To meet the real-time processing needs of compute-intensive big data applications,H-Storm heterogeneous computing platform TS developed based on Apache Storm.Through the Multi-process Service(MPS) feature,Graphic Process Unit(GPU) resource quantization and distributed calling mechanism are designed the task scheduling strategy of H-Storm heterogeneous clusters is proposed,and the task scheduling algorithm of GPU performance and load and adaptive flow distribution decision mechanism under cooperative computing are realized.Experimental results show that in the case of 512×512 matrix multiplication,the throughput of H-Storm heterogeneous computing platform increases by 54.9 times and the response delay decreases by 77 times compared with that of native Storm.
In the modern data center network,the partition aggregation transmission pattern easily leads to the throughput collapse on the bottleneck link,resulting in TCP Incast phenomenon.This paper proposes a random backoff method to reduce the concurrency degree of TCP flow burst transmission.By calculating the optimal random backoff time,the instantaneous congestion degree of bottleneck link is controlled and the TCP Incast problem is solved effectively.And it models and analyzes in theory the time-out probability and throughput when all TCP flows retreat randomly in the random backoff time interval.Experimental results show that the method can effectively avoid the transmission timeout event.The TCP flow concurrency and network throughput are increased by 2 times and 80 times respectively.
Tight security proofs need shorter security parameters and better efficiency.Therefor,a new Identity-based Signature(IBS) scheme named IDSSTR is proposed,which has a security specification for Computational Diffie-Hellman(CDH) problems and it is also naturally efficient on-line,no additional conditions is needed for the off-line stage and the verification process is unchanged.In order to shorten the total length of the signed message,a modified version of the IDSSTR with message recovery is goven.Analysis results show that,the difficulty of CDH problem is widely considered to be closely related to the discrete logarithm problem,therefor,the proposed signature scheme provides security assurance for such difficult problems.
For the presence of temporary,intermittent,and permanent errors,the processor capturing and executing an incorrect instruction will cause a control flow error to occur.This paper proposes a pure Error Detection of Software Signature(EDSS) algorithm based on table-driven with the study of Controlled Flow Detection by Software Signature(CFDSS) algorithm.It constructs two-dimensional table CFID to store the information of the control flow graph,and the illegal instruction jump is detected by comparing the signatures in the basic block with the signature stored in the CFID table.The EDSS algorithm can detect such errors successfully for the illegal branch jumps of the sharing-branch-fan-nodes that cannot be effectively detected by CFDSS algorithm.Experimental results show that the average error detection coverage of EDSS algorithm is 1.3% higher than that of CFDSS algorithm,and the average error detection rate of the sharing-branch-fan-nodes is about 1.9% higher than that of CFDSS algorithm.
In order to improve the classification accuracy of convolution neural network,an improved Stacking algorithm combining multiple convolution neural networks is proposed.The convolution neural network is used as the base classifier to classify the data,and the new sample is then classified by the meta-classifier.In order to reduce the dimension and correlation of the input data in meta-classifier,the dimension of output data of base classifier are reducted by Principal Component Analysis(PCA).Experimental results on classification accuracy of data sets show that,compared with traditional Stacking,average posteriori probability based algorithm and class voting based algorithm,the classification accuracy of the proposed algorithm is higher and more stable in similar networks and different types of networks.