Email alert
2023 Vol. 45, No. 3
Display Method:
2023, 45(3): 765-774.
doi: 10.11999/JEIT221047
Abstract:
Compared with the ε-Local Differential Privacy (LDP) mechanism, the scheme under (ε, δ)-local differential privacy has a smaller error bound and higher data utility. However, the current mean estimation mechanisms under (ε, δ)-local differential privacy still have problems such as large estimation error and low data utility. Therefore, for the mean estimation problem, two new mean estimation mechanisms under (ε, δ)-local differential privacy are proposed: the Interval-based Mechanism for mean estimation (IM) and the Neighbor-based Mechanism for mean estimation (NM). IM divides the perturbed data into three intervals. Then the real data is perturbed to the middle interval with a large probability, and the two sides are perturbed with a small probability. Collector averages directly the perturbed data to get an unbiased estimation. NM perturbs the real data to its near neighborhood with a large probability and perturbs it far away with a small probability. Then the collector applies the expectation maximization algorithm to obtain an estimated mean value with high accuracy. Finally, both IM and NM satisfy the privacy protection requirements are proved through theoretical analysis, and the data utility of IM and NM is superior to existing mechanisms is confirmed by experiments.
Compared with the ε-Local Differential Privacy (LDP) mechanism, the scheme under (ε, δ)-local differential privacy has a smaller error bound and higher data utility. However, the current mean estimation mechanisms under (ε, δ)-local differential privacy still have problems such as large estimation error and low data utility. Therefore, for the mean estimation problem, two new mean estimation mechanisms under (ε, δ)-local differential privacy are proposed: the Interval-based Mechanism for mean estimation (IM) and the Neighbor-based Mechanism for mean estimation (NM). IM divides the perturbed data into three intervals. Then the real data is perturbed to the middle interval with a large probability, and the two sides are perturbed with a small probability. Collector averages directly the perturbed data to get an unbiased estimation. NM perturbs the real data to its near neighborhood with a large probability and perturbs it far away with a small probability. Then the collector applies the expectation maximization algorithm to obtain an estimated mean value with high accuracy. Finally, both IM and NM satisfy the privacy protection requirements are proved through theoretical analysis, and the data utility of IM and NM is superior to existing mechanisms is confirmed by experiments.
2023, 45(3): 775-783.
doi: 10.11999/JEIT221104
Abstract:
Considering the problems of inefficiency and privacy leakage faced by the blockchain-based federated learning data sharing platform under massive data, an efficient model collaborative training and sharing scheme of blockchain based on hybrid privacy is proposed. In this scheme, a similarity-based training member selection algorithm according to Euclidean distance is first designed to select training members, forming a federated community, that is, to improve the efficiency and effect of training by selecting a small number of high-matching training nodes. Then, combined with threshold homomorphic encryption and differential privacy, a model collaborative training and sharing scheme based on hybrid privacy technology is constructed to ensure the privacy in the training and sharing process. The experimental results and system implementation show that the proposed scheme can achieve efficient training and data sharing under privacy protection while ensuring the accuracy of the training results.
Considering the problems of inefficiency and privacy leakage faced by the blockchain-based federated learning data sharing platform under massive data, an efficient model collaborative training and sharing scheme of blockchain based on hybrid privacy is proposed. In this scheme, a similarity-based training member selection algorithm according to Euclidean distance is first designed to select training members, forming a federated community, that is, to improve the efficiency and effect of training by selecting a small number of high-matching training nodes. Then, combined with threshold homomorphic encryption and differential privacy, a model collaborative training and sharing scheme based on hybrid privacy technology is constructed to ensure the privacy in the training and sharing process. The experimental results and system implementation show that the proposed scheme can achieve efficient training and data sharing under privacy protection while ensuring the accuracy of the training results.
2023, 45(3): 784-792.
doi: 10.11999/JEIT221064
Abstract:
Federated Learning and swarm Learning, as currently popular distributed machine learning paradigms, the former enables shared computation of model parameters in servers while protecting user data from third parties, while the latter uses blockchain technology to aggregate model parameters equally for all users without a central server. However, by analyzing the parameters after model training, such as the weights of deep neural network training, it is still possible to leak the user's private information. At present, there are several methods for protecting model parameters utilizing Local Differential Privacy (LDP) in federated learning, however it is challenging to reduce the gap in model testing accuracy when there is a limited privacy budget and user base. To solve this problem, a Positive and Negative Piecewise Mechanism (PNPM) is proposed, which perturbs the local model parameters before aggregation. First, it is proved that the mechanism satisfies the strict definition of differential privacy and ensures the privacy of the algorithm; Secondly, it is analyzed that the mechanism can ensure the accuracy of the model under a small number of users and ensure the effectiveness of the mechanism; Finally, it is compared with other state-of-the-art methods in terms of model accuracy and privacy protection on three mainstream image classification datasets and shows a better performance.
Federated Learning and swarm Learning, as currently popular distributed machine learning paradigms, the former enables shared computation of model parameters in servers while protecting user data from third parties, while the latter uses blockchain technology to aggregate model parameters equally for all users without a central server. However, by analyzing the parameters after model training, such as the weights of deep neural network training, it is still possible to leak the user's private information. At present, there are several methods for protecting model parameters utilizing Local Differential Privacy (LDP) in federated learning, however it is challenging to reduce the gap in model testing accuracy when there is a limited privacy budget and user base. To solve this problem, a Positive and Negative Piecewise Mechanism (PNPM) is proposed, which perturbs the local model parameters before aggregation. First, it is proved that the mechanism satisfies the strict definition of differential privacy and ensures the privacy of the algorithm; Secondly, it is analyzed that the mechanism can ensure the accuracy of the model under a small number of users and ensure the effectiveness of the mechanism; Finally, it is compared with other state-of-the-art methods in terms of model accuracy and privacy protection on three mainstream image classification datasets and shows a better performance.
2023, 45(3): 793-799.
doi: 10.11999/JEIT221092
Abstract:
The ability of Eight-Sided Fortress (ESF) algorithm to resist impossible differential cryptanalysis and linear cryptanalysis is studied in this paper. The ESF algorithm is a lightweight block cipher algorithm with Feistel structure, and its round function is Substitution-Permutation(SP) structure. Firstly, 12 rounds of ESF algorithm is analyzed in this paper by a new impossible differential distinguisher, and then 9 rounds of ESF algorithm is analyzed by linear cryptanalysis. It is calculated that the data complexity of 12 rounds of impossible differential analysis is about O(267), and the time complexity is about O(2110.7), while the data complexity of 9 rounds of linear cryptanalysis is only O(235), and the time complexity is no more than O(215.6). The results show that ESF algorithm is able to resist impossible differential cryptanalysis, while its ability to resist linear cryptanalysis is relatively weak.
The ability of Eight-Sided Fortress (ESF) algorithm to resist impossible differential cryptanalysis and linear cryptanalysis is studied in this paper. The ESF algorithm is a lightweight block cipher algorithm with Feistel structure, and its round function is Substitution-Permutation(SP) structure. Firstly, 12 rounds of ESF algorithm is analyzed in this paper by a new impossible differential distinguisher, and then 9 rounds of ESF algorithm is analyzed by linear cryptanalysis. It is calculated that the data complexity of 12 rounds of impossible differential analysis is about O(267), and the time complexity is about O(2110.7), while the data complexity of 9 rounds of linear cryptanalysis is only O(235), and the time complexity is no more than O(215.6). The results show that ESF algorithm is able to resist impossible differential cryptanalysis, while its ability to resist linear cryptanalysis is relatively weak.
2023, 45(3): 800-809.
doi: 10.11999/JEIT220602
Abstract:
In view of the fact that digital watermarking technology does not consider the preview needs of users in copyright image sharing scenarios, as well as the limitations of software control method and additional information method, a copyright image sharing scheme that supports previewing part of the visual content of the original image directly from the encrypted image is proposed. To this end, the idea of thumbnail preserving encryption is introduced into the user-side embedded watermarking scheme, and a blurred version of the original image content is presented on the encrypted image through pixel adjustment. The pixel bits to be adjusted are embedded into the hidden area by means of information hiding in advance to ensure the correctness of decryption. Moreover, the user watermark is embedded into the image at the same time of decryption, which is used to track the unauthorized redistribution behavior. In this way, not only the preview needs of the user during the sharing process is met, but also the copyright of the image is protected. The results of theoretical analysis and experimental tests demonstrate the security, feasibility, efficiency, and robustness of the proposed scheme.
In view of the fact that digital watermarking technology does not consider the preview needs of users in copyright image sharing scenarios, as well as the limitations of software control method and additional information method, a copyright image sharing scheme that supports previewing part of the visual content of the original image directly from the encrypted image is proposed. To this end, the idea of thumbnail preserving encryption is introduced into the user-side embedded watermarking scheme, and a blurred version of the original image content is presented on the encrypted image through pixel adjustment. The pixel bits to be adjusted are embedded into the hidden area by means of information hiding in advance to ensure the correctness of decryption. Moreover, the user watermark is embedded into the image at the same time of decryption, which is used to track the unauthorized redistribution behavior. In this way, not only the preview needs of the user during the sharing process is met, but also the copyright of the image is protected. The results of theoretical analysis and experimental tests demonstrate the security, feasibility, efficiency, and robustness of the proposed scheme.
2023, 45(3): 810-818.
doi: 10.11999/JEIT220957
Abstract:
In order to address these problems of coarse-grained access control, single point of service failure and low user decryption efficiency in existing smart home firmware update schemes, a fine-grained remote data security update scheme for smart home with privacy protection is proposed. The scheme realizes fine-grained access control through attribute-based encryption technology, and combines blockchain and Inter Planetary File System (IPFS) technology to store data. This scheme protects further user’s privacy by hiding access policies. And the Ciphertext Policy Attribute-Based Encryption (CP-ABE) is proposed. In addition, the outsourcing decryption algorithm for lightweight users is designed to reduce the computing burden of lightweight users effectively, and the fair payment in the outsourcing decryption process is realized by combining blockchain and smart contract technology. Finally, based on Decisional Bilinear Diffie-Hellman (DBDH) assumption, the proposed scheme is proved to be INDistinguishability under Chosen-Plaintext Attack (IND-CPA) security. The experimental results show that the proposed scheme reduces significantly the cost of terminal user decryption compared and communication overhead with existing schemes.
In order to address these problems of coarse-grained access control, single point of service failure and low user decryption efficiency in existing smart home firmware update schemes, a fine-grained remote data security update scheme for smart home with privacy protection is proposed. The scheme realizes fine-grained access control through attribute-based encryption technology, and combines blockchain and Inter Planetary File System (IPFS) technology to store data. This scheme protects further user’s privacy by hiding access policies. And the Ciphertext Policy Attribute-Based Encryption (CP-ABE) is proposed. In addition, the outsourcing decryption algorithm for lightweight users is designed to reduce the computing burden of lightweight users effectively, and the fair payment in the outsourcing decryption process is realized by combining blockchain and smart contract technology. Finally, based on Decisional Bilinear Diffie-Hellman (DBDH) assumption, the proposed scheme is proved to be INDistinguishability under Chosen-Plaintext Attack (IND-CPA) security. The experimental results show that the proposed scheme reduces significantly the cost of terminal user decryption compared and communication overhead with existing schemes.
2023, 45(3): 819-827.
doi: 10.11999/JEIT220618
Abstract:
As the basic component of smart grid, Smart Meter System (SMS) can regularly report the detailed electricity consumption data of users to power companies. However, SMS also bring some security problems, such as user privacy disclosure. This paper proposes a privacy protection scheme based on virtual ring for SMS based on fog computing. This scheme can provide the privacy of power consumption data and user identity, so that the attacker can not know the relationship between matching power data and user identity. In the proposed scheme, the SMS can use its virtual ring membership to anonymize its real identity, and it can also use asymmetric encryption and Paillier homomorphic system to generate ciphertext data from its power consumption data; Then the SMS sends the ciphertext data to the connected fog node, and the fog node collects regularly the ciphertext data of the SMS it manages. At the same time, the fog node verifies the virtual ring identity of these SMS, and then aggregates the collected ciphertext data and sends it to the control center; Finally, the control center decrypts the aggregated ciphertext to obtain the power consumption data. The experimental results show that the proposed scheme has some advantages in computing and communication costs.
As the basic component of smart grid, Smart Meter System (SMS) can regularly report the detailed electricity consumption data of users to power companies. However, SMS also bring some security problems, such as user privacy disclosure. This paper proposes a privacy protection scheme based on virtual ring for SMS based on fog computing. This scheme can provide the privacy of power consumption data and user identity, so that the attacker can not know the relationship between matching power data and user identity. In the proposed scheme, the SMS can use its virtual ring membership to anonymize its real identity, and it can also use asymmetric encryption and Paillier homomorphic system to generate ciphertext data from its power consumption data; Then the SMS sends the ciphertext data to the connected fog node, and the fog node collects regularly the ciphertext data of the SMS it manages. At the same time, the fog node verifies the virtual ring identity of these SMS, and then aggregates the collected ciphertext data and sends it to the control center; Finally, the control center decrypts the aggregated ciphertext to obtain the power consumption data. The experimental results show that the proposed scheme has some advantages in computing and communication costs.
2023, 45(3): 828-835.
doi: 10.11999/JEIT221050
Abstract:
Privacy protection is a hot topic in information security, where the privacy issues in Attribute Based Encryption(ABE) can be divided into data content privacy, policy privacy and attribute privacy. Considering the three privacy protection needs of data content, policy and attributes, an attribute-based Privacy-Preserving Encryption Scheme based on inner product predicates (PPES) is proposed. The privacy of data content is ensured by using confidentiality of encryption algorithm, furthermore the blind method of policy attributes and user attributes is constructed through vector commitment protocol to achieve policy privacy and attribute privacy. Based on the hybrid argument technology, adaptive chosen plaintext security of the scheme is proved under standard model. Besides commitment unforgeability of the scheme is also illustrated. The performance analysis results show that the proposed scheme has better operation efficiency compared to existing methods.
Privacy protection is a hot topic in information security, where the privacy issues in Attribute Based Encryption(ABE) can be divided into data content privacy, policy privacy and attribute privacy. Considering the three privacy protection needs of data content, policy and attributes, an attribute-based Privacy-Preserving Encryption Scheme based on inner product predicates (PPES) is proposed. The privacy of data content is ensured by using confidentiality of encryption algorithm, furthermore the blind method of policy attributes and user attributes is constructed through vector commitment protocol to achieve policy privacy and attribute privacy. Based on the hybrid argument technology, adaptive chosen plaintext security of the scheme is proved under standard model. Besides commitment unforgeability of the scheme is also illustrated. The performance analysis results show that the proposed scheme has better operation efficiency compared to existing methods.
2023, 45(3): 836-846.
doi: 10.11999/JEIT221077
Abstract:
In order to solve the problems of identity privacy easy to leak, difficult to monitor and difficult to share the traceability data in the existing edible agricultural products traceability scheme, a traceability schema of edible agricultural products based on novel fair blind signature and attribute-based encryption is proposed. Based on the authorized access and non-tampering characteristics of consortium blockchain, a novel fair blind signature method is proposed by combining elliptic curve and zero-knowledge proof, which achieves anonymity of the identity of the edible agricultural product data uploader and avoids the problem of enmiting the signer through the double ID mechanism. At the same time, the attribute-based encryption improved by Asmuth-Bloom threshold combined with smart contract technology is adopted to realize the secret sharing of traceability data of edible agricultural products with hierarchical permissions. The analysis and experimental results demonstrate that the proposed scheme has good security and functionality.
In order to solve the problems of identity privacy easy to leak, difficult to monitor and difficult to share the traceability data in the existing edible agricultural products traceability scheme, a traceability schema of edible agricultural products based on novel fair blind signature and attribute-based encryption is proposed. Based on the authorized access and non-tampering characteristics of consortium blockchain, a novel fair blind signature method is proposed by combining elliptic curve and zero-knowledge proof, which achieves anonymity of the identity of the edible agricultural product data uploader and avoids the problem of enmiting the signer through the double ID mechanism. At the same time, the attribute-based encryption improved by Asmuth-Bloom threshold combined with smart contract technology is adopted to realize the secret sharing of traceability data of edible agricultural products with hierarchical permissions. The analysis and experimental results demonstrate that the proposed scheme has good security and functionality.
2023, 45(3): 847-855.
doi: 10.11999/JEIT221174
Abstract:
In order to protect the privacy of\begin{document}${\rm{users}}' $\end{document} ![]()
![]()
\begin{document}${\rm{users}}' $\end{document} ![]()
![]()
\begin{document}${\rm{users}}' $\end{document} ![]()
![]()
\begin{document}${\rm{Users}}' $\end{document} ![]()
![]()
In order to protect the privacy of
2023, 45(3): 856-864.
doi: 10.11999/JEIT220956
Abstract:
Remote sensing image storage and retrieval outsourcing to a semi-trusted cloud platform may lead to image data leakage and return incomplete retrieval results. Although encryption can protect the security of image data, it can not ensure that the cloud platform provides accurate and complete storage and retrieval services. Blockchain technology guarantees the authenticity and integrity of storage and retrieval services, but the computation and storage capacity of blockchain are limited, which makes it a challenging problem to realize secure storage and retrieval of remote sensing images. This paper proposes a secure remote sensing image retrieval scheme based on blockchain and cloud computing platforms. To secure the validity of cloud-saved images, the image hash and lightweight data are first stored on the blockchain, while the cloud platform stores huge encrypted image data. Moreover, the blockchain performs attribute-based retrieval of remote sensing images, and the cloud platform performs content-based secure retrieval to ensure the integrity of the retrieval results. Finally, the remote sensing image retrieval transaction mechanism is designed using blockchain technology. The experiments show that the proposed scheme can achieve secure, reliable, and fair remote sensing image retrieval, enabling both trading sides to benefit from a high-trust and fair-trading environment.
Remote sensing image storage and retrieval outsourcing to a semi-trusted cloud platform may lead to image data leakage and return incomplete retrieval results. Although encryption can protect the security of image data, it can not ensure that the cloud platform provides accurate and complete storage and retrieval services. Blockchain technology guarantees the authenticity and integrity of storage and retrieval services, but the computation and storage capacity of blockchain are limited, which makes it a challenging problem to realize secure storage and retrieval of remote sensing images. This paper proposes a secure remote sensing image retrieval scheme based on blockchain and cloud computing platforms. To secure the validity of cloud-saved images, the image hash and lightweight data are first stored on the blockchain, while the cloud platform stores huge encrypted image data. Moreover, the blockchain performs attribute-based retrieval of remote sensing images, and the cloud platform performs content-based secure retrieval to ensure the integrity of the retrieval results. Finally, the remote sensing image retrieval transaction mechanism is designed using blockchain technology. The experiments show that the proposed scheme can achieve secure, reliable, and fair remote sensing image retrieval, enabling both trading sides to benefit from a high-trust and fair-trading environment.
2023, 45(3): 865-873.
doi: 10.11999/JEIT220284
Abstract:
Focusing on the problem of user privacy leakage during consortium blockchain transactions, IDentity Ring SignCryption based on Consortium Blockchain (CB-IDRSC) is devised in this paper. CB-IDRSC uses the smart contract technology to control the addition of new transactions, and so realizes its fairness and reliability; It uses the multiple Private Key Generators (PKGs) to generate the private key information for users, and so satisfies the requirements of partial decentralization of consortium blockchain and can protect the node privacy; In addition, it has the confidentiality, unforgeability and unconditional anonymity of ring signcryptors. In performance analysis, the smart contract used in CB-IDRSC is firstly deployed, and high computation efficiency of CB-IDRSC is shown by efficiency analysis. By ignoring the influence of network delay and other factors, the experiments show the influence of the number of PKGs to efficiency of setup phase with key generation phase is less than 3%.
Focusing on the problem of user privacy leakage during consortium blockchain transactions, IDentity Ring SignCryption based on Consortium Blockchain (CB-IDRSC) is devised in this paper. CB-IDRSC uses the smart contract technology to control the addition of new transactions, and so realizes its fairness and reliability; It uses the multiple Private Key Generators (PKGs) to generate the private key information for users, and so satisfies the requirements of partial decentralization of consortium blockchain and can protect the node privacy; In addition, it has the confidentiality, unforgeability and unconditional anonymity of ring signcryptors. In performance analysis, the smart contract used in CB-IDRSC is firstly deployed, and high computation efficiency of CB-IDRSC is shown by efficiency analysis. By ignoring the influence of network delay and other factors, the experiments show the influence of the number of PKGs to efficiency of setup phase with key generation phase is less than 3%.
2023, 45(3): 874-883.
doi: 10.11999/JEIT221081
Abstract:
In view of the fact that there is rare research on the incentive mechanism design in decentralized federated learning, and the existing incentive mechanisms for federated learning are seldom based on the global model effect, an incentive mechanism based on contract theory, is added into decentralized federated learning and a new incentivized federated learning model is proposed. A blockchain and an InterPlanetary File System (IPFS) are used to replace the central server of traditional federated learning for model parameter storage and distribution, based on which a contract publisher is responsible for contract formulation and distribution, and each federated learning participant chooses to sign a contract based on its local data quality. The contract publisher evaluates each local training model after each round of local training and issues a reward based on the agreed-upon conditions in the contract. The global model aggregation also aggregates model parameters based on the reward results. Experimental validation on the MNIST dataset and industry electricity consumption dataset show that the proposed incentivized federated learning model outperforms traditional federated learning and its decentralized structure improves its robustness.
In view of the fact that there is rare research on the incentive mechanism design in decentralized federated learning, and the existing incentive mechanisms for federated learning are seldom based on the global model effect, an incentive mechanism based on contract theory, is added into decentralized federated learning and a new incentivized federated learning model is proposed. A blockchain and an InterPlanetary File System (IPFS) are used to replace the central server of traditional federated learning for model parameter storage and distribution, based on which a contract publisher is responsible for contract formulation and distribution, and each federated learning participant chooses to sign a contract based on its local data quality. The contract publisher evaluates each local training model after each round of local training and issues a reward based on the agreed-upon conditions in the contract. The global model aggregation also aggregates model parameters based on the reward results. Experimental validation on the MNIST dataset and industry electricity consumption dataset show that the proposed incentivized federated learning model outperforms traditional federated learning and its decentralized structure improves its robustness.
2023, 45(3): 884-893.
doi: 10.11999/JEIT220070
Abstract:
The emergence of mobile medical social networks has greatly facilitated the communication of patients' conditions to each other, promoting efficient and high-quality communication and exchange between patients. However, it also causes problems of confidentiality and privacy of patient data. To solve this problem, an attribute-based signcryption scheme based on cloud-assisted verification is proposed, which can effectively protect the privacy of patient data. Patients signcrypt their health information and upload it to the cloud server. When the data user wants to access the patient's information, the cloud server helps the data user partially decrypt and verify the integrity of the data, which reduces the amount of calculation of the data user to a certain extent. At the same time, under the random oracle model, it is proved that the scheme satisfies the unforgeability under the adaptive selection message, the indistinguishability under the adaptive selection ciphertext attack, and the attribute privacy security. Theoretical analysis and numerical simulation experimental results show that the scheme is more efficient than the existing schemes in the signcryption and unsigncryption phases.
The emergence of mobile medical social networks has greatly facilitated the communication of patients' conditions to each other, promoting efficient and high-quality communication and exchange between patients. However, it also causes problems of confidentiality and privacy of patient data. To solve this problem, an attribute-based signcryption scheme based on cloud-assisted verification is proposed, which can effectively protect the privacy of patient data. Patients signcrypt their health information and upload it to the cloud server. When the data user wants to access the patient's information, the cloud server helps the data user partially decrypt and verify the integrity of the data, which reduces the amount of calculation of the data user to a certain extent. At the same time, under the random oracle model, it is proved that the scheme satisfies the unforgeability under the adaptive selection message, the indistinguishability under the adaptive selection ciphertext attack, and the attribute privacy security. Theoretical analysis and numerical simulation experimental results show that the scheme is more efficient than the existing schemes in the signcryption and unsigncryption phases.
2023, 45(3): 894-904.
doi: 10.11999/JEIT221088
Abstract:
Federated learning can effectively circumvent the data privacy issues of participants, but the parameters or gradients passed in model training may still leak the privacy of the participants. Also, the existence of malicious participants can seriously affect the aggregation process and model quality. In this paper, a trusted Federated Secure Aggregation method based on Similarity Clustering named FSA-SC is proposed. Firstly, the weight for each client can be measured based on the size of the client training data set and the communication distance between the client and the server, and those participants with higher weight are selected in the server-side model aggregation. Secondly, according to the similarity between the candidate clients, the candidate clients are divided into two groups, i.e., benign group and abnormal group. Finally, for the abnormal group, an intra-class broadcast and secondary negotiation are designed to replace and record the parameters of the members, so as to detect effectively malicious clients. In order to verify the effectiveness of FSA-SC, taking federated recommendation as the application scenario, experimental results on MovieLens 1M, Netflix and Amazon datasets indicate that FSA-SC can achieve efficient security aggregation and has greater robustness than baselines.
Federated learning can effectively circumvent the data privacy issues of participants, but the parameters or gradients passed in model training may still leak the privacy of the participants. Also, the existence of malicious participants can seriously affect the aggregation process and model quality. In this paper, a trusted Federated Secure Aggregation method based on Similarity Clustering named FSA-SC is proposed. Firstly, the weight for each client can be measured based on the size of the client training data set and the communication distance between the client and the server, and those participants with higher weight are selected in the server-side model aggregation. Secondly, according to the similarity between the candidate clients, the candidate clients are divided into two groups, i.e., benign group and abnormal group. Finally, for the abnormal group, an intra-class broadcast and secondary negotiation are designed to replace and record the parameters of the members, so as to detect effectively malicious clients. In order to verify the effectiveness of FSA-SC, taking federated recommendation as the application scenario, experimental results on MovieLens 1M, Netflix and Amazon datasets indicate that FSA-SC can achieve efficient security aggregation and has greater robustness than baselines.
2023, 45(3): 905-912.
doi: 10.11999/JEIT211259
Abstract:
Basic Probability Assignment (BPA) has no fixed generative model in the application of evidence theory. To solve this problem, a BPA generation method based on cloud model is proposed. Firstly, based on the normal cloud model of the sample attributes, the BPA model function of the single subset proposition is constructed, and the model function of the composite subset is expressed as Gaussian function product fusion. Secondly, a method of dynamically measuring attribute weights based on test samples is proposed to take into account the reliability of information sources. Finally, the BPA is obtained by modifying the output result of the model function with attribute weights. The classification and recognition experiments of iris and other data sets show that this method has high recognition accuracy and is suitable for situations with fewer samples.
Basic Probability Assignment (BPA) has no fixed generative model in the application of evidence theory. To solve this problem, a BPA generation method based on cloud model is proposed. Firstly, based on the normal cloud model of the sample attributes, the BPA model function of the single subset proposition is constructed, and the model function of the composite subset is expressed as Gaussian function product fusion. Secondly, a method of dynamically measuring attribute weights based on test samples is proposed to take into account the reliability of information sources. Finally, the BPA is obtained by modifying the output result of the model function with attribute weights. The classification and recognition experiments of iris and other data sets show that this method has high recognition accuracy and is suitable for situations with fewer samples.
2023, 45(3): 913-920.
doi: 10.11999/JEIT220014
Abstract:
Based on a length-N Golay Complementary sequence Pair (GCP) as a seed, a Z-optimal binary Z-Complementary sequence Pair (ZCP) with length N+3 is constructed by inserting three specific components at chosen positions in the aforementioned seed, where N is an integer. Compared with the known Type-II Z-optimal binary ZCPs of the same length, the resultant pairs have lower Peak-to-Mean Envelope Power Ratio (PMEPR). Both ZCPs and GCPs are widely used in Orthogonal Frequency Division Multiplexing (OFDM) systems and Code Division Multiple Access (CDMA) system, etc, however, the former has more flexible lengths and larger family sizes, which can better meet the requirements of applications.
Based on a length-N Golay Complementary sequence Pair (GCP) as a seed, a Z-optimal binary Z-Complementary sequence Pair (ZCP) with length N+3 is constructed by inserting three specific components at chosen positions in the aforementioned seed, where N is an integer. Compared with the known Type-II Z-optimal binary ZCPs of the same length, the resultant pairs have lower Peak-to-Mean Envelope Power Ratio (PMEPR). Both ZCPs and GCPs are widely used in Orthogonal Frequency Division Multiplexing (OFDM) systems and Code Division Multiple Access (CDMA) system, etc, however, the former has more flexible lengths and larger family sizes, which can better meet the requirements of applications.
2023, 45(3): 921-928.
doi: 10.11999/JEIT220499
Abstract:
An algebraic expression for the S-box of Camellia’s algorithm based on irreducible polynomials is proposed in this paper, and eight different isomorphic expressions are also given. Then combined with the characteristics of S-box, an optimization scheme based on polynomial basis is given by theoretical proof, in which some redundant linear operations are reduced. Compared with the same gate-limited scheme the circuit area is saved by 9.12% in the Semiconductor Manufacturing International Corporation (SMIC) 130 nm process library and by 8.31% in the SMIC 65 nm process library. Finally, according to the computational redundancy in the design of the S-box of Camellia algorithm, two completely equivalent representations on the finite field are given, which will have a positive impact on the optimization of the S-box of Camellia algorithm.
An algebraic expression for the S-box of Camellia’s algorithm based on irreducible polynomials is proposed in this paper, and eight different isomorphic expressions are also given. Then combined with the characteristics of S-box, an optimization scheme based on polynomial basis is given by theoretical proof, in which some redundant linear operations are reduced. Compared with the same gate-limited scheme the circuit area is saved by 9.12% in the Semiconductor Manufacturing International Corporation (SMIC) 130 nm process library and by 8.31% in the SMIC 65 nm process library. Finally, according to the computational redundancy in the design of the S-box of Camellia algorithm, two completely equivalent representations on the finite field are given, which will have a positive impact on the optimization of the S-box of Camellia algorithm.
2023, 45(3): 929-940.
doi: 10.11999/JEIT211605
Abstract:
During the modeling and analysis of neuronal activity, several biophysical effects should be taken into consideration. Due to fluctuations in intracellular and extracellular ion concentrations within the nervous system, internal fluctuations of electromagnetic fields and the effects of transmembrane magnetic flux need to be considered in the collective electrical activity and signal propagation between neuronal clusters. In this paper, a magnetic flux variable is introduced into a hybrid neuron, and a complex time-varying electromagnetic field is induced by modulating the membrane potential. Using analytical tools such as Xppauto, Matcont and Matlab, the existence and initial value of the equilibrium point of the new model, sensitivity and two-parameter bifurcation are discussed. When the external stimulus current and electromagnetic field change, the new model can be induced to generate abundant discharge modes, such as resting state, spike discharge, periodic (or chaotic) cluster discharge, especially coexisting discharge and hidden discharge benefited from the introduction of magnetic flux variable and memristor. According to the above analysis, the neuron model based on electromagnetic induction has high nonlinearity and more sensitive parameters, which enables the encryption algorithm to have a large key space. Based on this, an image encryption algorithm is designed in this paper. Pixels are first diffused once and then scrambled twice to their positions. Finally, through a series of numerical experiments, it is proved that the designed encryption algorithm can encrypt images effectively and has high security. The research takes into account the electromagnetic induction effect inside and outside the nerve cells, which is helpful for a more comprehensive understanding of the information encoding and transition laws between neurons. More bifurcation parameters and high complexity also make the designed neuron model has a good application prospect in image encryption.
During the modeling and analysis of neuronal activity, several biophysical effects should be taken into consideration. Due to fluctuations in intracellular and extracellular ion concentrations within the nervous system, internal fluctuations of electromagnetic fields and the effects of transmembrane magnetic flux need to be considered in the collective electrical activity and signal propagation between neuronal clusters. In this paper, a magnetic flux variable is introduced into a hybrid neuron, and a complex time-varying electromagnetic field is induced by modulating the membrane potential. Using analytical tools such as Xppauto, Matcont and Matlab, the existence and initial value of the equilibrium point of the new model, sensitivity and two-parameter bifurcation are discussed. When the external stimulus current and electromagnetic field change, the new model can be induced to generate abundant discharge modes, such as resting state, spike discharge, periodic (or chaotic) cluster discharge, especially coexisting discharge and hidden discharge benefited from the introduction of magnetic flux variable and memristor. According to the above analysis, the neuron model based on electromagnetic induction has high nonlinearity and more sensitive parameters, which enables the encryption algorithm to have a large key space. Based on this, an image encryption algorithm is designed in this paper. Pixels are first diffused once and then scrambled twice to their positions. Finally, through a series of numerical experiments, it is proved that the designed encryption algorithm can encrypt images effectively and has high security. The research takes into account the electromagnetic induction effect inside and outside the nerve cells, which is helpful for a more comprehensive understanding of the information encoding and transition laws between neurons. More bifurcation parameters and high complexity also make the designed neuron model has a good application prospect in image encryption.
2023, 45(3): 941-948.
doi: 10.11999/JEIT211405
Abstract:
Considering the deficiency that traditional intention recognition methods can only deal with certain types of uncertain information, a new information processing method of fuzzy Belief-Rule-Base (BRB) is proposed, which combines the advantages of fuzzy sets and Dempster-Shafer (DS) theory. Firstly, the connection relation of premise attributes is improved in the premise part of confidence rules, and fuzzy set segmentation is designed according to the statistical distribution characteristics of data sets. Cauchy distribution is selected as membership function to avoid the problem that confidence rules could not be activated effectively, which would lead to no effective output of the system. Secondly, the confidence distribution of different categories in the identification framework is fused, and the optimization model of rule weight and feature weight is established, and the input-output relationship between feature space and category space is constructed. On this basis, the matching degree and activation degree of the unknown intention data are calculated, and the recognition decision is made using the maximum confidence principle. Through experimental verification, sensitive parameter and interpretation of result, time complexity analysis, compared with other methods, the fuzzy belief-rule-base shows high accuracy rate, and effectiveness and reliability under the condition of small samples.
Considering the deficiency that traditional intention recognition methods can only deal with certain types of uncertain information, a new information processing method of fuzzy Belief-Rule-Base (BRB) is proposed, which combines the advantages of fuzzy sets and Dempster-Shafer (DS) theory. Firstly, the connection relation of premise attributes is improved in the premise part of confidence rules, and fuzzy set segmentation is designed according to the statistical distribution characteristics of data sets. Cauchy distribution is selected as membership function to avoid the problem that confidence rules could not be activated effectively, which would lead to no effective output of the system. Secondly, the confidence distribution of different categories in the identification framework is fused, and the optimization model of rule weight and feature weight is established, and the input-output relationship between feature space and category space is constructed. On this basis, the matching degree and activation degree of the unknown intention data are calculated, and the recognition decision is made using the maximum confidence principle. Through experimental verification, sensitive parameter and interpretation of result, time complexity analysis, compared with other methods, the fuzzy belief-rule-base shows high accuracy rate, and effectiveness and reliability under the condition of small samples.
2023, 45(3): 949-957.
doi: 10.11999/JEIT211594
Abstract:
Accurately predicting the evolution process and results of emergencies is of great reference to formulate the emergency response plans of the urban rail transit system and safeguard its secure operation. However, the prediction methods of emergency evolution results are lack of high intelligence, and excessively depend on the feature weighting and retrieval template set subjectively by policymakers, which is complicated, inaccurate, and short of applicability. Based on Knowledge Graph(KG) and Relational-Graph Convolution Neural network(R-GCN), a predicting method of evolution result of urban rail transit emergencies is proposed. A knowledge graph of urban rail transit emergencies is constructed to combine with contextual information related to the emergency for structured processing. Firstly, the knowledge graph of urban rail transit emergencies is constructed to combine with contextual information related to the emergency for structured processing. Then the predicting model of urban rail transit emergencies is constructed based on the relational-graph convolution neural network to achieve the result prediction of urban rail transit emergency. Finally, the verification is conducted via case base of urban rail transit emergency. The experimental result demonstrates that the predicting method proposed in this paper is of high accuracy and applicability, which can provide consolidated data and decision support for rail transit emergency management.
Accurately predicting the evolution process and results of emergencies is of great reference to formulate the emergency response plans of the urban rail transit system and safeguard its secure operation. However, the prediction methods of emergency evolution results are lack of high intelligence, and excessively depend on the feature weighting and retrieval template set subjectively by policymakers, which is complicated, inaccurate, and short of applicability. Based on Knowledge Graph(KG) and Relational-Graph Convolution Neural network(R-GCN), a predicting method of evolution result of urban rail transit emergencies is proposed. A knowledge graph of urban rail transit emergencies is constructed to combine with contextual information related to the emergency for structured processing. Firstly, the knowledge graph of urban rail transit emergencies is constructed to combine with contextual information related to the emergency for structured processing. Then the predicting model of urban rail transit emergencies is constructed based on the relational-graph convolution neural network to achieve the result prediction of urban rail transit emergency. Finally, the verification is conducted via case base of urban rail transit emergency. The experimental result demonstrates that the predicting method proposed in this paper is of high accuracy and applicability, which can provide consolidated data and decision support for rail transit emergency management.
2023, 45(3): 958-967.
doi: 10.11999/JEIT211577
Abstract:
Bone Age (BA) is one of the most important indicators in evaluating children's growth. The Bone Age Assessment (BAA) based on Chinese wrist bone development standard-CHN (CHN) scoring method is widely used in the evaluation of children's growth and development and height prediction. However, the adjacent developmental levels of some reference bones last longer, leading to the subjective judgment of developmental levels by experts based on personal experience, which affects the accuracy of predictions. When deep learning is used to evaluate the developmental levels of these atlases, the prediction results will be random. In this paper, based on more than 20000 X-ray images evaluated by experts, a new mature indicator with a large interval with a large interval is drawn to generate exquisite atlas to perform some reference bones. Additionally, the corresponding maturity score is determined by analyzing the level structure process to maximize the impact of error -level prediction on BAA. Combining Harris features and convolutional blocks of the convolutional neural network of the attention module is designed to evaluate automatically the level of bone maturity. In addition, an annotated database with an age distribution of 5-11 years is built to train and evaluate the method. The accuracy of predictions obtained by adding a new standard atlas to the CHN method reaches 94.6% and 99.13% when the tolerance is 0.5 years and 1 year, respectively. The experimental results show that the method proposed in this paper can distinguish the development degree of reference bones more precisely, and improve greatly the accuracy of BAA, proving the potential for practical clinical application.
Bone Age (BA) is one of the most important indicators in evaluating children's growth. The Bone Age Assessment (BAA) based on Chinese wrist bone development standard-CHN (CHN) scoring method is widely used in the evaluation of children's growth and development and height prediction. However, the adjacent developmental levels of some reference bones last longer, leading to the subjective judgment of developmental levels by experts based on personal experience, which affects the accuracy of predictions. When deep learning is used to evaluate the developmental levels of these atlases, the prediction results will be random. In this paper, based on more than 20000 X-ray images evaluated by experts, a new mature indicator with a large interval with a large interval is drawn to generate exquisite atlas to perform some reference bones. Additionally, the corresponding maturity score is determined by analyzing the level structure process to maximize the impact of error -level prediction on BAA. Combining Harris features and convolutional blocks of the convolutional neural network of the attention module is designed to evaluate automatically the level of bone maturity. In addition, an annotated database with an age distribution of 5-11 years is built to train and evaluate the method. The accuracy of predictions obtained by adding a new standard atlas to the CHN method reaches 94.6% and 99.13% when the tolerance is 0.5 years and 1 year, respectively. The experimental results show that the method proposed in this paper can distinguish the development degree of reference bones more precisely, and improve greatly the accuracy of BAA, proving the potential for practical clinical application.
2023, 45(3): 968-976.
doi: 10.11999/JEIT211600
Abstract:
Space targets have the characteristics of wide coverage, fast speed, high target density and similar movement, which lead to that in a relatively long time these targets can not be effectively distinguished. How to distinguish effectively the number and position of these non-cooperative space targets as soon as possible is very important. Therefore, based on Random Finite Set (RFS) theory and dynamic model of space targets, the number and position estimation method of unresolved space group targets is studied in this paper, which can effectively estimate the number and position of space group targets with high-speed and small spatial distribution range in the early stage of target monitoring. This method makes full use of the characteristics of Probability Hypothesis Density (PHD) filter, which can solve the number and state estimation of targets in unknown time-varying environment. The Gaussian Mixture PHD (GM-PHD) filter is combined with the space target dynamic equation to estimate the number of unresolved space targets, and the target state are estimated more effectively by the constraint of the dynamic equation. At the same time of target tracking, the resolution problem of unresolved space group targets can be solved. The correlation algorithms can provide data basis for state estimation, continuous stable tracking and reliable trajectory prediction of special value individual target in the group.
Space targets have the characteristics of wide coverage, fast speed, high target density and similar movement, which lead to that in a relatively long time these targets can not be effectively distinguished. How to distinguish effectively the number and position of these non-cooperative space targets as soon as possible is very important. Therefore, based on Random Finite Set (RFS) theory and dynamic model of space targets, the number and position estimation method of unresolved space group targets is studied in this paper, which can effectively estimate the number and position of space group targets with high-speed and small spatial distribution range in the early stage of target monitoring. This method makes full use of the characteristics of Probability Hypothesis Density (PHD) filter, which can solve the number and state estimation of targets in unknown time-varying environment. The Gaussian Mixture PHD (GM-PHD) filter is combined with the space target dynamic equation to estimate the number of unresolved space targets, and the target state are estimated more effectively by the constraint of the dynamic equation. At the same time of target tracking, the resolution problem of unresolved space group targets can be solved. The correlation algorithms can provide data basis for state estimation, continuous stable tracking and reliable trajectory prediction of special value individual target in the group.
2023, 45(3): 977-986.
doi: 10.11999/JEIT211619
Abstract:
In the radar electronic reconnaissance environment, the Discrete Frequency Coded (DFC) waveform signals emitted by non-cooperative targets with low probability of interception and anti-interference is hard to be accumulated and detected under low Signal-to-Noise Ratio (SNR) conditions. Consequently, a joint accumulation and detection algorithm is proposed in this paper. First, correlated accumulation and incoherent accumulation are jointly used to obtain signal envelopes from low SNR environments. Then, the bi-directional Constant False Alarm Rate (CFAR) threshold and pulse edge decision criteria are used to detect pulses and estimate accurate time of arrival and pulse width. Compared with conventional algorithms, the proposed algorithm could realize the accurate detection of discrete frequency coded waveform signals without any prior information, with low detection false alarm rate and good robustness. Simulation experiments verify the effectiveness and robustness of the algorithm in this paper.
In the radar electronic reconnaissance environment, the Discrete Frequency Coded (DFC) waveform signals emitted by non-cooperative targets with low probability of interception and anti-interference is hard to be accumulated and detected under low Signal-to-Noise Ratio (SNR) conditions. Consequently, a joint accumulation and detection algorithm is proposed in this paper. First, correlated accumulation and incoherent accumulation are jointly used to obtain signal envelopes from low SNR environments. Then, the bi-directional Constant False Alarm Rate (CFAR) threshold and pulse edge decision criteria are used to detect pulses and estimate accurate time of arrival and pulse width. Compared with conventional algorithms, the proposed algorithm could realize the accurate detection of discrete frequency coded waveform signals without any prior information, with low detection false alarm rate and good robustness. Simulation experiments verify the effectiveness and robustness of the algorithm in this paper.
2023, 45(3): 987-995.
doi: 10.11999/JEIT220063
Abstract:
Considering the issue of difference and complementarity of multi-source remote sensing images, this paper proposes a feature fusion classification method for optical image and SAR image based on spatial-spectral attention. Firstly, features of optical image and SAR image are extracted by the convolutional neural network, and an attention module composed of spatial attention and spectral attention is designed to analyze the importance of features. Features can be enhanced by the weights of the attention module, which can reduce the attention to irrelevant information, and thus improve the accuracy of fusion classification for optical and SAR images. Experimental results on two datasets of optical image and SAR image demonstrate that the proposed method is able to yield higher fusion classification accuracy.
Considering the issue of difference and complementarity of multi-source remote sensing images, this paper proposes a feature fusion classification method for optical image and SAR image based on spatial-spectral attention. Firstly, features of optical image and SAR image are extracted by the convolutional neural network, and an attention module composed of spatial attention and spectral attention is designed to analyze the importance of features. Features can be enhanced by the weights of the attention module, which can reduce the attention to irrelevant information, and thus improve the accuracy of fusion classification for optical and SAR images. Experimental results on two datasets of optical image and SAR image demonstrate that the proposed method is able to yield higher fusion classification accuracy.
2023, 45(3): 996-1005.
doi: 10.11999/JEIT220045
Abstract:
In order to solve the problem of in-band interference of enhanced LOng RAnge Navigation (eLORAN) system, an adaptive Weight-Function Notch Filter (WF-NF) algorithm is proposed based on the principle of Notch Filter (NF) to resist narrow-band interference. Firstly, the applicability of NF algorithm in eLORAN system and the pseudo convergence caused by detection frequency difference are analyzed. Then, by adding the weight function of frequency difference, a new WF-NF algorithm is derived to solve the notch logic flaw caused by the inherent frequency difference. Next, a gradient-jump optimization technique is designed to improve the convergence speed of weight coefficients. Finally, the stability, multiple applicability and practicability of the new algorithm are verified by simulation experiments in various environments. The analysis results show that the WF-NF algorithm can overcome the practical problem of detection frequency difference, realize efficient interference tracking and filtering, and compensate the accuracy of interference detection, so as to provide guarantee for the subsequent signal processing process.
In order to solve the problem of in-band interference of enhanced LOng RAnge Navigation (eLORAN) system, an adaptive Weight-Function Notch Filter (WF-NF) algorithm is proposed based on the principle of Notch Filter (NF) to resist narrow-band interference. Firstly, the applicability of NF algorithm in eLORAN system and the pseudo convergence caused by detection frequency difference are analyzed. Then, by adding the weight function of frequency difference, a new WF-NF algorithm is derived to solve the notch logic flaw caused by the inherent frequency difference. Next, a gradient-jump optimization technique is designed to improve the convergence speed of weight coefficients. Finally, the stability, multiple applicability and practicability of the new algorithm are verified by simulation experiments in various environments. The analysis results show that the WF-NF algorithm can overcome the practical problem of detection frequency difference, realize efficient interference tracking and filtering, and compensate the accuracy of interference detection, so as to provide guarantee for the subsequent signal processing process.
2023, 45(3): 1006-1014.
doi: 10.11999/JEIT220031
Abstract:
In order to solve the problem that the state estimation accuracy of mobile robot in Simultaneous Localization And Mapping (SLAM) is reduced due to the time-varying system noise and observed noise, a SLAM algorithm is proposed based on variational Bayes Double-Scale Adaptive time-varying noise Cubature Kalman Filter (DSACKF SLAM). The inverse Wishart distribution is used to model the one-step predicted error covariance matrix\begin{document}$ {{\boldsymbol{P}}_{k|k - 1}} $\end{document} ![]()
![]()
\begin{document}${{\boldsymbol{R}}_k}$\end{document} ![]()
![]()
\begin{document}${{\boldsymbol{X}}_k}$\end{document} ![]()
![]()
\begin{document}$ {{\boldsymbol{P}}_{k|k - 1}} $\end{document} ![]()
![]()
\begin{document}${{\boldsymbol{R}}_k}$\end{document} ![]()
![]()
In order to solve the problem that the state estimation accuracy of mobile robot in Simultaneous Localization And Mapping (SLAM) is reduced due to the time-varying system noise and observed noise, a SLAM algorithm is proposed based on variational Bayes Double-Scale Adaptive time-varying noise Cubature Kalman Filter (DSACKF SLAM). The inverse Wishart distribution is used to model the one-step predicted error covariance matrix
2023, 45(3): 1015-1022.
doi: 10.11999/JEIT220005
Abstract:
Large-range maritime vessel targets can be detected continuously by High-Frequency Surface Wave Radar (HFSWR), but the tracking trajectory of the target is easily broken in the presence of disturbing factors such as sea clutter. In current studies on HFSWR track association, the case of broken tracks is usually ignored and the track association is considered as a bipartite graph matching problem, which can lead to the possibility of judging broken tracks of a single target as multiple targets, and thus wrong target association results are obtained. For the above situation, fuzzy integrated evaluation and iterative search algorithms are considered in this paper. The Gale-Shapley (GS) algorithm is introduced into the field of track association for the first time, and it is improved to satisfy the many-to-many track association case when the track is broken , the Improved Gale-Shapley (IGS) algorithm is proposed. In this algorithm, the tendency sequences between the tracks can be obtained by calculating the fuzzy composite judgment values between the tracks. Then, the tracks are clustered by an iterative search method to obtain the track clusters. Finally, the track clusters and the propensity sequences are fed into the Gale-Shapley algorithm to perform several rounds of games to give the association results. The measured data and simulation data of dual-frequency HFSWR and Automatic Identification System (AIS) are used for experimental tests. Experimental tests are conducted using simulated and measured data from dual-frequency HFSWR and AIS. The experimental results show that the multi-sensor track association problem in the case of track break can be solved by the proposed algorithm, and the track association effect in dense areas is better than that of the conventional algorithm.
Large-range maritime vessel targets can be detected continuously by High-Frequency Surface Wave Radar (HFSWR), but the tracking trajectory of the target is easily broken in the presence of disturbing factors such as sea clutter. In current studies on HFSWR track association, the case of broken tracks is usually ignored and the track association is considered as a bipartite graph matching problem, which can lead to the possibility of judging broken tracks of a single target as multiple targets, and thus wrong target association results are obtained. For the above situation, fuzzy integrated evaluation and iterative search algorithms are considered in this paper. The Gale-Shapley (GS) algorithm is introduced into the field of track association for the first time, and it is improved to satisfy the many-to-many track association case when the track is broken , the Improved Gale-Shapley (IGS) algorithm is proposed. In this algorithm, the tendency sequences between the tracks can be obtained by calculating the fuzzy composite judgment values between the tracks. Then, the tracks are clustered by an iterative search method to obtain the track clusters. Finally, the track clusters and the propensity sequences are fed into the Gale-Shapley algorithm to perform several rounds of games to give the association results. The measured data and simulation data of dual-frequency HFSWR and Automatic Identification System (AIS) are used for experimental tests. Experimental tests are conducted using simulated and measured data from dual-frequency HFSWR and AIS. The experimental results show that the multi-sensor track association problem in the case of track break can be solved by the proposed algorithm, and the track association effect in dense areas is better than that of the conventional algorithm.
2023, 45(3): 1023-1031.
doi: 10.11999/JEIT211587
Abstract:
In the tactical communication environments, the characteristics of the wireless channel between mobile vehicle platforms become more complicated. The impact of the special tactical factors is not taken into account in the traditional mobile channel models. Therefore, these models are difficult to be applied directly to the design and optimization of vehicular systems in the tactical scenarios. In order to address the limitation of traditional mobile channel models, a Tactical Vehicle-to-Vehicle (T-V2V) channel model is proposed for the tactical communication environments, which considers jointly the influences of the mutual movement between two vehicle platforms, the alignment problem of directional antennas and the tactical terrain. Then, the proposed model is statistically analyzed based on the index of the Lever Crossing Rate (LCR) and the Average Duration of Fading (ADF). The simulation results show that the proposed model is more suitable for the actual situation and reflects the changing characteristics of V2V channel more accurately in the tactical environments. Finally, the relevant factors of the proposed model are analyzed and the conclusions can provide important guiding significance for the physical layer design in tactical communication environments.
In the tactical communication environments, the characteristics of the wireless channel between mobile vehicle platforms become more complicated. The impact of the special tactical factors is not taken into account in the traditional mobile channel models. Therefore, these models are difficult to be applied directly to the design and optimization of vehicular systems in the tactical scenarios. In order to address the limitation of traditional mobile channel models, a Tactical Vehicle-to-Vehicle (T-V2V) channel model is proposed for the tactical communication environments, which considers jointly the influences of the mutual movement between two vehicle platforms, the alignment problem of directional antennas and the tactical terrain. Then, the proposed model is statistically analyzed based on the index of the Lever Crossing Rate (LCR) and the Average Duration of Fading (ADF). The simulation results show that the proposed model is more suitable for the actual situation and reflects the changing characteristics of V2V channel more accurately in the tactical environments. Finally, the relevant factors of the proposed model are analyzed and the conclusions can provide important guiding significance for the physical layer design in tactical communication environments.
2023, 45(3): 1032-1042.
doi: 10.11999/JEIT220141
Abstract:
To address the disadvantages of small transmission rate and poor Bit Error Rate (BER) of M-ary differential chaos shift keying. An expanded capacity orthogonal noise suppression multi-level Differential Chaotic Shift Keying (DCSK) communication system is proposed. An improved orthogonal chaotic signal generator is designed at the transmitter of the system. It can generate four sets of orthogonal chaos-based signals, which can greatly increase the communication capacity. An integrated utility function is defined and a particle swarm algorithm is introduced to optimize each parameter of the system. The theoretical BER equation is derived and the system simulation is analyzed under the Additive White Gaussian Noise (AWGN) channel and Rayleigh fading channel. The integrated utility functions of different systems are also compared. The results show that the system has a lower BER and better integrated utility compared, and has a better practical application.
To address the disadvantages of small transmission rate and poor Bit Error Rate (BER) of M-ary differential chaos shift keying. An expanded capacity orthogonal noise suppression multi-level Differential Chaotic Shift Keying (DCSK) communication system is proposed. An improved orthogonal chaotic signal generator is designed at the transmitter of the system. It can generate four sets of orthogonal chaos-based signals, which can greatly increase the communication capacity. An integrated utility function is defined and a particle swarm algorithm is introduced to optimize each parameter of the system. The theoretical BER equation is derived and the system simulation is analyzed under the Additive White Gaussian Noise (AWGN) channel and Rayleigh fading channel. The integrated utility functions of different systems are also compared. The results show that the system has a lower BER and better integrated utility compared, and has a better practical application.
2023, 45(3): 1043-1054.
doi: 10.11999/JEIT220066
Abstract:
To solve the problem that the traditional jamming resource allocation algorithm needs relatively complete prior information when dealing with nonlinear combinatorial optimization problems, and meanwhile, the decision dimension is small, which can not meet the requirements of modern communication countermeasures, a Deep Reinforcement Learning communication jamming resource allocation algorithm Fused with Noise Network (FNNDRL) is proposed. Using the idea of noise network for reference, twin noise evaluation network, which can avoid the overestimation of Q value and improve the randomness of evaluation network to ensure the exploration of training process is designed by the algorithm. Based on the physical significance of the probability entropy, an improved strategy network loss function based on the strategy distribution entropy is designed to maximize the cumulative reward and the strategy distribution entropy to avoid convergence to local optimal in the process of strategy optimization. The simulation results show that the proposed algorithm is superior to the average allocation and reinforcement learning methods in solving the problem of jamming resource allocation. Meanwhile, the algorithm has high stability and strong adaptability to high-dimensional decision space.
To solve the problem that the traditional jamming resource allocation algorithm needs relatively complete prior information when dealing with nonlinear combinatorial optimization problems, and meanwhile, the decision dimension is small, which can not meet the requirements of modern communication countermeasures, a Deep Reinforcement Learning communication jamming resource allocation algorithm Fused with Noise Network (FNNDRL) is proposed. Using the idea of noise network for reference, twin noise evaluation network, which can avoid the overestimation of Q value and improve the randomness of evaluation network to ensure the exploration of training process is designed by the algorithm. Based on the physical significance of the probability entropy, an improved strategy network loss function based on the strategy distribution entropy is designed to maximize the cumulative reward and the strategy distribution entropy to avoid convergence to local optimal in the process of strategy optimization. The simulation results show that the proposed algorithm is superior to the average allocation and reinforcement learning methods in solving the problem of jamming resource allocation. Meanwhile, the algorithm has high stability and strong adaptability to high-dimensional decision space.
2023, 45(3): 1055-1064.
doi: 10.11999/JEIT220029
Abstract:
Device-to-Device (D2D) communication is one of the key technologies to solve the issue of spectrum resource scarcity. In this paper, a complex “many-to-many” scenario in cellular networks is investigated, where a Resource Block (RB) could be reused by more than one D2D pair, and a D2D pair is also allowed to use multiple RBs. Moreover, the number of D2D users is larger than that of Cellular User Equipments (CUEs) and RBs. Considering that CUEs have higher priority on resource, this optimization problem is decomposed into two stages: cellular users resource allocation and D2D user resource reuse. In the first stage, a Fairness-based Circular Bipartite Graph Matching (FCBGM) algorithm is proposed, which allocates the existing RBs to all CUEs to maximize the sum data rate of cellular users. In the second stage, a Bipartite Graph-based Resource Reuse (BGRR) algorithm and a Greedy-based Resource Reuse (GRR) algorithm is proposed, and the goal is to re-assign the RBs, which are already allocated to CUEs, to the D2D pairs in a way to maximize the system sum data rate while ensuring the basic data rate requirements of CUEs. Simulation results show that when the number of D2D pairs is much larger than that of CUEs and RBs, compared with the existing typical algorithms, the proposed algorithm can effectively improve system sum data rate and D2D access rate while guaranteeing user fairness and quality of service.
Device-to-Device (D2D) communication is one of the key technologies to solve the issue of spectrum resource scarcity. In this paper, a complex “many-to-many” scenario in cellular networks is investigated, where a Resource Block (RB) could be reused by more than one D2D pair, and a D2D pair is also allowed to use multiple RBs. Moreover, the number of D2D users is larger than that of Cellular User Equipments (CUEs) and RBs. Considering that CUEs have higher priority on resource, this optimization problem is decomposed into two stages: cellular users resource allocation and D2D user resource reuse. In the first stage, a Fairness-based Circular Bipartite Graph Matching (FCBGM) algorithm is proposed, which allocates the existing RBs to all CUEs to maximize the sum data rate of cellular users. In the second stage, a Bipartite Graph-based Resource Reuse (BGRR) algorithm and a Greedy-based Resource Reuse (GRR) algorithm is proposed, and the goal is to re-assign the RBs, which are already allocated to CUEs, to the D2D pairs in a way to maximize the system sum data rate while ensuring the basic data rate requirements of CUEs. Simulation results show that when the number of D2D pairs is much larger than that of CUEs and RBs, compared with the existing typical algorithms, the proposed algorithm can effectively improve system sum data rate and D2D access rate while guaranteeing user fairness and quality of service.
2023, 45(3): 1065-1073.
doi: 10.11999/JEIT220088
Abstract:
In Industrial Wireless Sensor Networks (IWSN), timely delivery of periodic control/sensing data flows and aperiodic event data flows is crucial to ensure production safety and efficiency. As a new metric of data freshness, Age of Information (AoI) can comprehensively measure the real-time performance of data delivery from the perspective of destination node. For industrial wireless sensor networks with hybrid periodic and aperiodic data, the data freshness metric of whole network is introduced. Considering that the freshness of periodic data exceeding the threshold may have a negative impact on industrial production, a joint optimization model is established, which minimizes the system average AoI and the probability of AoI overdue for periodic data, and then the optimization problem is formulated as a Markov Decision Process (MDP). Since the traditional optimal solution method based on relative value iteration is difficult to implement in large-scale networks result from dimensional disasters, Deep Reinforcement Learning (DRL) is used to reduce the state space dimension of the optimization problem. Moreover, the decision exploration mechanism is improved to speed up the learning speed, and a scheduling method of deep reinforcement learning based on Optimal Decision Exploration (DRL-ODE) is proposed. Simulation results show that the proposed method can improve the timeliness of network data delivery while reducing the probability of AoI overdue for periodic data effectively.
In Industrial Wireless Sensor Networks (IWSN), timely delivery of periodic control/sensing data flows and aperiodic event data flows is crucial to ensure production safety and efficiency. As a new metric of data freshness, Age of Information (AoI) can comprehensively measure the real-time performance of data delivery from the perspective of destination node. For industrial wireless sensor networks with hybrid periodic and aperiodic data, the data freshness metric of whole network is introduced. Considering that the freshness of periodic data exceeding the threshold may have a negative impact on industrial production, a joint optimization model is established, which minimizes the system average AoI and the probability of AoI overdue for periodic data, and then the optimization problem is formulated as a Markov Decision Process (MDP). Since the traditional optimal solution method based on relative value iteration is difficult to implement in large-scale networks result from dimensional disasters, Deep Reinforcement Learning (DRL) is used to reduce the state space dimension of the optimization problem. Moreover, the decision exploration mechanism is improved to speed up the learning speed, and a scheduling method of deep reinforcement learning based on Optimal Decision Exploration (DRL-ODE) is proposed. Simulation results show that the proposed method can improve the timeliness of network data delivery while reducing the probability of AoI overdue for periodic data effectively.
2023, 45(3): 1074-1082.
doi: 10.11999/JEIT220058
Abstract:
In order to solve the problem that slice migration lags behind by lacking awareness of physical network resources in 5G Network Slice (NS) scenarios, a Dynamic Slice Adjustment and Migration (DSAM) algorithm based on ensemble deep neural network traffic prediction is proposed. Firstly, a network total penalty model based on computing, memory and bandwidth resource allocation is established. Secondly, in order to predict the future traffic situation, a prediction algorithm based on ensemble deep neural network is proposed. Then the result of prediction are converted to perception of the physical network resource usage and resource requirements of slice in future according to the different types of traffic. Finally, in order to as large as possible to reduce operators punishment according to the result of perception, Virtual Network Functions (VNF) and virtual links are migrated to physical nodes and links that meet resource limits through dynamic slice adjustment and migration policies. The simulation results show that the proposed algorithm improves effectively the efficiency of slice migration and utilization of network resources.
In order to solve the problem that slice migration lags behind by lacking awareness of physical network resources in 5G Network Slice (NS) scenarios, a Dynamic Slice Adjustment and Migration (DSAM) algorithm based on ensemble deep neural network traffic prediction is proposed. Firstly, a network total penalty model based on computing, memory and bandwidth resource allocation is established. Secondly, in order to predict the future traffic situation, a prediction algorithm based on ensemble deep neural network is proposed. Then the result of prediction are converted to perception of the physical network resource usage and resource requirements of slice in future according to the different types of traffic. Finally, in order to as large as possible to reduce operators punishment according to the result of perception, Virtual Network Functions (VNF) and virtual links are migrated to physical nodes and links that meet resource limits through dynamic slice adjustment and migration policies. The simulation results show that the proposed algorithm improves effectively the efficiency of slice migration and utilization of network resources.
2023, 45(3): 1083-1093.
doi: 10.11999/JEIT211450
Abstract:
The coexistence performance of LTE-licensed Assisted Access (LAA) and WiFi networks has been extensively investigated. However, these works ignore capture effect, which is the phenomenon that the strongest signal may still be successfully received when more than two signals are transmitted simultaneously on the same channel, and which may occur more frequently in the coexistence scenario than in the pure WiFi network. Therefore, in this paper, the coexistence performance of LAA and WiFi networks with the capture effect is deeply investigated. More specifically, a capture model for more than two signals is firstly proposed in the coexistence scenario, and the capture probability is derived. Then the LAA access scheme is modeled as a new two-dimensional discrete Markov model with the capture effect, where the decrease of the backoff counter depends not only on the idle time slots but also on the time slots in which capture effect occurs. Finally, the expressions for coexistence performance are derived. A large number of simulation and numerical results verify the validity of the proposed Markov chain and capture model. The results prove the necessity of considering the capture effect in coexistence performance evaluation.
The coexistence performance of LTE-licensed Assisted Access (LAA) and WiFi networks has been extensively investigated. However, these works ignore capture effect, which is the phenomenon that the strongest signal may still be successfully received when more than two signals are transmitted simultaneously on the same channel, and which may occur more frequently in the coexistence scenario than in the pure WiFi network. Therefore, in this paper, the coexistence performance of LAA and WiFi networks with the capture effect is deeply investigated. More specifically, a capture model for more than two signals is firstly proposed in the coexistence scenario, and the capture probability is derived. Then the LAA access scheme is modeled as a new two-dimensional discrete Markov model with the capture effect, where the decrease of the backoff counter depends not only on the idle time slots but also on the time slots in which capture effect occurs. Finally, the expressions for coexistence performance are derived. A large number of simulation and numerical results verify the validity of the proposed Markov chain and capture model. The results prove the necessity of considering the capture effect in coexistence performance evaluation.
2023, 45(3): 1094-1101.
doi: 10.11999/JEIT211620
Abstract:
Nowadays, the existing vehicular applications have more stringent requirements for delay. Vehicular Edge Computing (VEC) is able to take advantage of network edges devices, such as Road Side Unit (RSU), for collaborative processing, which can effectively reduce the latency. Most existing studies assume that RSU has the sufficient computing resources to provide the unlimited services. But in fact, its computing resources will be limited with the increase of the number of processing tasks, which will restrict the delay sensitive vehicular applications. To solve this problem, a multi-task partial offloading scheme in vehicular edge computing is proposed in this paper. To minimize the total task processing delay, the remaining available computing resources of adjacent vehicles is considered under the condition of making full use of RSU computing resources in this scheme. Firstly, under the constrains of delay and resource, the optimal offloading decision variable ratio of local, RSU and adjacent vehicle for each task are allocated. Secondly, in order to minimize processing delay, the appropriate spare vehicle is selected in one-hop range as adjacent vehicles to process part of the task. Simulation results show the scheme proposed can reduce the delay better compared with other schemes.
Nowadays, the existing vehicular applications have more stringent requirements for delay. Vehicular Edge Computing (VEC) is able to take advantage of network edges devices, such as Road Side Unit (RSU), for collaborative processing, which can effectively reduce the latency. Most existing studies assume that RSU has the sufficient computing resources to provide the unlimited services. But in fact, its computing resources will be limited with the increase of the number of processing tasks, which will restrict the delay sensitive vehicular applications. To solve this problem, a multi-task partial offloading scheme in vehicular edge computing is proposed in this paper. To minimize the total task processing delay, the remaining available computing resources of adjacent vehicles is considered under the condition of making full use of RSU computing resources in this scheme. Firstly, under the constrains of delay and resource, the optimal offloading decision variable ratio of local, RSU and adjacent vehicle for each task are allocated. Secondly, in order to minimize processing delay, the appropriate spare vehicle is selected in one-hop range as adjacent vehicles to process part of the task. Simulation results show the scheme proposed can reduce the delay better compared with other schemes.
2023, 45(3): 1102-1110.
doi: 10.11999/JEIT220084
Abstract:
To resolve the problems of high computational complexity in strong noise environment, infeasibility of gaining large number of labeled samples and low detection probability, an Image Denoising and Classification driven Spectrum Sensing (IDCSS) method is proposed. Firstly, time-frequency transformation is employed to convert radio numerical signals into images. Then, as received signals of cognitive users and noise are highly correlated under strong noise environments, a novel Generative Adversarial Network (GAN) is designed to enhance the number and quality of samples of cognitive user signals. In the generator, residual-long-short-term memory network is designed to replace U-Net skip connection, realizing denoising and multi-scale features extraction. Loss function based on entropy is designed to optimize robustness to noise. A multi-dimensional discriminator is designed to enhance the quality of the generated image and retain the image details of the low signal-to-noise ratio cognitive user signals. Finally, the generated high-quality samples are used as labeled data, and the real samples combine to train the classifier to realize the recognition and classification of the spectrum occupancy state. Simulation results show that the proposed algorithm has better detection performance by comparing it with the state-of-the-art methods.
To resolve the problems of high computational complexity in strong noise environment, infeasibility of gaining large number of labeled samples and low detection probability, an Image Denoising and Classification driven Spectrum Sensing (IDCSS) method is proposed. Firstly, time-frequency transformation is employed to convert radio numerical signals into images. Then, as received signals of cognitive users and noise are highly correlated under strong noise environments, a novel Generative Adversarial Network (GAN) is designed to enhance the number and quality of samples of cognitive user signals. In the generator, residual-long-short-term memory network is designed to replace U-Net skip connection, realizing denoising and multi-scale features extraction. Loss function based on entropy is designed to optimize robustness to noise. A multi-dimensional discriminator is designed to enhance the quality of the generated image and retain the image details of the low signal-to-noise ratio cognitive user signals. Finally, the generated high-quality samples are used as labeled data, and the real samples combine to train the classifier to realize the recognition and classification of the spectrum occupancy state. Simulation results show that the proposed algorithm has better detection performance by comparing it with the state-of-the-art methods.
2023, 45(3): 1111-1120.
doi: 10.11999/JEIT211585
Abstract:
Li-ion Batteries (LiBs) have time-varying, dynamic, and nonlinear characteristics in application, as well as the capacity regeneration phenomenon, leading to inaccurate prediction of the Remaining Useful Life (RUL) of LiBs by the traditional models. This paper combines the Variational Modal Decomposition (VMD) method with Gaussian Process Regression (GPR) and Dynamic Adaptive Immune Particle Swarm Optimization (DAIPSO) to build a RUL prediction model. Firstly, the Health Indicator is extracted by using the time interval of equal discharging voltage difference analysis method, decomposing Health Indicator by using VMD to mine the internal information of the data and reduce the data complexity. For different components, the GPR prediction model is established using different covariance functions, which can effectively capture the long-term declining trend and short-term regeneration phenomenon. The GPR model is optimized using the DAIPSO algorithm to achieve the optimization of the hyperparameters of the kernel function, which establishes a more accurate degradation relationship model to achieve an accurate prediction of RUL, and uncertainty characterization. Finally, NASA battery data is used for verification. The offline prediction results show that the proposed method has high prediction accuracy and generalization adaptability.
Li-ion Batteries (LiBs) have time-varying, dynamic, and nonlinear characteristics in application, as well as the capacity regeneration phenomenon, leading to inaccurate prediction of the Remaining Useful Life (RUL) of LiBs by the traditional models. This paper combines the Variational Modal Decomposition (VMD) method with Gaussian Process Regression (GPR) and Dynamic Adaptive Immune Particle Swarm Optimization (DAIPSO) to build a RUL prediction model. Firstly, the Health Indicator is extracted by using the time interval of equal discharging voltage difference analysis method, decomposing Health Indicator by using VMD to mine the internal information of the data and reduce the data complexity. For different components, the GPR prediction model is established using different covariance functions, which can effectively capture the long-term declining trend and short-term regeneration phenomenon. The GPR model is optimized using the DAIPSO algorithm to achieve the optimization of the hyperparameters of the kernel function, which establishes a more accurate degradation relationship model to achieve an accurate prediction of RUL, and uncertainty characterization. Finally, NASA battery data is used for verification. The offline prediction results show that the proposed method has high prediction accuracy and generalization adaptability.
2023, 45(3): 1121-1133.
doi: 10.11999/JEIT211576
Abstract:
Underwater Wireless Sensor Networks (UWSNs) are widely used in disaster warning, resource exploration and other fields. However, UWSNs are vulnerable to malicious attacks. Therefore, it is urgent to develop a security mechanism that can adapt to its characteristics, e.g. communication band width, propagation time extension, and severe spatio-temporal uncertainty. First, based on the analysis of the characteristics and security requirements of UWSNs, the security threats UWSNs face are discussed in this paper. Then, the security mechanisms of UWSNs are summarized, including encryption, authentication, trust management, intrusion detection, secure location, secure synchronization and secure routing. Finally, the challenges of lack of practical tests and relevant data sets in the security research of UWSNs are discussed, as well as the future research direction of developing security mechanism based on network characteristics.
Underwater Wireless Sensor Networks (UWSNs) are widely used in disaster warning, resource exploration and other fields. However, UWSNs are vulnerable to malicious attacks. Therefore, it is urgent to develop a security mechanism that can adapt to its characteristics, e.g. communication band width, propagation time extension, and severe spatio-temporal uncertainty. First, based on the analysis of the characteristics and security requirements of UWSNs, the security threats UWSNs face are discussed in this paper. Then, the security mechanisms of UWSNs are summarized, including encryption, authentication, trust management, intrusion detection, secure location, secure synchronization and secure routing. Finally, the challenges of lack of practical tests and relevant data sets in the security research of UWSNs are discussed, as well as the future research direction of developing security mechanism based on network characteristics.
2023, 45(3): 1134-1146.
doi: 10.11999/JEIT211572
Abstract:
The technology of cloud storage is an effective way to solve the problems in high-capacity data storage, interaction and management. Using encrypted data in cloud servers is an important means to protect the privacy and security of user data in remote servers. Searchable encryption technology can improve the system availability on the premise of ensuring the security of user data. For its search efficiency, the Symmetric Searchable Encryption (SSE) has become a hot research topic. In general, the related research can be summarized into three aspects: the system model, efficiency and security, and usability. Firstly, the typical models of Symmetric Searchable Encryption(SSE) system are introduced. Then, common methods for search efficiency optimisation and security analysis are analysed in depth in this paper. Finally, from the aspects of scene adaptability, sentence expression ability and query result optimization, the research on scheme usability is combed and the hot spots and difficulties of the current research are summarized. On this basis, the possible research hotspots in the future are further analyzed.
The technology of cloud storage is an effective way to solve the problems in high-capacity data storage, interaction and management. Using encrypted data in cloud servers is an important means to protect the privacy and security of user data in remote servers. Searchable encryption technology can improve the system availability on the premise of ensuring the security of user data. For its search efficiency, the Symmetric Searchable Encryption (SSE) has become a hot research topic. In general, the related research can be summarized into three aspects: the system model, efficiency and security, and usability. Firstly, the typical models of Symmetric Searchable Encryption(SSE) system are introduced. Then, common methods for search efficiency optimisation and security analysis are analysed in depth in this paper. Finally, from the aspects of scene adaptability, sentence expression ability and query result optimization, the research on scheme usability is combed and the hot spots and difficulties of the current research are summarized. On this basis, the possible research hotspots in the future are further analyzed.
