ISSN 1000-1239 CN 11-1777/TP

Table of Content

01 November 2021, Volume 58 Issue 11
Study of Wechat Sybil Detection
Yang Zheng, Yin Qilei, Li Haoran, Miao Yuanli, Yuan Dong, Wang Qian, Shen Chao, Li Qi
2021, 58(11):  2319-2332.  doi:10.7544/issn1000-1239.2021.20210461
Asbtract ( 1068 )   HTML ( 354)   PDF (1653KB) ( 767 )  
Related Articles | Metrics
Online social networks (OSNs) are efficient platforms for information dissemination and facilitate our daily life. The value of OSN accounts increases with the popularity of OSNs. In order to obtain profits illegally, attackers leverage OSNs to construct various attacks such as fraud and gambling. A number of solutions have been proposed to protect users’ security, which mainly focuses on detecting malicious accounts (or Sybils) by analyzing user behavior or the propagation of user relations. Unfortunately, it usually takes much time to collect enough data to perform malicious account detection. Attackers can perform different kinds of attacks during the data collection phase. To detect Sybils efficiently, we propose a new approach that leverages account registration attributes to detect Sybils. First, we analyze the existing detection methods in sybil detection. Then, we analyze the registration data of WeChat. We analyze and compare the distribution of Sybils and benign accounts in different registration attributes, and find that Sybils are prone to cluster with some registration attributes. According to these statistics, we extract two kinds of features from different attributes, i.e., synchronization-based features and anomaly-based features, and calculate the similarity of two accounts based on those features. The accounts that have high similarity are more likely to be malicious. Finally, we build a graph upon accounts having a high similarity to cluster malicious users. We calculate a malicious score for each user to infer whether it is a Sybil. We prototype our approach, and the experimental results with real WeChat show that our approach can achieve 96% precision and 60% recall.
Multi-Mode Attack Detection and Evaluation of Abnormal States for Industrial Control Network
Xu Lijuan, Wang Bailing, Yang Meihong, Zhao Dawei, Han Jideng
2021, 58(11):  2333-2349.  doi:10.7544/issn1000-1239.2021.20210598
Asbtract ( 390 )   HTML ( 9)   PDF (2714KB) ( 381 )  
Related Articles | Metrics
The ultimate intentions of various attack strategies leads the control system to a critical states or dangerous states for industrial control network. As a consequence, the attack detection method based on abnormal device status exceeds any other methods in terms of reliability. Oriented to the difficulty of accurately determining the ending of attack, this paper established the attack strategies model and the abnormal status description model, and then constructed corresponding datasets under a variety of attack strategies, proposed time slice partitioning algorithm based on inflection point fusion and state feature clustering algorithm, finally constructed an anomaly detection scheme based on state transition probability graph. Experimental results indicate that this scheme can effectively detect a variety of attack strategies. In addition, the research on the quantitative evaluation of semantic attack impacting on system states is relatively weaker than any other attack pattern, such as data injection attack, denial of service attack, and man-in the middle attack. In response to the above phenomenon, with results of anomaly detection as the cornerstone, this paper proposed the scheme of quantitative evaluation of attack impact on system states, according to the fusion analysis of abnormal features and threat degree indicators, for the state changes of the system at different stages. This work has important theoretical valuation and practical significance for identifying attack intention.
Stealthy Attack Towards Speaker Recognition Based on One-“Audio Pixel” Perturbation
Shen Yijie, Li Liangcheng, Liu Ziwei, Liu Tiantian, Luo Hao, Shen Ting, Lin Feng, Ren Kui
2021, 58(11):  2350-2363.  doi:10.7544/issn1000-1239.2021.20210632
Asbtract ( 320 )   HTML ( 2)   PDF (1560KB) ( 247 )  
Related Articles | Metrics
Attacks towards the speaker recognition system need to inject a long-time perturbation, so it is easy to be detected by machines or administrators. We propose a novel attack towards the speaker recognition based on one-“audio pixel”. Such attack uses the black-box characteristics and search mode of the differential evolution algorithm that does not rely on the model and the gradient information. It overcomes the problem in previous works that the disturbance duration cannot be constrained. Thus, our attack effectively spoofs the speaker recognition via one-“audio pixel” perturbation. In particular, we design a candidate point construction model based on the audio-point-disturbance tuple targeting time series of audio data. It solves the problem that candidate points of differential evolution algorithm are difficult to be described against our attack. The success rate of our attack achieves 100% targeting 60 people in LibriSpeech dataset. In addition, we also conduct abundant experiments to explore the impact of different conditions (e.g., gender, dataset and speaker recognition method) on the performance of our stealthy attack. The result of above experiments provides guidance for effective attacks. At the same time, we put forward ideas based on denoising, reconstruction algorithm and speech compression to defend against our stealthy attack, respectively.
Federated Learning Backdoor Attack Scheme Based on Generative Adversarial Network
Chen Dawei, Fu Anmin, Zhou Chunyi, Chen Zhenzhu
2021, 58(11):  2364-2373.  doi:10.7544/issn1000-1239.2021.20210659
Asbtract ( 823 )   HTML ( 16)   PDF (3164KB) ( 658 )  
Related Articles | Metrics
Federated learning enables users to participate in collaborative model training while keeping their data in local, which ensures the privacy and security of users’ data. It has been widely used in smart finance, smart medical and other fields. However, federated learning shows inherent vulnerability to backdoor attacks, where the attacker implants the backdoor by uploading the model parameters. Once the global model recognizes the input with the trigger, it will misclassify the input as the label specified by the attacker. This paper proposes a new federated learning backdoor attack scheme, Bac_GAN. By combining generative adversarial network, triggers are implanted in clean samples in the form of watermarks, which reduces the discrepancy between trigger features and clean sample features, and enhance the imperceptibility of triggers. By scaling the backdoor model, the problem of offsetting the contribution of the backdoor during parameter aggregation is avoided, so that the backdoor model can converge in a short time, thus significantly increasing the attack success rate. In addition, we conduct experimental tests on the core elements of backdoor attacks, such as trigger generation, watermark coefficient and scaling coefficient, and give the best parameters that affect the performance of backdoor attack. Also, we validate the attack effectiveness of the Bac_GAN scheme on MNIST and CIFAR-10.
Hybrid Feature Fingerprint-Based Wireless Device Identification
Song Yubo, Chen Bing, Zheng Tianyu, Chen Hongyuan, Chen Liquan, Hu Aiqun
2021, 58(11):  2374-2399.  doi:10.7544/issn1000-1239.2021.20210676
Asbtract ( 242 )   HTML ( 13)   PDF (6830KB) ( 296 )  
Related Articles | Metrics
Wireless networks transmit data over open wireless channels, so they are vulnerable to impersonation attacks and information forgery attacks. To prevent such attacks, accurate device identification is required. The device identification technology based on channel state information (CSI) fingerprinting uses the wireless channel characteristics of device for identification. Since CSI can provide fine-grained channel characteristics and can be easily obtained from OFDM wireless devices, this technology has received wide attention. However, since CSI fingerprints identify the wireless channel characteristics of device, they change with the location or the environment of device. What’s more, the existing technologies usually use machine learning for fingerprint matching for increasing identification accuracy, but the computational complexity of fingerprint matching increases, which in turn cannot be implemented in embedded devices with limited computational ability. To address these problems, this paper proposes a hybrid feature fingerprint-based device identification scheme, which includes the identification in access stage and communication stage. Packet arrival interval distribution (PAID) fingerprint, which is independent of device’s location, is introduced for identification in access stage to compensate for the shortcomings of the CSI fingerprint. In communication stage, CSI fingerprints are extracted from each data packet and identified in real time with the feature that CSI can be acquired packet by packet. In addition, this paper proposes a fingerprint matching scheme with low computational complexity to ensure fast and accurate device identification even in devices with limited computational ability. We implement the identification system on Raspberry Pi and perform some experiments, which show that the identification accuracy is up to 98.17% and 98.7% in access stage and communication stage, and the identification time of a single packet in communication stage is only 0.142ms.
A Feature Watermarking Generation and Embedding Scheme for IPv6 Network
Tao Jun, Zhu Zhenchao, Wang Zhaoyue, Li Wenqiang, Sun Weice
2021, 58(11):  2400-2415.  doi:10.7544/issn1000-1239.2021.20210654
Asbtract ( 257 )   HTML ( 7)   PDF (5134KB) ( 257 )  
Related Articles | Metrics
Under the limitation of space and time resources, researchers exploit the network covert channel, which based on a small amount of watermark information, to trace the attack flow and locate the real attack source. However, the self-similarity of the tracked traffic would appear because of the relatively fixed content and location of the watermark. What’s more, the IPSec encryption protocol embedded in the IPv6 protocol limits the range of carrier choice, which may threaten the watermarking based on the single carrier. In this paper, Targeting at optimizing the watermark invisibility, combined with intermediate node not dividing the packet for IPv6 environment, considering the feature extraction limitation of intermittent transmission network and slow flow network, the feature watermarking sequence extraction strategy associated with the target stream is designed. Aiming at different network transmission scenarios, a packet-dependent watermarking generation based on mixed covert channel and a time-dependent watermarking generation based on mixed time slot are proposed. Extensive experiments show that the watermarking generation technology proposed in this paper can reduce the impact of watermarking embedding on the original traffic, decrease the probability of watermarking being recognized and attack, and improve the imperceptibility of watermark under the premise of certain accuracy.
A Byzantine-Robust Federated Learning Algorithm Based on Matrix Mapping
Liu Biao, Zhang Fangjiao, Wang Wenxin, Xie Kang, Zhang Jianyi
2021, 58(11):  2416-2429.  doi:10.7544/issn1000-1239.2021.20210633
Asbtract ( 452 )   HTML ( 5)   PDF (1924KB) ( 502 )  
Related Articles | Metrics
Federated learning can better protect data privacy because the parameter server only collects the client model and does not touch the local data of the client. However, its basic aggregation algorithm FedAvg is vulnerable to Byzantine client attacks. In response to this problem, many studies have proposed different aggregation algorithms, but these aggregation algorithms have insufficient defensive capabilities, and the model assumptions do not fit the reality. Therefore, we propose a new type of Byzantine robust aggregation algorithm. Different from the existing aggregation algorithms, our algorithm focuses on detecting the probability distribution of the Softmax layer. Specifically, after collecting the client model, the parameter server obtains the Softmax layer probability distribution of the model through the generated matrix to map the updated part of the model, and eliminates the client model with abnormal distribution. The experimental results show that without reducing the accuracy of FedAvg, the Byzantine tolerance rate is increased from 40% to 45% in convergence prevention attacks, and the defense against edge-case backdoor attacks is realized in backdoor attacks. In addition, according to the current state-of-the-art adaptive attack framework, an adaptive attack is designed specifically for our algorithm, and experimental evaluations have been carried out. The experimental results show that our aggregation algorithm can defend at least 30% of Byzantine clients.
A Safe Storage and Release Method of Trajectory Data Satisfying Differential Privacy
Wu Wanqing, Zhao Yongxin, Wang Qiao, Di Chaofan
2021, 58(11):  2430-2443.  doi:10.7544/issn1000-1239.2021.20210589
Asbtract ( 341 )   HTML ( 14)   PDF (1854KB) ( 247 )  
Related Articles | Metrics
In recent years, although location-based service software facilitates people’s life, it brings the risk of privacy leakage. In order to solve this problem, we propose a trajectory data publishing method that is based on the noise prefix tree structure. In the first part, the trajectory equivalence class is constructed according to the space-time characteristics of the trajectory, and then the locus location points are divided by Hilbert curve to obtain the central points of the divided region. Finally, the obtained central points are converged into the new trajectory, so as to reduce the spatial complexity. The second part builds a prefix tree for storing location points according to the nature of the prefix tree, and stores the aggregated track location points into the prefix tree, which can improve query efficiency. In the third part, in order to protect the sensitive information stored in the nodes, this article will add Laplace noise to the nodes of the prefix tree, so that safer trajectory data can be released. Considering that the published data should be of high availability, this paper uses the arithmetic privacy budget allocation method to add Laplace noise to the node data, and limits the amount of noise by the threshold value of each layer, so as to finally publish trajectory data with high availability satisfying the differential privacy model. Through the experimental verification of real data sets, and comparing with the existing NTPT algorithm, our proposed TDPP algorithm is lower than the NTPT algorithm in different error values, and can provide better privacy protection. It is verified that the algorithm proposed in this paper improves data availability while ensuring data privacy.
End Spreading Multi-User Secure Communication System Based on SCMA
Shi Leyi, Lan Ru, Duan Pengfei, Han Qiang
2021, 58(11):  2444-2455.  doi:10.7544/issn1000-1239.2021.20210615
Asbtract ( 198 )   HTML ( 4)   PDF (1502KB) ( 73 )  
Related Articles | Metrics
End spreading technology uses a sequence of multiple end information to represent the identity information, and each piece of end information was irrelevant to the information it conveys. Thus it can hide the real information of the user. However, the end spreading sequence has the problems of low resource utilization, poor autocorrelation and unable to realize secure communication among multiple users. The sparse code multiple access (SCMA) technology is introduced into the generation process of end spreading sequence, establish the system model of end spreading multi-user secure communication system based on SCMA, and elaborating on the codebook design allocation strategy and code-word loading and sending strategy designed in the model. After that, the prototype system is analyzed and experimentally verified from two aspects of security capability and quality of communication service. The experiment result shows that, end spreading multi-user secure communication system based on SCMA realizes the covert transmission of user information, and the receiver can correctly distinguish user information. With sparse coding, the system has a low error rate and good transmission performance under certain overload conditions.
InterDroid: An Interpretable Android Malware Detection Method for Conceptual Drift
Zhang Bing, Wen Zheng, Wei Xiaoyu, Ren Jiadong
2021, 58(11):  2456-2474.  doi:10.7544/issn1000-1239.2021.20210560
Asbtract ( 399 )   HTML ( 8)   PDF (2803KB) ( 303 )  
Related Articles | Metrics
Aiming at the problems in Android malware detection, which are high subjectivity of feature definition, poor interpretability of feature selection process, and lack of temporal instability of training model detection accuracy, an interpretable Android malware detection method for concept drift called InterDroid is proposed. Firstly, four characteristics of the detection model: permission, API package name, intention and Dalvik bytecode are inferred through the high-quality artificial Android malware analysis report. And InterDroid training and comparison algorithm are obtained through automatic machine learning algorithm TPOT (tree-based tipeline optimization tool), thus abandoning the complicated process of model selection and parameter adjustment in traditional methods. After that, the traditional feature wrapper method is improved by integrating the model interpretation algorithm SHAP (shapley additive explanations), and the feature set with high contribution to the classification results is obtained for detection model training. Finally, the existence of concept drift in Android malware detection is proved by the double tests of MWU(Mann-Whitney U) and machine learning model. Based on the JDA(joint distribution adaptation), the accuracy of the detection model for Android malware in the new era is improved. The experimental results show that the feature screened by InterDroid is stable and interpretable. At the same time, the feature-representation transfer module in InterDroid can improve the detection accuracy of Android malware in 2019 and 2020 by 46% and 44%.
Document-Level Event Temporal Relation Extraction with Context Information
Wang Jun, Shi Cunhui, Zhang Jin, Yu Xiaoming, Liu Yue, Cheng Xueqi
2021, 58(11):  2475-2484.  doi:10.7544/issn1000-1239.2021.20200627
Asbtract ( 514 )   HTML ( 11)   PDF (2123KB) ( 370 )  
Related Articles | Metrics
Event temporal relation extraction is an important natural language understanding task, which can be widely used in downstream tasks such as construction of knowledge graph, question answering system and narrative generation. Existing event temporal relation extraction methods often treat the task as a sentence-level event pair classification problem, and solve it by some classification model. However, based on limited local sentence information, the accuracy of the extraction of temporal relations among events is low and the global consistency of the temporal relations cannot be guaranteed. For this problem, this paper proposes a document-level event temporal relation extraction with context information, which uses the neural network model based on Bi-LSTM (bidirectional long short-term memory) to learn the temporal relation expressions of event pairs, and then uses the self-attention mechanism to combine the information of other event pairs in the context, to obtain a better event temporal relation expression for temporal relation classification. At last, that event temporal relation expression with context information will improve the global event temporal relation extraction by enhancing temporal relation classification of all event pairs in the document. Experiments on TB-Dense (timebank dense) dataset and MATRES (multi-axis temporal relations for start-points) dataset show that this method can achieve better results than the latest sentence-level methods.
Space Transformation Based Random Forest Algorithm
Guan Xiaoqiang, Wang Wenjian, Pang Jifang, Meng Yinfeng
2021, 58(11):  2485-2499.  doi:10.7544/issn1000-1239.2021.20200523
Asbtract ( 294 )   HTML ( 5)   PDF (2916KB) ( 211 )  
Related Articles | Metrics
Random forest is a commonly used classification algorithm in the field of machine learning, which has the advantages of wide application and not easy overfitting. In order to improve the overall performance of random forest in dealing with multi-classification problems, a space transformation based random forest algorithm (ST-RF) is proposed. Firstly, a priority class based linear discriminant analysis (PCLDA) method is designed. On the basis of obtaining the projection matrix for priority class, the discrimination effect between priority class samples and other classes samples is enhanced by spatial transformation. Then, PCLDA method is introduced into the process of random forest construction. By selecting the priority class randomly for each decision tree, the diversity among decision trees in random forests is guaranteed. By using the PCLDA method to create decision trees with different priority classes, the classification accuracy of individual decision tree is improved. Thus, the overall classification performance of the integrated model is effectively improved. By comparing the ST-RF algorithm with seven typical random forest algorithms in 10 standard datasets, the effectiveness of the proposed algorithm is verified. Moreover, the spatial transformation strategy based on PCLDA is applied to the above comparison algorithms, and the performance of the algorithms before and after adding the spatial transformation strategy are compared and analyzed. The experimental results show that ST-RF algorithm has obvious advantages in dealing with multi-classification problems, and the proposed spatial transformation strategy has strong universality, which can significantly improve the classification performance of the original algorithm.
Closed High Utility Itemsets Mining over Data Stream Based on Sliding Window Model
Cheng Haodong, Han Meng, Zhang Ni, Li Xiaojuan, Wang Le
2021, 58(11):  2500-2514.  doi:10.7544/issn1000-1239.2021.20200554
Asbtract ( 216 )   HTML ( 3)   PDF (2927KB) ( 179 )  
Related Articles | Metrics
It is a challenging task to mine high utility itemsets from the data stream, because the incoming data stream must be processed in real time within the constraints of time and storage memory. Data stream mining usually generates a large number of redundant itemsets. In order to reduce the number of these useless itemsets and ensure lossless compression of complete high utility itemsets, it is necessary to mine closed itemsets, which can be several orders of magnitude smaller than the collection of complete high utility itemsets. In order to solve the above problem, a high utility itemsets mining algorithm (sliding-window-model-based closed high utility itemsets mining on data stream, CHUI_DS) is proposed to achieve mining closed high utility itemsets on data stream. A new utility-list structure is designed in CHUI_DS, which is very effective in increasing the speed of batch insertion and deletion. In addition, effective pruning strategies are applied to improve the closed itemset mining process and eliminate potential low-utility candidates. Extensive experimental evaluation of the proposed algorithm on real datasets and synthetic datasets shows the efficiency and feasibility of the algorithm. In terms of speed, it is superior to the previously proposed algorithms that mainly run in batch mode. Moreover, it is suitable for sliding windows of different sizes, and has strong scalability in terms of the number of transactions.
Computing Propositional Minimal Models: MiniSAT-Based Approaches
Zhang Li, Wang Yisong, Xie Zhongtao, Feng Renyan
2021, 58(11):  2515-2523.  doi:10.7544/issn1000-1239.2021.20200370
Asbtract ( 260 )   HTML ( 2)   PDF (1747KB) ( 166 )  
Related Articles | Metrics
Computing minimal models is an essential task in many reasoning systems in artificial intelligence. However, even for a positive conjunctive normal form (CNF) formula, the tasks of computing and checking its minimal model are not tractable. To compute a minimal model of a clause theory, one of the current main methods is to convert the clause theory into a disjunction logic program and use an answer set programming (ASP) solver to compute its stable model. This paper proposes a method MMSAT for computing minimal models based on SAT(satisfiability problem) solvers. In terms of the recently proposed minimal reduct based minimal checking algorithm CheckMinMR, a minimal model decomposing based minimal model computing algorithm MRSAT is proposed. Finally, the two algorithms are evaluated by a large number of randomly generated 3CNF formulas and industrial benchmarks from the SAT international competition. Experimental results show that the two methods proposed in this paper compute minimal models at about three times faster than clingo for random 3CNF formulas and slightly faster than clingo for industrial SAT benchmarks. Thus, they are effective. In addition, it also reveals that clingo sometimes incorrectly computes a minimal model. Thus, the two proposed methods in this paper are more stable than clingo.
Shared-Account Cross-Domain Sequential Recommendation with Self-Attention Network
Guo Lei, Li Qiuju, Liu Fang’ai, Wang Xinhua
2021, 58(11):  2524-2537.  doi:10.7544/issn1000-1239.2021.20200564
Asbtract ( 404 )   HTML ( 5)   PDF (969KB) ( 265 )  
Related Articles | Metrics
Shared-account cross-domain sequential recommendation (SCSR) is the task of recommending next items in a particular context, where users share a single account, and their behavior records are available in multiple domains. Compared with traditional sequential recommendation tasks, SCSR is challenging due to: 1) The interactions generated by an account is a mixture of multiple users. 2) The behaviors in one domain might be helpful to improve recommendations in another domain. Recently,most of the related work is based on recurrent neural network(RNN). Due to the inherent drawbacks of RNN, RNN-based methods are time consuming and more importantly they fail to capture long-range dependencies of accounts’ interactions. In this work, we target at SCSR and propose a self-attention-based cross-domain recommendation model(SCRM) to address these two challenges. Specifically, to model the mixed interactions from multiple users of a single account, a multi-head self-attention network is first introduced. Then, to leverage the domain information in one domain to improve the recommendation in another domain, the cross-domain transfer network based on a multi-layer cross-map perceptual network is innovatively proposed. Finally, a hybrid recommendation decoder is explored to consider the information from both domains to achieve recommendation in each domain. We conduct experiments on a real-world dataset HVIDEO, and the experimental results show that SCRM outperforms all the baseline methods in terms of MRR and Recall. In terms of training efficiency, SCRM achieves shorter training and learning time than RNN-based methods.
Survey of Text Stance Detection
Li Yang, Sun Yuqing, Jing Weipeng
2021, 58(11):  2538-2557.  doi:10.7544/issn1000-1239.2021.20200518
Asbtract ( 953 )   HTML ( 101)   PDF (2222KB) ( 614 )  
Related Articles | Metrics
Text stance detection is a basic study of text opinion mining, which aims to analyze the stance expressed in the text towards a specific target. Due to the rapid development of the Internet, the discussions of users for public events and consumer products are growing exponentially. The research of text stance detection is of great importance for product marketing and public opinion decision-making. This paper reviews the research of text stance detection from three angles: target type, text granularity and research method. First, from the perspective of target type, this paper focuses on three aspects: single-target stance detection, multi-target stance detection and cross-target stance detection; from the perspective of text granularity, the paper compares different application scenarios and methods of sentence level stance detection, document level stance detection and debate text stance detection; from the perspective of research methods, the paper introduces the traditional machine learning, topic model, deep learning and “two-stage” methods, and points out the advantages and disadvantages of various methods. Then, the evaluation tasks of text stance detection and the open data resources are summarized. Finally, based on the current research, the paper summarizes the application fields and looks forward to the future development trends and challenges of text stance detection.
Resource Deployment with Prediction and Task Scheduling Optimization in Edge Cloud Collaborative Computing
Su Mingfeng, Wang Guojun, Li Renfa
2021, 58(11):  2558-2570.  doi:10.7544/issn1000-1239.2021.20200621
Asbtract ( 572 )   HTML ( 13)   PDF (2334KB) ( 467 )  
Related Articles | Metrics
The cloud computing model of data centralized processing is facing new challenges for providing diversified application services with rapid interaction and green efficiency. In this paper, the cloud computing capability is extended to the edge devices, and an edge cloud collaborative computing framework is proposed. A resource deployment algorithm based on task prediction (RDTP) is designed. The tasks are predicted by two-dimensional time series in cloud service center, and the task resource deployment of edge server is optimized by classification aggregation and delay threshold determination. A task scheduling algorithm based on Pareto improvement (TSPI) is proposed. At the edge servers, the Pareto progressive comparison is conducted in two stages to obtain the tangent point or any intersection point of the two objective curves of quality of user service and effect of system service to optimize task scheduling. The experimental results show that combining the resource deployment algorithm based on task prediction and the task scheduling algorithm based on Pareto improvement (RDTP-TSPI) increases the average user task hit rate. In addition, in the application scenarios of varying user task scales and different Zipf distribution parameters α, the average service completion time of users, the overall service effectiveness of system, and the total task delay rate of RDTP-TSPI are better than the TSPI and BA (benchmark task scheduling algorithm based on FIFO).