A novel DDoS detection approach based on Cellular Neural Network (CNN) model in cloud computing is proposed in this paper. Cloud computing is a new generation of computation and information platform, which faces many security issues owing to the characteristics such as widely distributed and heterogeneous environment, voluminous, noisy and volatile data, difficulty in communication, changing attack patterns. CNN is an artificial neural network which features a multi-dimensional array of neurons and local interconnections among cells and CNN can be used to solve the cloud security difficulties according to the nature of non-linear and dynamic. RPLA and Tabu optimized algorithm is employed to learn the CNN classifier templates and bias for DDoS intrusion detection in cloud computing. Experiments on DDoS attacks detection show that whether RPLA-CNN or Tabu-CNN models are effective for DDoS Attacks detection. Results show that CNN model for DDoS attacks detection in cloud computing exhibits an excellent performance with the higher attack detection rate with lower false positive rate.

No abstract

With the development and popularization of computer, network video teaching is the new medium of school teaching. But because the scale of network video teaching system and broadband carrying capacity are limited, network video is hard to satisfy the massive need in school, thereby it could lead to sharp drop in network video teaching quality. Through establishing a simulation platform based on OPNET, the article analyse the video teaching system under network construction of school network; compare and discuss a optimization program of network current-limiting; and make a simulation test on optimization program of network cureent-limiting. The result show that through selecting right optimization program of network cureent-limiting, we can make s big promotion in system carrying capacity, and enhance the user’s video-teaching experience.

Data replication techniques are used in cloud computing to reduce access latency, network bandwidth and enhance data availability, system reliability. Replica selection involves selecting the best replica location to access the data for job execution in cloud computing. In order to select the best replica, a novel response time-driven replica selection approach based on Dirichlet probability distribution (DPRS) is proposed in this paper. Dirichlet PDF is the conjugate prior of categorical distribution to predict the posteriori value. The response time is calculated based on the network parameters such as network bandwidth, file size and access latency. The best replica can be predicted in corresponding with the historical log file by using Dirichlet PDF. Simulation results show that DPRS method conducts high performance in lower mean response time, while compared with No replica, LRU and LFU strategies.