Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2020-09-28 Cooperative journals: 《计算机应用研究》
Abstract： The infectious disease prevention information system has the problems that data is difficult to circulate and share between monitoring systems. To solve these problems, this paper based on Delegated Proof of Stake consensus algorithm combined with blockchain technology proposed an Infectious Disease Data Sharing model. By virtue of the characteristics of the blockchain, such as decentralization, non-tampering and collective maintenance, double-chain be used as the blockchain architecture, which improves the work efficiency. Combined with Inter-Planetary File System, large capacity storage space is obtained, which solves the space problem faced by block data storage and ensures the stability and sharing security of data storage. A consensus algorithm for disease prevention and control to realize the efficient operation is proposed by this paper. In addition, by comparing with other data sharing models, the advantages of this model in data storage and sharing security are highlighted.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-11-29 Cooperative journals: 《计算机应用研究》
Abstract： At present, the traditional method of cracking WinRAR is to use the CPU and GPU, but the potential password space is very large which requiries a higher performance computing platform to find the correct password within a limited time. Therefore, this paper uses the hardware platform of multi-FPGA to achieve a high-performance WinRAR crack algorithm. The SHA-1 algorithm was optimized by adding pre-calculation and carry saving adder under the full-pipeline architecture, so as to improve the throughput of the algorithm. And the use of state machine control to optimize data splicing, to improve algorithm parallelism. At the same time, it used the asynchronous clock and multiple FIFO buffers to read and write the overall structure of the data optimization algorithm to reduce the coupling within the algorithm. The experimental results show that the final optimized resource utilization rate is 75% and the frequency reaches 200MHz. The 4-byte password deciphering speed is 102, 796 per second, which is 100 times faster than CPU and 3.5 times faster than GPU.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-24 Cooperative journals: 《计算机应用研究》
Abstract： In order to reduce the influence of fingerprint data amount and AP number in the RSSI fingerprint database on the computational efficiency of the KNN algorithm, this paper proposed a Wi-Fi indoor positioning algorithm based on MD5-KNN, which optimized the fingerprint database of large place. In the offline phase, the algorithm converted each fingerprint into a MD5 sequence that included a 32-bit hexadecimal representation. In the online phase, the time it took for the algorithm to locate was linear with the number of APs and did not increase linearly with the number of fingerprints, which reduced the time and calculation amount required for positioning effectively. At the same time, this algorithm calculated the appropriate K adaptively, so as to effectively solve the problem that the RSSI-KNN algorithm needs to manually set the value of K. The experimental results show that the proposed algorithm can effectively improve the positioning accuracy and location efficiency of the indoor positioning technology based on Wi-Fi.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-24 Cooperative journals: 《计算机应用研究》
Abstract： In order to exploit the performance advantages of many core processors and solve more large-scale problems, this paper studied the parallelization of large integer multiplication on many core processors. Based on the parallel analysis of Written multiplication and Comba multiplication, this paper proposed a variety of solutions to solve the problem of load balancing when Comba multiplication is parallelized. Then according to the structural characteristics of SW26010, this paper selected the improved Comba multiplication based on Written multiplication. And the implementation process used some optimization methods such as vectorization, register communication and so on. The test results show that the improved Comba algorithm has better parallelism and can make good use of the performance advantages of the SW26010 many core processor.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-18 Cooperative journals: 《计算机应用研究》
Abstract： In the Multi-objective evolutionary algorithm, the solution of the rank distribution takes up most of the computation time. According to the Pareto dominance relationship, we research the solution of the rank and distribution in this paper, and we found that there are some redundancy operation of rank and allocation in the classic methods. The classic algorithms need to rank all solutions, and select the next generation after sort, so it caused some unnecessary sort operations. To reduce the redundancy, we introduce a fast method of non-dominated sorting. The proposed algorithm only deals the individual which belong the highest level in current population, and meanwhile it can choose individual into the next generation during the process of distribution level. When the next generation individuals were selected enough, the program is ended. This method has reduced the number of the sorting of individuals, and the time complexity is also reduced greatly. Then, we use three objectives optimization problem as an example, combine the fast non-dominated sorting method with the differential evolution algorithm, and propose a fast three objective differential evolution algorithm based on non-dominated sorting. Finally, we compare our method with the famous ranking method and ENS/BNS method. The simulation results show that our proposed algorithm in time complexity and convergence speed is better than the classical method, and is not worse than the ENS/BNS algorithm. On the performance of the standard test function DTLZ1-DTLZ6, the proposed method is similar to or better than the comparison algorithm. Based on above experiment, we have verified the correctness and effectiveness of our proposed method.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： In view of the current big data applications mainly use the general processor as the computing core, and the system structure is simple, energy efficiency ratio is low, can’t fully meet the big data computing needs. In this paper, based on mimic computing model, a design method of big data high-efficiency platform is put forward. This method took computing grain as the basic research object, deeply analyzed the features of big data application algorithms, and reasonably divided the computational subtasks. Secondly, an architecture matching matrix was constructed and the subtasks were assigned to the right processing units. Finally, dynamic voltage/frequency scaling technology and data layout algorithm were used to control the voltage of non-critical tasks and optimize the structure layout of critical tasks. The experimental results show that the mimic computing can integrate the heterogeneous computing components in depth, establish a flexible and scalable architecture, give full play to the overall efficiency of the system, reduce the power consumption and improve the energy efficiency ratio.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-19 Cooperative journals: 《计算机应用研究》
Abstract： Micro-blog sentiment analysis aims to study the emotional views of users on hot events, and research shows that deep learning is feasible in micro-blog's sentiment analysis. In view of traditional convolutional neural network, micro-blog sentiment analysis ignores the correlation between discontinuous words. Therefore, this paper applied attention mechanism to the input end of convolutional neural network (CNN) model to improve this problem. Because Chinese micro-blog belongs to the short text category, there was a possibility of losing too many semantic features in the selection of pooling layer features in the process of convolutional neural network forward propagation, so into the long short term memory neural network tree at the output of the convolutional neural network terminal (LSTM) , by adding the sentence structure to strengthen the deep semantic learning. Based on the two improvements, it constructed a Chinees micro-blog sentiment analysis model (Att-CTL) . Experiments show that the model has excellent characteristics in Chinese micro-blog sentiment analysis, especially in polarity shifting, and maintains a high F1 value.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-17 Cooperative journals: 《计算机应用研究》
Abstract： In order to solve the problem of multiplexing gain loss in Alamouti space-time block coding, this paper proposed a two-groups Alamouti transmission scheme. On this basis, in order to improve the bit error ratio (BER) with lower computational complexity at the receiver, this paper proposed a two-groups Alamouti transmission scheme by combining Geometric Mean Decomposition (GMD) algorithm and non-linear precoding technique. The design method of the proposed solution consists of three main operations. Firstly, obtain the effective channel matrix of two-group Alamouti transmission systems. Secondly, jointly design the transceiver with the obtained effective channel matrix by exploiting the geometric mean decomposition (GMD) algorithm. Finally, mitigate the inter-channel interference by utilizing non-linear precoding techniques of Dirty Paper Coding (DPC) and Tomlinson-Harashima Precoding (THP) . The simulation results demonstrate that the proposed scheme is capable of significantly improving the achievable BER performance.
Hits
Hits
Downloads
Comment