module are defined,and the workflow of each module is systematically introduced.Finally,the function and performance of the system are verified.Through theoretical and experimental verification,it is proved that the detection method proposed in this paper can effectively detect DDoS attacks. |
