In the practical application scenario, the wheat kernel image background is single, and the imperfect characteristics of wheat imperfect grains are mostly local features while most of the image features are not different from normal grains. In order to solve the problems, an imperfect wheat kernel recognition method based on detail Image Enhancement (IE) was proposed. Firstly, the alternate minimization algorithm was used to constrain the L0 norms of the original image in the horizontal and vertical directions to smooth the original image as the base layer, and the original image was subtracted from the base layer to obtain the detail layer of the image. Then, the detail layer was delighted and superimposed with the base layer to enhance the image. Finally, the enhanced image was used as the input of the Convolutional Neural Network (CNN), and the CNN with Batch Normalization (BN) layer was used for recognition of the image. The classic classification networks LeNet-5, ResNet-34, VGG-16 and these networks with the BN layer were used as classification networks, and the images before and after enhancement were used as input to carry out classification experiments, and the accuracy of the test set was used to evaluate the performance. Experimental results show that by adding the BN layer and using the same input, all three classic classification networks have the accuracy of the test set increased by 5 percentage points, and when using the images with enhanced detail as input, the three networks have the accuracy of the test set increased by 1 percentage point, and when the above two are used together, all the three networks obtain the accuracy of the test set improved by more than 7 percentage points.

Aiming at high energy consumption of data center in container cloud, a virtual resource deployment strategy based on host selection algorithm with Power Full (PF) was proposed. Firstly, the allocation and migration scheme of virtual resource in container cloud was proposed and the significant impact of host selection strategy on energy consumption of data center was found. Secondly, by studying the mathematical relationship between the utilization of host and the utilization of containers, between the utilization of host and the utilization of virtual machines and between the utilization of host and energy consumption of data center, a mathematical model of the energy consumption of data center in container cloud was constructed and an optimization objective function was defined. Finally, the function of host's energy consumption was simulated using linear interpolation method, and a host selection algorithm with PF was proposed according to the clustering property of the objects. Simulation results show that compared with First Fit (FF), Least Full (LF) and Most Full (MF), the energy consumption of the proposed algorithm is averagely reduced by 45%,53% and 49% respectively in the computing service of regular tasks and different host clusters; is averagely reduced by 56%,46% and 58% respectively in the computing service of regular tasks and same host cluster; is averagely reduced by 32%,24% and 12% respectively in the computing service of irregular tasks and different host clusters. The results indicate that the proposed algorithm realizes reasonable virtual resource deployment in container cloud, and has advantage in data center energy saving.

The leaf switch failure would seriously affect the use of high performance computer system with K-Ary N-Bridge topology. In order to improve the usability and maintainability of that topology, a routing fault-tolerant strategy based on misrouting algorithm was proposed. The basic idea was to bypass the failed leaf switch leveraging misrouting, jump to other leaf switches in the same dimension, and then reached the destination node through the normal route. The proposed fault-tolerant strategy could shield the failed leaf switch without affecting the system usage. A fault-tolerant experiment was carried out in a practical K-Ary N-Bridge topology. The result shows that this fault-tolerant strategy can quickly shield the failed leaf switch as expected and can effectively improve the efficiency of system maintenance.

Concerning the performance of the Orthogonal Space Time Block Code based Multiple-Input Multiple-Output (OSTBC-MIMO) system over independent but not necessarily identically distributed (i.n.i.d.) generalized-K fading channels, the system model of the OSTBC-MIMO system was established in i.n.i.d. generalized-K fading channel by adopting M-QAM modulation scheme. The equivalent Signal-to-Noise Ratio (SNR) of the receiver was approximated by the product of two variables that composed of multiple independent Gamma distributed random variables. On the basis of that, the probability density function expression of the equivalent SNR was derived as well as the expression of the average symbol error probability, channel capacity and outage probability. The simulation results show that, in addition to the parameter m, the parameter k also has a significant impact on the overall system performance, which produces the non-negligible differences on the system performance between the i.n.i.d. generalized-K fading channels and the i.i.d. generalized-K fading channels.

Since most malwares are designed using decentralized architecture to resist detection and countering, in order to fast and accurately detect Peer-to-Peer (P2P) bots at the stealthy stage and minimize their destructiveness, a real-time detection system for stealthy P2P bots based on statistical features was proposed. Firstly, all the P2P hosts inside a monitored network were detected using means of machine learning algorithm based on three P2P statistical features. Secondly, P2P bots were discriminated based on two P2P bots statistical features. The experimental results show that the proposed system is able to detect stealthy P2P bots with an accuracy of 99.7% and a false alarm rate below 0.3% within 5 minutes. Compared to the existing detection methods, this system requires less statistical characteristics and smaller time window, and has the ability of real-time detection.