����Ŀ¼

15 August 2018, Volume 4 Issue 8

Previous Issue    Next Issue

Comprehensive Review

Machine learning security and privacy:a survey

Lei SONG, Chunguang MA, Guanghan DUAN

2018, 4(8):  1-11.  doi:10.11959/j.issn.2096-109x.2018067

Asbtract ( 6094 )   HTML ( 280)   PDF (701KB) ( 10070 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

As an important method to implement artificial intelligence,machine learning technology is widely used in data mining,computer vision,natural language processing and other fields.With the development of machine learning,it brings amount of security and privacy issues which are getting more and more attention.Firstly,the adversary model was described according to machine learning.Secondly,the common security threats in machine learning was summarized,such as poisoning attacks,adversarial attacks,oracle attacks,and major defense methods such as regularization,adversarial training,and defense distillation.Then,privacy issues such were summarized as stealing training data,reverse attacks,and membership tests,as well as privacy protection technologies such as differential privacy and homomorphic encryption.Finally,the urgent problems and development direction were given in this field.

Papers

Defense decision-making method based on incomplete information stochastic game

Junnan YANG,Hongqi ZHANG,Chuanfu ZHANG

2018, 4(8):  12-20.  doi:10.11959/j.issn.2096-109x.2018065

Asbtract ( 775 )   HTML ( 10)   PDF (652KB) ( 2036 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Most of the stochastic game models to select network defense strategies are composed of matrix game and Markov decision,which assumes that the defender has known the attacker's revenue.This assumption does not conform to the actual situation.The uncertainty of the attacker's income was converted into the indeterminacy of the attacker type,and an incomplete information stochastic game model,which was combined with the static Bias game and Markov decision,was constructed.The equilibrium solution method of the incomplete information stochastic game model was given,and the strategy selection of the defender was guided by the stable Bias Nash equilibrium.Finally,a practical example was given to demonstrate the feasibility and effectiveness of the model.

Malware prediction technique based on program gene

Da XIAO,Bohan LIU,Baojiang CUI,Xiaochen WANG,Suoxing ZHANG

2018, 4(8):  21-30.  doi:10.11959/j.issn.2096-109x.2018069

Asbtract ( 1716 )   HTML ( 24)   PDF (816KB) ( 1834 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

With the development of Internet technology,malicious programs have risen explosively.In the face of executable files without source,the current mainstream malware detection uses feature detection based on similarity,with lack of analysis of malicious sources.To resolve this status,the definition of program gene was raised,a generic method of extracting program gene was designed,and a malicious program prediction method was proposed based on program gene.Utilizing machine learning and deep-learning algorithms,the forecasting system has good prediction ability,with the accuracy rate of 99.3% in the deep-learning model,which validates the role of program gene theory in the field of malicious program analysis.

Message encryption and authentication mechanism based on wireless physical layer key

Xingzhi,JINLiang LI,Zhou ZHONG,Yangming LOU

2018, 4(8):  31-38.  doi:10.11959/j.issn.2096-109x.2018055

Asbtract ( 1282 )   HTML ( 8)   PDF (883KB) ( 1505 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to solve the problem of hidden key leakage in traditional high-level message authentication and passive eavesdropping in physical layer message authentication,a message encryption and authentication mechanism based on the characteristics of wireless physical layer was proposed.By extracting the characteristics of the wireless channel,the physical layer key was generated and the seed key was generated by combining the identity key.The key stream generator then generated the key stream to decrypt the message according to the extracted seed,protects the privacy of the transmitted data,and the final receiver authenticated the authenticity and integrity of the received message through the CRC verification result.Simulation results show that the proposed method has better performance in preventing passive eavesdropping and active attack,and the randomness of key stream is greatly improved and the rate of authentication leak detection is reduced by half.

Research on key management scheme for military vehicle network

Junjie LIU,Jia ZHAO,Qiang ZHANG,Jiqiang LIU,Lei HAN

2018, 4(8):  39-46.  doi:10.11959/j.issn.2096-109x.2018068

Asbtract ( 863 )   HTML ( 0)   PDF (608KB) ( 1130 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

A scheme of military vehicle network security based on the hierarchical tree structure was proposed,which could improve communication efficiency and security of military vehicle network key management.The proposed scheme provides a good solution to solve the key management in military vehicle network communication.

5G network slicing function migration mechanism based on particle swarm optimization algorithm

Qiang CHEN,Caixia LIU,Lingshu LI

2018, 4(8):  47-55.  doi:10.11959/j.issn.2096-109x.2018066

Asbtract ( 1397 )   HTML ( 8)   PDF (641KB) ( 1210 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In multi-application scenarios of 5G,data traffic often increases dramatically.Virtual machine resources in network slicing may not meet the normal needs of users.In view of this,a network slicing function migration mechanism aiming at load balancing was proposed.The mechanism simulates the virtual machine into particles based on particle swarm optimization algorithm.In the process of migration,all particles were divided into several subgroups,and particle swarm optimization algorithm was applied within and among groups.According to the historical optimal solution and the current global optimal solution,the particle location was updated,and the best target particles were determined by selecting the smaller particle size of the particle in real time.The mechanism not only improves the convergence speed,but also improves the accuracy of the algorithm.Compared with other migration methods,the results show that the proposed migration mechanism has the advantages of high accuracy and fast convergence.And it can also improve the efficiency of resource utilization,reduce the energy consumption of data center,and has better adaptability.

Research on the implementation mechanism of multipath traffic proportion of IPv6 Overlay network

Shangming ZHU,Yun CHU,Yalan WANG

2018, 4(8):  56-62.  doi:10.11959/j.issn.2096-109x.2018045

Asbtract ( 965 )   HTML ( 3)   PDF (638KB) ( 796 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Based on IPv6 overlay network,two key technologies,multihoming and multipath routing were analyzed.According to logic functions,architectures of overlay nodes were presented.The mechanism of multipath traffic proportion of IPv6 overlay network was discussed on emphasis.A simplified scheme of traffic proportion according to path bandwidth was proposed,and the design and implementation of the scheme were carried out.Experimental results show that the efficiency of the algorithm of multipath traffic proportion is much higher than that of single path,and the advantage is more obvious with increasing data packets.

Data protection impact assessment system in the mode of risk management

Ying CHENG

2018, 4(8):  63-70.  doi:10.11959/j.issn.2096-109x.2018064

Asbtract ( 1940 )   HTML ( 35)   PDF (525KB) ( 3035 )   Knowledge map   

References | Related Articles | Metrics

In the era of big data,the risk management approach has been broadly applied in the field of personal information protection.Data protection impact assessment has become an important system to promote data protection.It takes the provisions of the data protection impact assessment of the European General Data Protection Regulation (GDPR) 2016 as the sample.By using the literature research and empirical analysis method,it analyzes in depth the theoretical background,rise and evolution,meaning and scope of data protection impact assessment to establish a standardized and specific impact assessment system as well as promote personal information protection.Assessment content includes not only privacy risk assessment,but also data security,data quality and non-discrimination.Data protection impact assessment should be set as a mandatory obligation for data processing activities that are likely to result in high risks.The evaluation process shall take the advices from stakeholders to reflect their benefits.The external supervision should be strengthened and the assessment report shall be published properly.

Identity preserving face completion with generative adversarial networks

Xudong WANG,Hongquan WEI,Chao GAO,Ruiyang HUANG

2018, 4(8):  71-76.  doi:10.11959/j.issn.2096-109x.2018070

Asbtract ( 1678 )   HTML ( 20)   PDF (954KB) ( 1554 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

As a special application of image completion technology,face image completion has an irreplaceable role in the occlusion of face recognition,portrait restoration and other issues.The existing face completion algorithm only aims at complementing the authenticity of the image without considering its identity consistency after completion.A face complement algorithm based on improved generative confrontation network was designed.By introducing SN-GAN algorithm,the stability of model training was improved.At the same time,the identity recognition constraint was added to the generated image using the face recognition model.Experiments have shown that the proposed method can effectively maintain the identity of the complementary image when generating high-authenticity images.