Xiao Jiali, Nie Gaofeng, Deng Gang, Tian Hui, Zhang Chong
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1001
摘要 ( 253 ) PDF (1828 KB)( 58 )
Sparse code multiple access-based uplink grant-free transmission (SCMA-UGFT) has been proposed to realize ultra reliable and low latency communication (URLLC) in the fifth generation (5G) system. Without the process of resource request and grant, users may collide in the same resource. To compensate the potential user performance decline, resource scheduling becomes a tough issue in the SCMA-UGFT system. This article proposes a duplicated transmission-based resource scheduling (DTBRS) scheme for SCMA-UGFT system by considering the URLLC scenario. Different from the existing schemes, not only one shared basic transmission units (BTUs) are allocated to a user equipment ( UE) in the proposed DTBRS scheme for initial transmission to realize the duplicated
transmission and to guarantee the transmission reliability. Besides, according to the proposed DTBRS scheme, one or two exclusive BTUs are assigned to a UE for retransmission to avoid the re-collision. At last, each packet is given a lifetime to limit the transmission latency to meet the URLLC latency requirement. The simulation demonstrates that the DTBRS scheme can achieve a better performance than the existing state-of-the-art scheme in terms of the average packet drop rate.
Hou Huanhuan, Jiang Jing, Lei Ming, Liu Ben
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1002
摘要 ( 237 ) PDF (1076 KB)( 93 )
Ultra-dense network (UDN) deployment of small cells introduces novel technical challenges, one of which is that the interference levels increase considerably with the network density. This paper proposes interference suppression scheme based on compressive sensing (CS) framework for UDN. Firstly, the measurement matrix is designed by exploiting the sparsity of millimeter wave channels. CS technique is employed to transform the high dimension sparse signal into low dimension signal. Then, the interference is canceled in the compressed domain. Finally, the stagewise weak orthogonal matching pursuit (SWOMP) algorithm is used to reconstruct the useful signal after interference suppression. The analysis and simulation results demonstrate the effectiveness of the algorithm. Simulation results demonstrate that the proposed interference suppression in compressive domain yields performance gains compared to other classical interference suppression schemes. The proposed algorithm can reduce the computational complexity of interference suppression algorithm.
Pan Ziyu, Hu Han
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1003
摘要 ( 251 ) PDF (723 KB)( 119 )
In the research of green communication, considering the base station (BS) power allocation from the perspective of energy efficiency (EE) is meaningful for heterogeneous cellular networks (HCNs) optimization. The EE of two-tier HCNs was analyzed and a new method for the network EE optimization was proposed by adjusting the small BS transmitting power. First, the HCNs ware modeled by homogeneous Poisson point processes (PPPs), and the coverage probability of BSs in each tier was derived. Second, according to the definition of EE, and the closed-form of EE was given by deriving the total power consumption and total throughput of HCNs respectively. At last, the analytical performance of the EE of HCNs on the small BS transmission power was analyzed, and a small BS power optimization algorithm was proposed to maximize the EE. Simulation results show that, the transmission power of small BS has a significant impact on the EE of HCNs. Furthermore, by optimizing the transmission power of small BS, the EE of HCNs can be improved effectively.
Zhao Guosheng, Liao Yuting, Wang Jian
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1004
摘要 ( 283 ) PDF (601 KB)( 120 )
In the mobile crowd sensing (MCS) network environment, it is very important to establish an evolutionary process that can dynamically depict the trust degree of task participants. To address this issue, this paper proposes a dynamic trust evaluation model for task participants. Firstly, according to the security requirements and trust strategy of the perceived tasks, the attribute reduction algorithm (ARA) based on rough set is used to obtain the multi-attribute indexes that affect the participants' trust information. Removing the redundant attributes can avoid the lag of trust evaluation and reduce the time cost. Secondly, the grey correlation analysis method is used to solve the correlation degree between the target sequence and the comparison sequence on the trust attributes by integrating the multi-attribute decision-making method, which avoids the distortion of the trust evaluation caused by human subjective factors and improves the quality of the perceived data. Finally, a dynamic trust evaluation model for participants in complex sensing network environment is established. The simulation results show that the proposed model can not only dynamically depict the trust degree of participants in real time, but also have higher accuracy and less time cost.
Guo Qinghao, Shuai Liguo, Hu Sunying
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1005
摘要 ( 254 ) PDF (3255 KB)( 52 )
The training efficiency and test accuracy are important factors in judging the scalability of distributed deep learning. In this dissertation, the impact of noise introduced in the mixed national institute of standards and technology database (MNIST) and CIFAR-10 datasets is explored, which are selected as benchmark in distributed deep learning. The noise in the training set is manually divided into cross-noise and random noise, and each type of noise has a different ratio in the dataset. Under the premise of minimizing the influence of parameter interactions in distributed deep learning, we choose a compressed model (SqueezeNet) based on the proposed flexible communication method. It is used to reduce the communication frequency and we evaluate the influence of noise on distributed deep training in the synchronous and asynchronous stochastic gradient descent algorithms. Focusing on the experimental platform TensorFlowOnSpark, we obtain the training accuracy rate at different noise ratios and the training time for different numbers of nodes. The existence of cross-noise in the training set not only decreases the test accuracy and increases the time for distributed training. The noise has positive effect on destroying the scalability of distributed deep learning.
Tang Xianlun, Chen Yingjie, Xu Jin, Yu Xinxian
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1006
摘要 ( 321 ) PDF (2725 KB)( 59 )
Text classification is a classic task innatural language process (NLP). Convolutional neural networks (CNNs) have demonstrated its effectiveness in sentence and document modeling. However, most of existing CNN models are applied to the fixed-size convolution filters, thereby unable to adapt different local interdependency. To address this problem, a deep global-attention based convolutional network with dense connections (DGA-CCN) is proposed. In the framework, dense connections are applied to connect each convolution layer to each of the other layers which can accept information from all previous layers and get multiple sizes of local information. Then the local information extracted by the convolution layer is reweighted by deep global-attention to obtain a sequence representation with more valuable information of the whole sequence. A series of experiments are conducted on five text classification benchmarks, and the experimental results show that the proposed model improves upon the state of-the-art baselines on four of five datasets, which can show the effectiveness of our model for text classification.
Yuan Ye, Yu Minmin, Liu Jiming
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1007
In order to improve the accuracy of text similarity calculation, this paper presents a text similarity function part of speech and word order-smooth inverse frequency (PO-SIF) based on sentence vector, which optimizes the classical SIF calculation method in two aspects: part of speech and word order. The classical SIF algorithm is to calculate sentence similarity by getting a sentence vector through weighting and reducing noise. However, the different methods of weighting or reducing noise would affect the efficiency and the accuracy of similarity calculation. In our proposed PO-SIF, the weight parameters of the SIF sentence vector are first updated by the part of speech subtraction factor, to determine the most crucial words. Furthermore, PO-SIF calculates the sentence vector similarity taking into the account of word order, which overcomes the drawback of similarity analysis that is mostly based on the word frequency. The experimental results validate the performance of our proposed PO-SIF on improving the accuracy of text similarity calculation.
Deng Junyong, Xie Xiaoyan, Liu Yang, Tian Pu
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1008
Primitive assembly is an inevitable procedure of graphics rendering which performs the objects preparation for the following steps, however, the conventional approaches suffer from some issues, such as the missing of surface attribute, mismatch of color mode for clipped primitives, and performance bottleneck of rendering pipeline. This paper takes all these issues into considerations, and proposes a parallel primitive assembly accelerator (PPAA) which can solve not only the functional problems but also improve the shading performance. The register transfer level (RTL) circuit is designed and the detailed approach is presented. The prototype systems are implemented on Xilinx field programmable gate array (FPGA) XC6VLX550T and Altera FPGA EP2C70F896C6. The experimental results show that PPAA can accomplish the assembly tasks correctly and with higher performance of 1.5x and 2.5x of two previous implementations. For the most frequently independent primitives, the PPAA can efficiently enhance the throughput by squeezing out the pipeline bubbles and by balancing the pipeline stages.
Lin Qiaomin, Fa Anqi, Pan Min, Xie Qiang, Du Kun, Sheng Michael
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1009
摘要 ( 242 ) PDF (534 KB)( 139 )
Currentlyradio frequency identification (RFID) technology has been widely used in many kinds of applications. Store retailers use RFID readers with multiple antennas to monitor all tagged items. However, because of the interference from environment and limitations of the radio frequency technology, RFID tags are identified by more than one RFID antenna, leading to the false positive readings. To address this issue, we propose a RFID data stream cleaning method based on K-means to remove those false positive readings within sampling time. First, we formulate a new data stream model which adapts to our cleaning algorithm. Then we present the preprocessing method of the data stream model, including sliding window setting, feature extraction of data stream and normalization. Next, we introduce a novel way using K-means clustering algorithm to clean false positive readings. Last, the effectiveness and efficiency of the proposed method are verified by experiments. It achieves a good balance between performance and price.
Ma Yan, Du Donggao
中国邮电高校学报(英文版), 2020, 27 (2). doi： 10.19682/j.cnki.1005-8885.2020.1010
摘要 ( 248 ) PDF (1200 KB)( 65 )
Application programming interface (API) is a procedure call interface to operation system resource. API-based behavior features can capture the malicious behaviors of malware variants. However, existing malware detection approaches have a deal of complex operations on constructing and matching. Furthermore, graph matching is adopted in many approaches, which is a nondeterministic polynominal (NP)-complete problem because of computational complexity. To address these problems, a novel approach is proposed to detect malware variants. Firstly, the API of the malware are divided by their functions and parameters. Then, the classified behavior graph (CBG) is constructed from the API call sequences. Finally, the signature based on CBGs for each malware family is generated. Besides, the malware variants are classified by ensemble learning algorithm. Experiments on 1 220 malware samples show that the true positive rate (TPR) is up to 89.0% with the low false positive rate (FPR) 3.7% by ensemble learning.