Email alert
2021 Vol. 43, No. 7
Display Method:
2021, (7): 1-4.
Abstract:
2021, 43(7): 1803-1815.
doi: 10.11999/JEIT200487
Abstract:
Lithium-ion battery has been widely used in electric vehicle industry because of its long cycle life, high energy density, low self-discharge rate and low environmental pollution. The Battery Management System (BMS) in electric vehicles can maintain and monitor the battery status to ensure the security and reliability of the battery. The battery’s State of Charge (SoC) represents the remaining power in the battery and is one of the important parameters of the BMS. Real-time and accurate SoC estimation can extend battery life and ensure driving safety. However, Lithium-ion battery is indeed a highly complex nonlinear time-varying system, many unknown factors, such as battery life, ambient temperature, battery self-discharge and so on, will affect the estimation accuracy, which greatly increases the difficulty of estimation. To meet the requirements of accurate, rapid and real-time SoC estimation for Lithium-ion batteries under different conditions, further research and improvement of SoC estimation algorithm are needed. In recent years, some review literature on the Lithium-ion battery SoC estimation has been published. However, the existing literature has not summarized the estimation methods comprehensively and lack of process description. In this paper, the working principle of Lithium-ion battery is firstly introduced, as well as the factors affecting the SoC estimation of battery. Secondly, summary and analysis of the battery SoC estimation methods are conducted by the latest research achievement. According to the different characteristics of various algorithms, the battery SoC estimation can be divided into five categories: look-up table method, ampere hour integration method, model-based method, data-driven method and hybrid method. The main characteristics of various estimation methods are explained and the advantages and disadvantages of models or algorithms are comprehensively compared and discussed. Finally, the future development direction of SoC estimation methods for Lithium-ion batteries in electric vehicles is prospected.
Lithium-ion battery has been widely used in electric vehicle industry because of its long cycle life, high energy density, low self-discharge rate and low environmental pollution. The Battery Management System (BMS) in electric vehicles can maintain and monitor the battery status to ensure the security and reliability of the battery. The battery’s State of Charge (SoC) represents the remaining power in the battery and is one of the important parameters of the BMS. Real-time and accurate SoC estimation can extend battery life and ensure driving safety. However, Lithium-ion battery is indeed a highly complex nonlinear time-varying system, many unknown factors, such as battery life, ambient temperature, battery self-discharge and so on, will affect the estimation accuracy, which greatly increases the difficulty of estimation. To meet the requirements of accurate, rapid and real-time SoC estimation for Lithium-ion batteries under different conditions, further research and improvement of SoC estimation algorithm are needed. In recent years, some review literature on the Lithium-ion battery SoC estimation has been published. However, the existing literature has not summarized the estimation methods comprehensively and lack of process description. In this paper, the working principle of Lithium-ion battery is firstly introduced, as well as the factors affecting the SoC estimation of battery. Secondly, summary and analysis of the battery SoC estimation methods are conducted by the latest research achievement. According to the different characteristics of various algorithms, the battery SoC estimation can be divided into five categories: look-up table method, ampere hour integration method, model-based method, data-driven method and hybrid method. The main characteristics of various estimation methods are explained and the advantages and disadvantages of models or algorithms are comprehensively compared and discussed. Finally, the future development direction of SoC estimation methods for Lithium-ion batteries in electric vehicles is prospected.
2021, 43(7): 1816-1820.
doi: 10.11999/JEIT200460
Abstract:
For the speech impulse noise caused by the speckle effect in the laser vibration measurement system, a kurtosis detection algorithm based on the fourth-order cumulant is studied, and the mathematical iterative formulas for the kurtosis coefficient and the normalized kurtosis coefficient are deduced, and a method for determining the dynamic threshold of kurtosis coefficient is proposed in FPGA. The simulation results based on experimental data show that the algorithm reduces the calculation by about 25% compared with the theoretical formula and saves hardware resources, and has a more sensitive detection performance for lower amplitude impulse noise.
For the speech impulse noise caused by the speckle effect in the laser vibration measurement system, a kurtosis detection algorithm based on the fourth-order cumulant is studied, and the mathematical iterative formulas for the kurtosis coefficient and the normalized kurtosis coefficient are deduced, and a method for determining the dynamic threshold of kurtosis coefficient is proposed in FPGA. The simulation results based on experimental data show that the algorithm reduces the calculation by about 25% compared with the theoretical formula and saves hardware resources, and has a more sensitive detection performance for lower amplitude impulse noise.
2021, 43(7): 1821-1827.
doi: 10.11999/JEIT200876
Abstract:
The speed of existing signature and verification architecture is difficult to meet the requirement of the specific applications domain, to solve this problem a high-performance hardware architecture of Ed25519 algorithm is developed. The scalar multiplication algorithm is implemented by using the window method with 2 bit width to reduce the total cycle numbers of the algorithm significantly. By optimizing the order of operations of point addition and point doubling, the hardware utilization rate of multiplier is improved. The module multiplication is realized by using fast module reduction with low computational complexity, thus the overall operation speed is improved. The modular L algorithm based on Barrett reduction is proposed to reuse the fast modular reduction in scalar multiplications. By optimizing the modular power computation in the decompression process, the steps are simplified and the modular multiplication can be reused. Under the TSMC 55 nm CMOS process, the area of the proposed hardware architecture is 7.46×105 equivalent gate, and the maximum frequency is up to 360 MHz. It can perform 9.06×104 key generations, 8.82×104 signatures and 3.99×104 verifications per second.
The speed of existing signature and verification architecture is difficult to meet the requirement of the specific applications domain, to solve this problem a high-performance hardware architecture of Ed25519 algorithm is developed. The scalar multiplication algorithm is implemented by using the window method with 2 bit width to reduce the total cycle numbers of the algorithm significantly. By optimizing the order of operations of point addition and point doubling, the hardware utilization rate of multiplier is improved. The module multiplication is realized by using fast module reduction with low computational complexity, thus the overall operation speed is improved. The modular L algorithm based on Barrett reduction is proposed to reuse the fast modular reduction in scalar multiplications. By optimizing the modular power computation in the decompression process, the steps are simplified and the modular multiplication can be reused. Under the TSMC 55 nm CMOS process, the area of the proposed hardware architecture is 7.46×105 equivalent gate, and the maximum frequency is up to 360 MHz. It can perform 9.06×104 key generations, 8.82×104 signatures and 3.99×104 verifications per second.
2021, 43(7): 1828-1833.
doi: 10.11999/JEIT200202
Abstract:
To solve the problem that most existing image recognition systems are implemented in software which can not utilize the parallel computing power of neural networks, this paper proposes a FPGA image recognition system based on improved RBF neural network hardware. The multiplication operation in the neural networks is complex and inconvenient for hardware implementation. Furthermore, a sort circuit based on bit comparison is designed to solve the problem of fast sorting of a large number of data. Then, a multi-target image recognition application system is developed. The feature extraction part in the developed system is implemented by FPGA, and the image recognition part is implemented by ASIC circuit. The experimental results show that the average recognition time of the improved RBF neural network algorithm proposed is 50% shorter than that of LeNet-5, AlexNet and VGG16, and the time for the developed hardware system to recognize 10000 sample pictures is 165μs, which is reduced by about 60% compared with 426.6μs required by a DSP chip system.
To solve the problem that most existing image recognition systems are implemented in software which can not utilize the parallel computing power of neural networks, this paper proposes a FPGA image recognition system based on improved RBF neural network hardware. The multiplication operation in the neural networks is complex and inconvenient for hardware implementation. Furthermore, a sort circuit based on bit comparison is designed to solve the problem of fast sorting of a large number of data. Then, a multi-target image recognition application system is developed. The feature extraction part in the developed system is implemented by FPGA, and the image recognition part is implemented by ASIC circuit. The experimental results show that the average recognition time of the improved RBF neural network algorithm proposed is 50% shorter than that of LeNet-5, AlexNet and VGG16, and the time for the developed hardware system to recognize 10000 sample pictures is 165μs, which is reduced by about 60% compared with 426.6μs required by a DSP chip system.
2021, 43(7): 1834-1840.
doi: 10.11999/JEIT200513
Abstract:
By analyzing the circuit structure of Differential Pass-transistor Precharge Logic (DP2L), it is found that the circuit can not achieve the complete constant power consumption, and there is still a risk of being attacked by power attack. To solve this problem, the circuit structure of DP2L is improved by this paper, and the circuits before and after the improvement are simulated using Hspice. The experimental results show that the improved DP2L circuit structure has better characteristics of constant power consumption and can better meet the design requirements of the logic circuit.
By analyzing the circuit structure of Differential Pass-transistor Precharge Logic (DP2L), it is found that the circuit can not achieve the complete constant power consumption, and there is still a risk of being attacked by power attack. To solve this problem, the circuit structure of DP2L is improved by this paper, and the circuits before and after the improvement are simulated using Hspice. The experimental results show that the improved DP2L circuit structure has better characteristics of constant power consumption and can better meet the design requirements of the logic circuit.
2021, 43(7): 1841-1848.
doi: 10.11999/JET200110
Abstract:
In order to realize on-demand physical resource allocation in network function virtualization platform, an adaptive virtualized network function scaling method is proposed. The proposed method first use long short term memory network to realize traffic forecasting. Then combining with the forecasting result, a forward neural network-based approach is designed to predict resource demand of requested virtualized network function. Finally, according to the result of resource demand prediction, a dynamic encoding genetic algorithm is proposed to realize dynamic deployment of virtualized network function instances. The experiment results show that compared with existing scaling methods, the proposed scaling method can reduce the negative impact of inaccurate traffic forecasting, decrease the relative error of resource demand prediction as well as the total number of servers occupied by requested virtualized network function instances.
In order to realize on-demand physical resource allocation in network function virtualization platform, an adaptive virtualized network function scaling method is proposed. The proposed method first use long short term memory network to realize traffic forecasting. Then combining with the forecasting result, a forward neural network-based approach is designed to predict resource demand of requested virtualized network function. Finally, according to the result of resource demand prediction, a dynamic encoding genetic algorithm is proposed to realize dynamic deployment of virtualized network function instances. The experiment results show that compared with existing scaling methods, the proposed scaling method can reduce the negative impact of inaccurate traffic forecasting, decrease the relative error of resource demand prediction as well as the total number of servers occupied by requested virtualized network function instances.
2021, 43(7): 1849-1855.
doi: 10.11999/JEIT200519
Abstract:
According to the characteristics and propagation law of zero-day virus, the propagation model and stability of zero-day virus are studied. Firstly, the mechanism of zero-day virus transmission is analyzed. Based on the Susceptible-Infected-Removed-Susceptible(SIRS) virus transmission model, the node of infection state is redefined, the node of execution state and the node of damage state are introduced, and the zero-day virus transmission Susceptible - Initial-state-of infection- Zero-day - Damaged – Recovery (SIZDR) dynamic model is established. Secondly, the local stability of the system equilibrium point, the basic regeneration number and its influence on the scale of virus transmission are analyzed by using Rous stability criterion. Finally, the local stability of the model is verified by simulation, and the influence of node infection rate, node degree and node damage rate on zero-day virus transmission is analyzed. Theoretical analysis and simulation results show that the proposed model can objectively reflect the law of zero-day virus transmission, and the magnitude of zero-day virus spread is positively correlated with node degree and node infection rate, and negatively correlated with node damage rate. Targeted prevention and control of known viruses can effectively improve the defense effect against zero-day viruses.
According to the characteristics and propagation law of zero-day virus, the propagation model and stability of zero-day virus are studied. Firstly, the mechanism of zero-day virus transmission is analyzed. Based on the Susceptible-Infected-Removed-Susceptible(SIRS) virus transmission model, the node of infection state is redefined, the node of execution state and the node of damage state are introduced, and the zero-day virus transmission Susceptible - Initial-state-of infection- Zero-day - Damaged – Recovery (SIZDR) dynamic model is established. Secondly, the local stability of the system equilibrium point, the basic regeneration number and its influence on the scale of virus transmission are analyzed by using Rous stability criterion. Finally, the local stability of the model is verified by simulation, and the influence of node infection rate, node degree and node damage rate on zero-day virus transmission is analyzed. Theoretical analysis and simulation results show that the proposed model can objectively reflect the law of zero-day virus transmission, and the magnitude of zero-day virus spread is positively correlated with node degree and node infection rate, and negatively correlated with node damage rate. Targeted prevention and control of known viruses can effectively improve the defense effect against zero-day viruses.
2021, 43(7): 1856-1863.
doi: 10.11999/JEIT200399
Abstract:
In the storage and deletion service provided by cloud computing, due to the separation of outsourced data ownership and management, the cloud server may not follow the user’s request to delete the corresponding data, and the outsourced data can be easily exposed to unauthorized users due to the widely-adopted logical deletion. Therefore, an efficient and secure cloud data storage and deletion scheme is proposed. Firstly, an attribute-based encryption based on ciphertext policy is constructed based on elliptic curves to achieve fine-grained access control. Secondly, publicly verifiable data deletion is realized by using blockchain. The proposed scheme has the characteristics of responsibility traceability, two-party deletion and verifiability. Theoretical analysis and experimental results show that the presented scheme has more desirable security and performance, and can meet the needs of cloud data sharing and secure deletion.
In the storage and deletion service provided by cloud computing, due to the separation of outsourced data ownership and management, the cloud server may not follow the user’s request to delete the corresponding data, and the outsourced data can be easily exposed to unauthorized users due to the widely-adopted logical deletion. Therefore, an efficient and secure cloud data storage and deletion scheme is proposed. Firstly, an attribute-based encryption based on ciphertext policy is constructed based on elliptic curves to achieve fine-grained access control. Secondly, publicly verifiable data deletion is realized by using blockchain. The proposed scheme has the characteristics of responsibility traceability, two-party deletion and verifiability. Theoretical analysis and experimental results show that the presented scheme has more desirable security and performance, and can meet the needs of cloud data sharing and secure deletion.
2021, 43(7): 1864-1871.
doi: 10.11999/JEIT200124
Abstract:
To solve the problem of data security and privacy preservation brought by the centralization of cloud storage, a cloud-assisted Ciphertext Policy Attribute Based Eencryption(CP-ABE) data sharing encryption scheme based on BlockChain(BC) is proposed. In this scheme, the symmetric key of the encrypted file is encrypted by attribute-based encryption and the encrypted file is uploaded to cloud server for realizing the data security and fine-grained access control. Searchable encryption technology is adopted to encrypt the keyword, and the keyword ciphertext is uploaded to the BlockChain. Keyword search is executed by the BlockChain to ensure the security of keyword ciphertext, which effectively solves the security problems existing in the cloud storage and sharing system. This scheme can satisfy the indiscernibility, trap indiscernibility and series resistance under the selective plaintext attack. Finally, the effectiveness of the scheme is verified by performance evaluation.
To solve the problem of data security and privacy preservation brought by the centralization of cloud storage, a cloud-assisted Ciphertext Policy Attribute Based Eencryption(CP-ABE) data sharing encryption scheme based on BlockChain(BC) is proposed. In this scheme, the symmetric key of the encrypted file is encrypted by attribute-based encryption and the encrypted file is uploaded to cloud server for realizing the data security and fine-grained access control. Searchable encryption technology is adopted to encrypt the keyword, and the keyword ciphertext is uploaded to the BlockChain. Keyword search is executed by the BlockChain to ensure the security of keyword ciphertext, which effectively solves the security problems existing in the cloud storage and sharing system. This scheme can satisfy the indiscernibility, trap indiscernibility and series resistance under the selective plaintext attack. Finally, the effectiveness of the scheme is verified by performance evaluation.
2021, 43(7): 1872-1878.
doi: 10.11999/JEIT200307
Abstract:
To cope with the issue of locating multi-target in mitigating satellite interference, a localization method is proposed based on Compressive Sensing (CS). The sources of satellite interference can be identified by using Received Signal Strength (RSS) measurement only, relying on the spatial sparsity of the target source and the fact that multi-beam antenna has different gain at the position of interference. The conclusions show that positioning performance is related to node distribution, target number, coverage radius and decision threshold. Furthermore, over the Primal-Dual Interior Point (PDIP) algorithm, the simulation result represents that the target number is four under certain conditions, and the position accuracy is closed to 7.7 km with SNR of 20 dB. In addition, the study result also confirms that the proposed algorithm is better than the classic methods of Rotating Interferometer (RI) and Direction Of Arrival (DOA) estimation
To cope with the issue of locating multi-target in mitigating satellite interference, a localization method is proposed based on Compressive Sensing (CS). The sources of satellite interference can be identified by using Received Signal Strength (RSS) measurement only, relying on the spatial sparsity of the target source and the fact that multi-beam antenna has different gain at the position of interference. The conclusions show that positioning performance is related to node distribution, target number, coverage radius and decision threshold. Furthermore, over the Primal-Dual Interior Point (PDIP) algorithm, the simulation result represents that the target number is four under certain conditions, and the position accuracy is closed to 7.7 km with SNR of 20 dB. In addition, the study result also confirms that the proposed algorithm is better than the classic methods of Rotating Interferometer (RI) and Direction Of Arrival (DOA) estimation
2021, 43(7): 1879-1885.
doi: 10.11999/JEIT200497
Abstract:
Discontinuous Reception (DRX) is power-saving mechanism in 5G unlicensed band deployments. Legacy DRX scheme for licensed band does not work well in unlicensed band. Wakeup window size is fixed and can not adjust with channel busy level. Guaranteed transmission delay is at the cost of more power consumption. A novel DRX scheme of is proposed for the 5th Generation New Radio-Unlicensed (5G NR-U) standalone scenario. In the novel scheme, NR-U equipment runs energy detection continuously during its wakeup period to get channel state: busy or free and adjusts its wakeup window size adaptively according to energy detection results. Comparing to legacy method with wakeup window fixed, it is demonstrated by analysis and simulation that the novel method saves more power than traditional method while both of them meet the average delay requirement. In the analysis scenario of this paper, novel method saves 11% power more than legacy method.
Discontinuous Reception (DRX) is power-saving mechanism in 5G unlicensed band deployments. Legacy DRX scheme for licensed band does not work well in unlicensed band. Wakeup window size is fixed and can not adjust with channel busy level. Guaranteed transmission delay is at the cost of more power consumption. A novel DRX scheme of is proposed for the 5th Generation New Radio-Unlicensed (5G NR-U) standalone scenario. In the novel scheme, NR-U equipment runs energy detection continuously during its wakeup period to get channel state: busy or free and adjusts its wakeup window size adaptively according to energy detection results. Comparing to legacy method with wakeup window fixed, it is demonstrated by analysis and simulation that the novel method saves more power than traditional method while both of them meet the average delay requirement. In the analysis scenario of this paper, novel method saves 11% power more than legacy method.
2021, 43(7): 1886-1892.
doi: 10.11999/JEIT200388
Abstract:
The energy-efficient power allocation is studied in the uplink of the multi-carrier Ultra-Dense Networks (UDN). Based on the non-cooperative game theory, a distributed anti-interference power allocation scheme is proposed so that each cell can independently optimize energy efficiency while suppressing the inter-cell interference. Due to the fact that the energy efficiency problem under the constrains of the Quality of Service(QoS)and the maximum transmitter power is a challenging nonconvex problem and small cells suffer from the severe inter-cell interference, an accurate and low-complexity stair water-filling algorithm is proposed to solve the nonconvex problem in the best response process. Based on this algorithm, a multi-user anti-interference power allocation algorithm is proposed using interference channel gains. Simulation results and numerical analysis show that this algorithm can improve the system energy efficiency with no reduction in spectrum efficiency performance.
The energy-efficient power allocation is studied in the uplink of the multi-carrier Ultra-Dense Networks (UDN). Based on the non-cooperative game theory, a distributed anti-interference power allocation scheme is proposed so that each cell can independently optimize energy efficiency while suppressing the inter-cell interference. Due to the fact that the energy efficiency problem under the constrains of the Quality of Service(QoS)and the maximum transmitter power is a challenging nonconvex problem and small cells suffer from the severe inter-cell interference, an accurate and low-complexity stair water-filling algorithm is proposed to solve the nonconvex problem in the best response process. Based on this algorithm, a multi-user anti-interference power allocation algorithm is proposed using interference channel gains. Simulation results and numerical analysis show that this algorithm can improve the system energy efficiency with no reduction in spectrum efficiency performance.
2021, 43(7): 1893-1899.
doi: 10.11999/JEIT200209
Abstract:
In order to solve the problems of large estimation error, prolonged identification and high time complexity, which exist in tag quantity estimation algorithm in Radio Frequency IDentification (RFID) system, The Twice Labels Number Estimation algorithm based on Gaussian fitting and Chebyshev inequality (TLNEGC) is proposed. Firstly, a collision model is established based on the relationship between the collision factor and the collision time slot ratio, and a Gaussian estimation model is obtained by fitting the Gaussian function to the discrete data points. Afterward, the Gaussian estimation model is used to initially estimate the number of labels, and then according to the results of the initial estimation, judge whether a second estimation is required. The second estimation is performed by using Chebyshev's inequality to search the estimation interval twice to obtain the best estimate. The MATLAB simulation analysis indicates that the average estimation error and total time consumption of the TLNEGC algorithm are significantly lower than those of existing high-precision label estimation algorithms, and it also has lower time complexity and higher stability.
In order to solve the problems of large estimation error, prolonged identification and high time complexity, which exist in tag quantity estimation algorithm in Radio Frequency IDentification (RFID) system, The Twice Labels Number Estimation algorithm based on Gaussian fitting and Chebyshev inequality (TLNEGC) is proposed. Firstly, a collision model is established based on the relationship between the collision factor and the collision time slot ratio, and a Gaussian estimation model is obtained by fitting the Gaussian function to the discrete data points. Afterward, the Gaussian estimation model is used to initially estimate the number of labels, and then according to the results of the initial estimation, judge whether a second estimation is required. The second estimation is performed by using Chebyshev's inequality to search the estimation interval twice to obtain the best estimate. The MATLAB simulation analysis indicates that the average estimation error and total time consumption of the TLNEGC algorithm are significantly lower than those of existing high-precision label estimation algorithms, and it also has lower time complexity and higher stability.
2021, 43(7): 1900-1906.
doi: 10.11999/JEIT200136
Abstract:
According to the space theory for error correction, a Polar decoding algorithm for medium and short code lengths, based on the perturbation with a Convolution Neural Network (CNN), is presented to overcome the poor performance of the Successive Cancellation (SC) decoding algorithm and the high complexity of the Successive Cancellation List (SCL) decoding algorithm. For any receiving signals that failing to decode, a perturbation noise, generated through the CNN, is added to the receiving signal, and the likelihood information is then recalculated for further decoding. The simulation results show that the proposed algorithm has a gain of about 0.6 dB compared with the SC decoding algorithm, and an improvement of about 0.1 dB and a lower average complexity than that of SCL decoding algorithm when L=16.
According to the space theory for error correction, a Polar decoding algorithm for medium and short code lengths, based on the perturbation with a Convolution Neural Network (CNN), is presented to overcome the poor performance of the Successive Cancellation (SC) decoding algorithm and the high complexity of the Successive Cancellation List (SCL) decoding algorithm. For any receiving signals that failing to decode, a perturbation noise, generated through the CNN, is added to the receiving signal, and the likelihood information is then recalculated for further decoding. The simulation results show that the proposed algorithm has a gain of about 0.6 dB compared with the SC decoding algorithm, and an improvement of about 0.1 dB and a lower average complexity than that of SCL decoding algorithm when L=16.
2021, 43(7): 1907-1912.
doi: 10.11999/JEIT200374
Abstract:
The Peak to Average Power Ratio (PAPR) problem of OFDM system reduces the transmission efficiency of the system, increases the difficulty of demodulation, and makes the High Power Amplifier (HPA) to saturation. In order to solving the problem of nonlinear distortion caused by clipping and HPA, the proposed scheme uses preprocessing method based on Taylor series to deal with nonlinear interference so that to minimize the influence of signal distortion. At the transmitter, the PAPR of the signal is reduced by clipping, and the transmission process is considered as a whole based on the sparsity of clipping noise in the time domain. Finally, the total nonlinear distortion signal is recovered by using the Orthogonal Matching Pursuit (OMP) algorithm. The simulation results show that the proposed method can effectively suppress the interference signals of the system, reduce the nonlinear influence of useful signals, and verify the correctness of the scheme.
The Peak to Average Power Ratio (PAPR) problem of OFDM system reduces the transmission efficiency of the system, increases the difficulty of demodulation, and makes the High Power Amplifier (HPA) to saturation. In order to solving the problem of nonlinear distortion caused by clipping and HPA, the proposed scheme uses preprocessing method based on Taylor series to deal with nonlinear interference so that to minimize the influence of signal distortion. At the transmitter, the PAPR of the signal is reduced by clipping, and the transmission process is considered as a whole based on the sparsity of clipping noise in the time domain. Finally, the total nonlinear distortion signal is recovered by using the Orthogonal Matching Pursuit (OMP) algorithm. The simulation results show that the proposed method can effectively suppress the interference signals of the system, reduce the nonlinear influence of useful signals, and verify the correctness of the scheme.
2021, 43(7): 1913-1921.
doi: 10.11999/JEIT200238
Abstract:
In view of transporting various spatial data service types, an Advanced Orbiting Systems (AOS) virtual channel hybrid scheduling model is established based on AOS virtual channel multiplexing technology. In the model, a algorithm based on genetic-particle swarm ordering is proposed for asynchronous Virtual Channel (VC). Service priority, scheduling delay urgency and frame remaining urgency are the key constraints that affect the scheduling order of VC.The algorithm establishes the genetic-particle swarm fitness function model according to the constraints, and further enables the particles in the particle swarm to update the position according to the evolution operator of the genetic algorithm, thereby finding the optimal asynchronous VC scheduling sequence.At the same time, a dynamic weighted round-robin scheduling algorithm is designed for the synchronous VC, so that each synchronous VC occupies the physical channel according to the weighting factor and the allocated number of time slots. Simulation results show that the VC hybrid scheduling algorithm in this paper takes into account the priority of asynchronous data, the isochronism of synchronous data, and the urgency of VIP data. It has a smaller average scheduling delay and less frame remaining, so the algorithm meets the transmission requirements of different services.
In view of transporting various spatial data service types, an Advanced Orbiting Systems (AOS) virtual channel hybrid scheduling model is established based on AOS virtual channel multiplexing technology. In the model, a algorithm based on genetic-particle swarm ordering is proposed for asynchronous Virtual Channel (VC). Service priority, scheduling delay urgency and frame remaining urgency are the key constraints that affect the scheduling order of VC.The algorithm establishes the genetic-particle swarm fitness function model according to the constraints, and further enables the particles in the particle swarm to update the position according to the evolution operator of the genetic algorithm, thereby finding the optimal asynchronous VC scheduling sequence.At the same time, a dynamic weighted round-robin scheduling algorithm is designed for the synchronous VC, so that each synchronous VC occupies the physical channel according to the weighting factor and the allocated number of time slots. Simulation results show that the VC hybrid scheduling algorithm in this paper takes into account the priority of asynchronous data, the isochronism of synchronous data, and the urgency of VIP data. It has a smaller average scheduling delay and less frame remaining, so the algorithm meets the transmission requirements of different services.
2021, 43(7): 1922-1929.
doi: 10.11999/JEIT200318
Abstract:
In order to solve the limitation of low information transmission rate in Direct Sequence Spread Spectrum (DSSS), some techniques such as multi-band spread spectrum and index modulation appear. Because the additional information in this series of technologies is mapped by pseudo code transformation, the larger the modulation order is, the greater the complexity will be, and when the complexity of the system is limited, the modulation order will not be improved. In order to overcome the limitation of high-order information transmission rate, an UltraHigh Order Code Index Modulation (UHO-CIM) method with low complexity is proposed. The multi code set index is carried out by two-dimensional information grouping, so as to reduce the number of channels, and then the cyclic shift index is carried out by three-dimensional information grouping, so as to improve greatly the transmission rate without increasing the channels. Moreover, the relationship between shift channel and non-shift channel is used to remove the environmental impact. This method can not only effectively transmit multi-dimensional information, but also reduce greatly the complexity compared with the existing methods, and has obvious advantages in comprehensive performance. In addition, the existing methods are difficult to achieve high-order and ultra-high-order information transmission, while the proposed method can achieve ultra-high-order information transmission with modulation order greater than 15, which provides reference technology for the efficient spread spectrum communication applications.
In order to solve the limitation of low information transmission rate in Direct Sequence Spread Spectrum (DSSS), some techniques such as multi-band spread spectrum and index modulation appear. Because the additional information in this series of technologies is mapped by pseudo code transformation, the larger the modulation order is, the greater the complexity will be, and when the complexity of the system is limited, the modulation order will not be improved. In order to overcome the limitation of high-order information transmission rate, an UltraHigh Order Code Index Modulation (UHO-CIM) method with low complexity is proposed. The multi code set index is carried out by two-dimensional information grouping, so as to reduce the number of channels, and then the cyclic shift index is carried out by three-dimensional information grouping, so as to improve greatly the transmission rate without increasing the channels. Moreover, the relationship between shift channel and non-shift channel is used to remove the environmental impact. This method can not only effectively transmit multi-dimensional information, but also reduce greatly the complexity compared with the existing methods, and has obvious advantages in comprehensive performance. In addition, the existing methods are difficult to achieve high-order and ultra-high-order information transmission, while the proposed method can achieve ultra-high-order information transmission with modulation order greater than 15, which provides reference technology for the efficient spread spectrum communication applications.
2021, 43(7): 1930-1938.
doi: 10.11999/JEIT200165
Abstract:
In order to solve the shortcomings of high Bit Error Rate (BER) in traditional Correlation Delay Shift Keying (CDSK) chaotic communication system, a Noise Reduction Multi-Carrier Correlated Delay Shift Keying (NR-MC-CDSK) chaotic communication system based on Schmidt orthogonalization is proposed. At the transmitter, Schmidt orthogonalization algorithm is used to generate N completely orthogonal chaotic signals which are copied P times as reference signals. N information signals are added up for transmission in each group and multi-carrier is used to transmit MN user information per frame. At the receiver, the signal is demodulated by a matched filter, then the signal is averaged by a moving average filter to suppress the noise and demodulated by correlation. The BER formula of the system in Additive White Gaussian Noise (AWGN) channel and multipath Rayleigh fading channel are derived and simulated. The results show that the BER performance of the system is better than that of many multi-carrier chaotic communication systems and the data transmission rate is improved obviously compared with CDSK system. Theoretical basis for the future application of the system to practical communication systems is provided in the paper and shows strong feasibilities in future engineering application.
In order to solve the shortcomings of high Bit Error Rate (BER) in traditional Correlation Delay Shift Keying (CDSK) chaotic communication system, a Noise Reduction Multi-Carrier Correlated Delay Shift Keying (NR-MC-CDSK) chaotic communication system based on Schmidt orthogonalization is proposed. At the transmitter, Schmidt orthogonalization algorithm is used to generate N completely orthogonal chaotic signals which are copied P times as reference signals. N information signals are added up for transmission in each group and multi-carrier is used to transmit MN user information per frame. At the receiver, the signal is demodulated by a matched filter, then the signal is averaged by a moving average filter to suppress the noise and demodulated by correlation. The BER formula of the system in Additive White Gaussian Noise (AWGN) channel and multipath Rayleigh fading channel are derived and simulated. The results show that the BER performance of the system is better than that of many multi-carrier chaotic communication systems and the data transmission rate is improved obviously compared with CDSK system. Theoretical basis for the future application of the system to practical communication systems is provided in the paper and shows strong feasibilities in future engineering application.
2021, 43(7): 1939-1946.
doi: 10.11999/JEIT200299
Abstract:
Interrupted-Sampling Repeater Jamming (ISRJ) is an advanced radar coherent interference based on the principle of under-sampling method, and achieves a train of false targets. In this paper, a joint transmitted waveform and mismatched filter design method focus on suppressing the ISRJ is proposed. Firstly, the mathematical optimization model is established by minimizing the transmitted signal pulse compression integrated sidelobe level and jamming signal mismatched filter output integrated level under the constraint of constant modulus waveform. Secondly, the analytical solution of transmitted waveform and mismatched filter is obtained by simplifing the optimization problem. By utilizing cyclic iterative algorithm, unimodular waveform and mismatched filter are produced. Finally, simulations are presented to assess the waveform and filter performance and anti-ISRJ performance. Simulation results demonstrated that the ISRJ is suppressed by jointly designing the waveform and filter.
Interrupted-Sampling Repeater Jamming (ISRJ) is an advanced radar coherent interference based on the principle of under-sampling method, and achieves a train of false targets. In this paper, a joint transmitted waveform and mismatched filter design method focus on suppressing the ISRJ is proposed. Firstly, the mathematical optimization model is established by minimizing the transmitted signal pulse compression integrated sidelobe level and jamming signal mismatched filter output integrated level under the constraint of constant modulus waveform. Secondly, the analytical solution of transmitted waveform and mismatched filter is obtained by simplifing the optimization problem. By utilizing cyclic iterative algorithm, unimodular waveform and mismatched filter are produced. Finally, simulations are presented to assess the waveform and filter performance and anti-ISRJ performance. Simulation results demonstrated that the ISRJ is suppressed by jointly designing the waveform and filter.
2021, 43(7): 1947-1953.
doi: 10.11999/JEIT200250
Abstract:
To address track association problem of radar and Electronic Support Measurements (ESM) under complex conditions such as system biases and asynchronous track, an asynchronous anti-bias track association algorithm based on discrete degree of interval sequence is proposed. The discrete information measurement of mixed interval sequence is defined and the interval method of system biases is given. The association determination is performed by calculating the interval discrete degree and using classical assignment method. Compared with the traditional algorithm, the asynchronous track can be directly correlated without time-domain registration in the presence of system errors and it is not sensitive to noise distribution. The simulation results show that the algorithm has good robust performance and is not affected by the target moving position. The algorithm can be applied to the configuration of sensors at same or different sites.
To address track association problem of radar and Electronic Support Measurements (ESM) under complex conditions such as system biases and asynchronous track, an asynchronous anti-bias track association algorithm based on discrete degree of interval sequence is proposed. The discrete information measurement of mixed interval sequence is defined and the interval method of system biases is given. The association determination is performed by calculating the interval discrete degree and using classical assignment method. Compared with the traditional algorithm, the asynchronous track can be directly correlated without time-domain registration in the presence of system errors and it is not sensitive to noise distribution. The simulation results show that the algorithm has good robust performance and is not affected by the target moving position. The algorithm can be applied to the configuration of sensors at same or different sites.
2021, 43(7): 1954-1961.
doi: 10.11999/JEIT200474
Abstract:
In recent years, countries around the world have paid more and more attention to the development of wind power. The existence of wind farms may have a negative impact on the performance of air traffic control surveillance radars. Therefore, the research on the clutter suppression technology of wind farms is of great significance to improve the work performance of air traffic control surveillance radars and ensure the safety of air traffic. When the Morphological Component Analysis(MCA)algorithm is applied to the wind farm clutter suppression based on the difference of sparse characteristics for the signals, the calculation burden is lower and the performance is better. However, the clutter suppression performance of the MCA algorithm is affected when the spectral resolution is reduced due to the short Coherent Processing Interval(CPI)and the signal characteristics are not obvious. Therefore, the sparse reconstruction algorithm and the MCA algorithm are combined to suppress the clutter in the wind farm with a small number of coherent pulses. It is considered that the short CPI received echo data is the default of tail data on the basis of the longer CPI radar echo data, and then the sparse reconstruction algorithm is used to recover the default data, and the MCA algorithm is used to suppress wind farm clutter. The experimental results verify the effectiveness of the proposed method.
In recent years, countries around the world have paid more and more attention to the development of wind power. The existence of wind farms may have a negative impact on the performance of air traffic control surveillance radars. Therefore, the research on the clutter suppression technology of wind farms is of great significance to improve the work performance of air traffic control surveillance radars and ensure the safety of air traffic. When the Morphological Component Analysis(MCA)algorithm is applied to the wind farm clutter suppression based on the difference of sparse characteristics for the signals, the calculation burden is lower and the performance is better. However, the clutter suppression performance of the MCA algorithm is affected when the spectral resolution is reduced due to the short Coherent Processing Interval(CPI)and the signal characteristics are not obvious. Therefore, the sparse reconstruction algorithm and the MCA algorithm are combined to suppress the clutter in the wind farm with a small number of coherent pulses. It is considered that the short CPI received echo data is the default of tail data on the basis of the longer CPI radar echo data, and then the sparse reconstruction algorithm is used to recover the default data, and the MCA algorithm is used to suppress wind farm clutter. The experimental results verify the effectiveness of the proposed method.
2021, 43(7): 1962-1969.
doi: 10.11999/JEIT200498
Abstract:
A novel method Interacting Multiple Mode Multi-Sensor Multi-target Multi-Bernoulli (IMM-MS-MeMBer) filter to track multiple maneuvering targets in low detection probability scenario is proposed. At the prediction stage of the IMM-MS-MeMBer filter, model probability of the target is adaptively updated by utilizing the current measurement information, and then the mixed prediction of the target state is executed; At the update stage of the IMM-MS-MeMBer filter, the greedy multi-sensor measurement partitioning strategy is employed in measurement partition step, the posterior probability density of the target is updated by using the divided set of measurements and the IMM-MS-MeMBer filter; In addition, the IMM-MS-MeMBer filter utilizes the target angle and Doppler information to realize the simultaneous estimation of the position and speed of multiple maneuvering targets. Numerical experiments verify the superior performance of the IMM-MS-MeMBer filter.
A novel method Interacting Multiple Mode Multi-Sensor Multi-target Multi-Bernoulli (IMM-MS-MeMBer) filter to track multiple maneuvering targets in low detection probability scenario is proposed. At the prediction stage of the IMM-MS-MeMBer filter, model probability of the target is adaptively updated by utilizing the current measurement information, and then the mixed prediction of the target state is executed; At the update stage of the IMM-MS-MeMBer filter, the greedy multi-sensor measurement partitioning strategy is employed in measurement partition step, the posterior probability density of the target is updated by using the divided set of measurements and the IMM-MS-MeMBer filter; In addition, the IMM-MS-MeMBer filter utilizes the target angle and Doppler information to realize the simultaneous estimation of the position and speed of multiple maneuvering targets. Numerical experiments verify the superior performance of the IMM-MS-MeMBer filter.
2021, 43(7): 1970-1977.
doi: 10.11999/JEIT200529
Abstract:
In modern radar electronic battlefield, target detection and parameter estimation have great significance. Therefore, a high-speed multi-target parameter estimation method for Frequency Agile-Orthogonal Frequency Division Multiplexing (FA-OFDM) radar based on Random sampling consensus (Ransac) algorithm is proposed in this paper. Firstly, multiple narrowband OFDM subcarriers with random frequency hopping are simultaneously transmitted in each pulse of conventional frequency agile radar. The echo signals of all subcarriers in a single pulse are compressed, and then the high-resolution range of the target is synthesized by Iterative Adaptive Approach (IAA) algorithm. Furthermore, the echoes of each pulse are compressed and iterative adaptive spectrum estimated, and the high-resolution distance of different pulse time is obtained to form the observation data set. Then, according to the steps of the Ransac algorithm to estimate the signal parameter model, multiple time-distance lines are fitted, and then parameters of multiple high-speed moving targets are estimated at the same time. Finally, the influence of the Signal-to-Noise Ratio (SNR) on detection probability and the target velocity on relative error of estimation are analyzed, respectively. Simulations are provided to verify the effectiveness of the proposal.
In modern radar electronic battlefield, target detection and parameter estimation have great significance. Therefore, a high-speed multi-target parameter estimation method for Frequency Agile-Orthogonal Frequency Division Multiplexing (FA-OFDM) radar based on Random sampling consensus (Ransac) algorithm is proposed in this paper. Firstly, multiple narrowband OFDM subcarriers with random frequency hopping are simultaneously transmitted in each pulse of conventional frequency agile radar. The echo signals of all subcarriers in a single pulse are compressed, and then the high-resolution range of the target is synthesized by Iterative Adaptive Approach (IAA) algorithm. Furthermore, the echoes of each pulse are compressed and iterative adaptive spectrum estimated, and the high-resolution distance of different pulse time is obtained to form the observation data set. Then, according to the steps of the Ransac algorithm to estimate the signal parameter model, multiple time-distance lines are fitted, and then parameters of multiple high-speed moving targets are estimated at the same time. Finally, the influence of the Signal-to-Noise Ratio (SNR) on detection probability and the target velocity on relative error of estimation are analyzed, respectively. Simulations are provided to verify the effectiveness of the proposal.
2021, 43(7): 1978-1984.
doi: 10.11999/JEIT200143
Abstract:
For the underwater moving platforms, the near-range filtering of active sonar is seriously affected by reverberation disturbance. Usually, the real target echo will be masked by numerous reverberation highlights, which will greatly increase the false alarm rate of subsequent detection. Among adjacent periods of the bearing-time-record from array processing in some scenarios, this paper utilizes the potential coherent structure of reverberation, and then assumes that reverberation component satisfies the low-rank property. In addition, the relative motion may assume that target echoes of interest are irrelevant and sparse. Accordingly, the bearing-time-record can be decomposed as low-rank reverberation, sparse moving target echo and noise components. To suppress reverberation and enhance target echoes, the Accelerated Proximal Gradient(APG) and Fast Data Projection Method(FDPM) are proposed to realize offline and online decomposition, respectively. The experimental results validate the assumed models, and both approaches can effectively enhance target echoes.
For the underwater moving platforms, the near-range filtering of active sonar is seriously affected by reverberation disturbance. Usually, the real target echo will be masked by numerous reverberation highlights, which will greatly increase the false alarm rate of subsequent detection. Among adjacent periods of the bearing-time-record from array processing in some scenarios, this paper utilizes the potential coherent structure of reverberation, and then assumes that reverberation component satisfies the low-rank property. In addition, the relative motion may assume that target echoes of interest are irrelevant and sparse. Accordingly, the bearing-time-record can be decomposed as low-rank reverberation, sparse moving target echo and noise components. To suppress reverberation and enhance target echoes, the Accelerated Proximal Gradient(APG) and Fast Data Projection Method(FDPM) are proposed to realize offline and online decomposition, respectively. The experimental results validate the assumed models, and both approaches can effectively enhance target echoes.
2021, 43(7): 1985-1991.
doi: 10.11999/JEIT200447
Abstract:
Due to the scarcity of sea clutter data, the high cost and long period of obtaining sea clutter data greatly limit the research of sea clutter characteristics and the application of ocean remote sensing. The method of sea clutter data generation based on the Generative Adversarial Networks (GAN) is studied. By extending the traditional GAN framework, a one-dimensional sea clutter data generation and identification model is formed. Based on the radar measured sea clutter data set, the generation and identification model training in the adversarial network is carried out. The amplitude distribution characteristics and time and spatial correlation of the sea clutter data generated by the model are analyzed. Based on the measured data, it is verified that the method can generate more sea clutter data with more variety, and similar distribution to the real sea clutter data.
Due to the scarcity of sea clutter data, the high cost and long period of obtaining sea clutter data greatly limit the research of sea clutter characteristics and the application of ocean remote sensing. The method of sea clutter data generation based on the Generative Adversarial Networks (GAN) is studied. By extending the traditional GAN framework, a one-dimensional sea clutter data generation and identification model is formed. Based on the radar measured sea clutter data set, the generation and identification model training in the adversarial network is carried out. The amplitude distribution characteristics and time and spatial correlation of the sea clutter data generated by the model are analyzed. Based on the measured data, it is verified that the method can generate more sea clutter data with more variety, and similar distribution to the real sea clutter data.
2021, 43(7): 1992-1999.
doi: 10.11999/JEIT200456
Abstract:
The existence of high-squint angle and three-dimensional acceleration makes the motion error of maneuvering platform SAR have obvious two-dimensional spatial variability, which greatly increases the difficulty of imaging. A sparse autofocus method based on the estimation and compensation of two-dimensional spatial-variant motion error is proposed. Based on the Keystone transform and the frequency domain phase filtering method, a frequency-domain approximate observation operator is constructed to correct the spatial-variant imaging parameters. In the process of autofocus, firstly, a sparse autofocus model based on the frequency-domain approximate observation operator is constructed to focus the image roughly and estimate the non-spatial-variant motion error parameters, and the Iterative Shrinkage-Thresholding Algorithm (ISTA) is used to solve the constructed sparse autofocus model. Then, the precise phase error curves of multiple sub-regions are obtained by the sparse autofocus model and the least square method can be used to estimate the spatial-variant motion error parameters. Finally, the compensation of spatial-variant motion error is realized by correcting the approximate observation operator. The simulation results show the effectiveness of the proposed method.
The existence of high-squint angle and three-dimensional acceleration makes the motion error of maneuvering platform SAR have obvious two-dimensional spatial variability, which greatly increases the difficulty of imaging. A sparse autofocus method based on the estimation and compensation of two-dimensional spatial-variant motion error is proposed. Based on the Keystone transform and the frequency domain phase filtering method, a frequency-domain approximate observation operator is constructed to correct the spatial-variant imaging parameters. In the process of autofocus, firstly, a sparse autofocus model based on the frequency-domain approximate observation operator is constructed to focus the image roughly and estimate the non-spatial-variant motion error parameters, and the Iterative Shrinkage-Thresholding Algorithm (ISTA) is used to solve the constructed sparse autofocus model. Then, the precise phase error curves of multiple sub-regions are obtained by the sparse autofocus model and the least square method can be used to estimate the spatial-variant motion error parameters. Finally, the compensation of spatial-variant motion error is realized by correcting the approximate observation operator. The simulation results show the effectiveness of the proposed method.
2021, 43(7): 2000-2006.
doi: 10.11999/JEIT200442
Abstract:
It is one of the important approaches to improve the direction finding accuracy of interferometer to reduce the measurement error by averaging the signal phase difference between the channels in the interferometer. In this process, there are two methods: scalar accumulation and vector accumulation. In order to analyze on these two methods, after a brief introduction to the direction finding model of interferometer and the formation process of phase difference, the statistical characteristics of phase difference are deduced by using the signal vector method. Then a detailed contrast between scalar accumulation and vector accumulation of phase difference is made by using the probability distribution results derived. This not only reveals the threshold effect in scalar accumulation, but also the infinite approximation process of vector accumulation to the real value is theoretically proved. Finally, the validity and correctness of the theoretical analysis are verified by simulations. It provides an important theoretical guidance on the data processing of interferometer phase difference measurement for engineering application.
It is one of the important approaches to improve the direction finding accuracy of interferometer to reduce the measurement error by averaging the signal phase difference between the channels in the interferometer. In this process, there are two methods: scalar accumulation and vector accumulation. In order to analyze on these two methods, after a brief introduction to the direction finding model of interferometer and the formation process of phase difference, the statistical characteristics of phase difference are deduced by using the signal vector method. Then a detailed contrast between scalar accumulation and vector accumulation of phase difference is made by using the probability distribution results derived. This not only reveals the threshold effect in scalar accumulation, but also the infinite approximation process of vector accumulation to the real value is theoretically proved. Finally, the validity and correctness of the theoretical analysis are verified by simulations. It provides an important theoretical guidance on the data processing of interferometer phase difference measurement for engineering application.
2021, 43(7): 2007-2013.
doi: 10.11999/JEIT200180
Abstract:
To improve the description accuracy of the characteristics of microwave link induced by rain attenuation and expand the available parameters of microwave link signals, the 15 GHz, 18 GHz and 23 GHz microwave links and rain gauges deployed in Jiangyin area of Jiangsu Province are used to carry out synchronous comparative observation, and the rain attenuation relationship at three frequency bands are fitted. The relationship between the 13 features of the received signal level (including the average, median, 25% quantile, 75% quantile, standard deviation, maximum, minimum, etc.) and the rain/no-rain period and rainfall intensity is extracted and analyzed. The conclusions are as follows. There is an obvious negative correlation between the signal of microwave link and the rainfall intensity. There is a general good consistency between the fitted rain attenuation relationship and the ITU-R empirical rain attenuation relationship, but there are certain of differences in different frequencies; All 13 parameters have a certain probability of overlap in the rain period and no-rain period, which is the main reason why it is difficult to distinguish between rain and no-rain; The higher is the frequency, the more significant is the impact of rainfall on the signal change, the more conducive to the microwave link inversion of rainfall. The results provide an important basis for improving the discrimination of rain and no-rain, and determination of reference value and inversion of rain intensity.
To improve the description accuracy of the characteristics of microwave link induced by rain attenuation and expand the available parameters of microwave link signals, the 15 GHz, 18 GHz and 23 GHz microwave links and rain gauges deployed in Jiangyin area of Jiangsu Province are used to carry out synchronous comparative observation, and the rain attenuation relationship at three frequency bands are fitted. The relationship between the 13 features of the received signal level (including the average, median, 25% quantile, 75% quantile, standard deviation, maximum, minimum, etc.) and the rain/no-rain period and rainfall intensity is extracted and analyzed. The conclusions are as follows. There is an obvious negative correlation between the signal of microwave link and the rainfall intensity. There is a general good consistency between the fitted rain attenuation relationship and the ITU-R empirical rain attenuation relationship, but there are certain of differences in different frequencies; All 13 parameters have a certain probability of overlap in the rain period and no-rain period, which is the main reason why it is difficult to distinguish between rain and no-rain; The higher is the frequency, the more significant is the impact of rainfall on the signal change, the more conducive to the microwave link inversion of rainfall. The results provide an important basis for improving the discrimination of rain and no-rain, and determination of reference value and inversion of rain intensity.
2021, 43(7): 2014-2022.
doi: 10.11999/JEIT200357
Abstract:
In order to solve the problems of detection and state analysis of high-speed railway catenary droppers, this paper proposes a multi-scale detection method for dropper states based on Refinedet network and Hough transform. First, the positioning result of droppers through Refinedet network is obtained, and Hough transform is used to locate where the dropper line is; Then the surrounding area of the dropper line is extracted with a ralated twiddle factor. Those extracted areas, replacing the results of detection net, are fed into classification network for training the final dropper state analysis mode. Experiments show that the accuracy of dropper detection model is over 95.3%, and the dropper state analysis model can eliminate the impact of meaningless area pixels while accelerating training process, the final state analysis model achieves a high accuracy over 97.5%.
In order to solve the problems of detection and state analysis of high-speed railway catenary droppers, this paper proposes a multi-scale detection method for dropper states based on Refinedet network and Hough transform. First, the positioning result of droppers through Refinedet network is obtained, and Hough transform is used to locate where the dropper line is; Then the surrounding area of the dropper line is extracted with a ralated twiddle factor. Those extracted areas, replacing the results of detection net, are fed into classification network for training the final dropper state analysis mode. Experiments show that the accuracy of dropper detection model is over 95.3%, and the dropper state analysis model can eliminate the impact of meaningless area pixels while accelerating training process, the final state analysis model achieves a high accuracy over 97.5%.
2021, 43(7): 2023-2031.
doi: 10.11999/JEIT200414
Abstract:
Compared to H.264/AVC coding standard, High Efficiency Video Coding (HEVC) improves the compression efficiency, but the consequent disadvantage is the significant increase in encoding complexity by using the quad-tree partition. A Multi-Layer Feature Transfer Convolutional Neural Network (MLFT-CNN) for Coding Unit (CU) division and characterization vector prediction in HEVC intra coding mode is proposed, which greatly reduces the complexity of video coding. Firstly, a reduced-resolution feature extraction module incorporating CU partition structure information is proposed. Then, the channel attention mechanism is improved for a better texture expression performance of the feature. After that, the feature transfer mechanism is designed to use the feature division of high-depth coding unit to guide the division of low-depth coding unit. Finally, the target loss function represented by the segmented feature is established, and the end-to-end CU division represents the vector prediction network. The experimental results show that the proposed algorithm effectively reduces the encoding complexity of HEVC without affecting the video coding quality. Specifically, compared to the standard method, the encoding complexity on the standard test sequence is reduced by 70.96% on average.
Compared to H.264/AVC coding standard, High Efficiency Video Coding (HEVC) improves the compression efficiency, but the consequent disadvantage is the significant increase in encoding complexity by using the quad-tree partition. A Multi-Layer Feature Transfer Convolutional Neural Network (MLFT-CNN) for Coding Unit (CU) division and characterization vector prediction in HEVC intra coding mode is proposed, which greatly reduces the complexity of video coding. Firstly, a reduced-resolution feature extraction module incorporating CU partition structure information is proposed. Then, the channel attention mechanism is improved for a better texture expression performance of the feature. After that, the feature transfer mechanism is designed to use the feature division of high-depth coding unit to guide the division of low-depth coding unit. Finally, the target loss function represented by the segmented feature is established, and the end-to-end CU division represents the vector prediction network. The experimental results show that the proposed algorithm effectively reduces the encoding complexity of HEVC without affecting the video coding quality. Specifically, compared to the standard method, the encoding complexity on the standard test sequence is reduced by 70.96% on average.
2021, 43(7): 2032-2037.
doi: 10.11999/JEIT200413
Abstract:
The study of odor recognition based on ElectroEncephaloGram (EEG) signals has important application value to objectively evaluating olfactory function and diagnosing olfactory disorders. Because of the inconvenience caused by using too many EEG channels in practical application scenarios, it is particularly important to study how to choose EEG channels. In this paper, a new ReliefF-Pearson channel selection algorithm is proposed to solve the channel selection problem in the classification of olfactory EEG signals. The algorithm combines the weight idea of ReliefF and the correlation principle of Pearson coefficient to select EEG channels. Experimental results show that compared with the traditional ReliefF-based channel selection algorithm, the proposed algorithm could significantly reduce the number of channels used while ensuring a certain classification accuracy, and the result of channel selection does not depend on human experience and classifiers. In addition, the spatial distribution of the selected channels is consistent with the existing olfactory neurophysiological position, which further confirms the scientificity and effectiveness of this method. The proposed method provides new idea for the research of olfactory EEG channel selection.
The study of odor recognition based on ElectroEncephaloGram (EEG) signals has important application value to objectively evaluating olfactory function and diagnosing olfactory disorders. Because of the inconvenience caused by using too many EEG channels in practical application scenarios, it is particularly important to study how to choose EEG channels. In this paper, a new ReliefF-Pearson channel selection algorithm is proposed to solve the channel selection problem in the classification of olfactory EEG signals. The algorithm combines the weight idea of ReliefF and the correlation principle of Pearson coefficient to select EEG channels. Experimental results show that compared with the traditional ReliefF-based channel selection algorithm, the proposed algorithm could significantly reduce the number of channels used while ensuring a certain classification accuracy, and the result of channel selection does not depend on human experience and classifiers. In addition, the spatial distribution of the selected channels is consistent with the existing olfactory neurophysiological position, which further confirms the scientificity and effectiveness of this method. The proposed method provides new idea for the research of olfactory EEG channel selection.
2021, 43(7): 2038-2045.
doi: 10.11999/JEIT200164
Abstract:
To resist the geometric attack of watermarked images effectively, a robust watermarking algorithm based on Blob-Harris feature region combined with NonSubsampled Contourlet Transform (NSCT) and pseudo Zernike moment is proposed. First, the original image is extracted from its low-frequency image after two-layer NSCT. Then, Blob-Harris detection operator is used to extract the feature points of the low-frequency image. The feature regions are determined according to the feature scale of each feature point, and the stable non-overlapping feature areas are optimized and filtered out and zero padding around them to obtain square feature areas as watermark embedding areas. Finally, the Zernike moments of each square feature area are calculated, the watermarking information is embeded to quantized modulation regularized Zernike moments. The experimental results show that when the peak signal-to-noise ratio of the Lena reaches more than 40 dB, the algorithm has relatively strong robustness to conventional image processing, geometric attacks such as scaling, rotation, and shearing and combined attacks.
To resist the geometric attack of watermarked images effectively, a robust watermarking algorithm based on Blob-Harris feature region combined with NonSubsampled Contourlet Transform (NSCT) and pseudo Zernike moment is proposed. First, the original image is extracted from its low-frequency image after two-layer NSCT. Then, Blob-Harris detection operator is used to extract the feature points of the low-frequency image. The feature regions are determined according to the feature scale of each feature point, and the stable non-overlapping feature areas are optimized and filtered out and zero padding around them to obtain square feature areas as watermark embedding areas. Finally, the Zernike moments of each square feature area are calculated, the watermarking information is embeded to quantized modulation regularized Zernike moments. The experimental results show that when the peak signal-to-noise ratio of the Lena reaches more than 40 dB, the algorithm has relatively strong robustness to conventional image processing, geometric attacks such as scaling, rotation, and shearing and combined attacks.
2021, 43(7): 2046-2054.
doi: 10.11999/JEIT200389
Abstract:
For the problem that Bayesian variational inference with low convergence precision is easy to fall into local optimum during search process, a Bayesian variational inference algorithm based on Expectation-Maximization (EM) and Simulated Annealing (SA) is proposed. The influence of the initial prior on the final result and the optimization efficiency of the variational free energy in the process of variational inference can not be ignored. The double EM is introduced to construct the initial prior of the variational parameter to reduce the sensitivity of the initial prior. And the inverse temperature parameter is introducted to improve the free energy function, which makes the energy be effectively controlled in the optimization process. This paper uses convergence criterion theory to analyze the convergence of the algorithm. The proposed algorithm is used for experiments with an Gaussian mixture model and the experimental results show that the proposed algorithm has better convergence results.
For the problem that Bayesian variational inference with low convergence precision is easy to fall into local optimum during search process, a Bayesian variational inference algorithm based on Expectation-Maximization (EM) and Simulated Annealing (SA) is proposed. The influence of the initial prior on the final result and the optimization efficiency of the variational free energy in the process of variational inference can not be ignored. The double EM is introduced to construct the initial prior of the variational parameter to reduce the sensitivity of the initial prior. And the inverse temperature parameter is introducted to improve the free energy function, which makes the energy be effectively controlled in the optimization process. This paper uses convergence criterion theory to analyze the convergence of the algorithm. The proposed algorithm is used for experiments with an Gaussian mixture model and the experimental results show that the proposed algorithm has better convergence results.
2021, 43(7): 2055-2062.
doi: 10.11999/JEIT200509
Abstract:
The coalition structure generation is an important domain of distributed artificial intelligence. Most coalition formation models are only based on the utility, and any number of coalitions are permitted, which makes it be NP complexity difficult to generate the optimal coalition structure. Actually, Trust is the base of cooperation and has direct effect on the final utility. So, not only utility but also trust relationship should be seriously considered. To this end, the utility constraint is extended to trust and utility constraint, a two-tuples is used to represent utility and trust, which is the base of coalition structure generation. Inspired by the classic s-t-cut algorithm for graph cut, coalition structure generation constrained by trust and utility relationship is investigated. Assuming that individual rationality of agents and the stability of coalition (there is no block) is satisfied, the network is cut by the relationship of utility and trust to formation coalitions. The proposed algorithms of coalition structure generation named MT-s-t-cut and MTU-s-t-cut (Trust s-t-cut) can output the optimal coalition structure in polynomial time. The results of simulated experiments show that the social utility increases with the number of agents, and the running time of the algorithms are far less than that of Dynamic Programming (DP) and Optimal Dynamic Programming and Integer Partition (ODP-IP) algorithms.
The coalition structure generation is an important domain of distributed artificial intelligence. Most coalition formation models are only based on the utility, and any number of coalitions are permitted, which makes it be NP complexity difficult to generate the optimal coalition structure. Actually, Trust is the base of cooperation and has direct effect on the final utility. So, not only utility but also trust relationship should be seriously considered. To this end, the utility constraint is extended to trust and utility constraint, a two-tuples is used to represent utility and trust, which is the base of coalition structure generation. Inspired by the classic s-t-cut algorithm for graph cut, coalition structure generation constrained by trust and utility relationship is investigated. Assuming that individual rationality of agents and the stability of coalition (there is no block) is satisfied, the network is cut by the relationship of utility and trust to formation coalitions. The proposed algorithms of coalition structure generation named MT-s-t-cut and MTU-s-t-cut (Trust s-t-cut) can output the optimal coalition structure in polynomial time. The results of simulated experiments show that the social utility increases with the number of agents, and the running time of the algorithms are far less than that of Dynamic Programming (DP) and Optimal Dynamic Programming and Integer Partition (ODP-IP) algorithms.
2021, 43(7): 2063-2070.
doi: 10.11999/JEIT200436
Abstract:
As a classic subject in computer vision and image processing, pedestrian detection has a wide range of applications to intelligence driving and video monitoring fields. However, most of pedestrian detection methods based on visible or infrared images have no satisfying result in some complex environments or situations, such as rain, smog, occlusion, variation of illuminance and target scales, no matter in terms of detection accuracy or speed. This paper analyzes and finds out that, pedestrians usually have quite different characteristics in visible and infrared image, and which have their own advantages in different environments. Therefore, combining fusion and multi-scale technology, a real-time multi-scale pedestrian detection algorithm suitable for complex environment named FPDNet (Fusion Pedestrian Detection Network) is proposed. The detection framework is consisted by three main modules: feature extraction backbone network, multi-scale detection network and decision-level fusion network. The proposed method is able to extract multi-scale pedestrian characteristics under visible or infrared background adaptively. Experimental results prove that the detection network has good adaptability in complex visual environments, and can meet the demands of practical applications to detection accuracy and speed.
As a classic subject in computer vision and image processing, pedestrian detection has a wide range of applications to intelligence driving and video monitoring fields. However, most of pedestrian detection methods based on visible or infrared images have no satisfying result in some complex environments or situations, such as rain, smog, occlusion, variation of illuminance and target scales, no matter in terms of detection accuracy or speed. This paper analyzes and finds out that, pedestrians usually have quite different characteristics in visible and infrared image, and which have their own advantages in different environments. Therefore, combining fusion and multi-scale technology, a real-time multi-scale pedestrian detection algorithm suitable for complex environment named FPDNet (Fusion Pedestrian Detection Network) is proposed. The detection framework is consisted by three main modules: feature extraction backbone network, multi-scale detection network and decision-level fusion network. The proposed method is able to extract multi-scale pedestrian characteristics under visible or infrared background adaptively. Experimental results prove that the detection network has good adaptability in complex visual environments, and can meet the demands of practical applications to detection accuracy and speed.
2021, 43(7): 2071-2078.
doi: 10.11999/JEIT200448
Abstract:
To solve the problems that current behavior recognition methods can not effectively extract the spatial-temporal information in non-European 3D skeleton sequence and lack attention for specific joints, an action recognition model based on 3D graph convolution and attention enhanced is proposed in this paper. Firstly, the specific working principles of the 3D convolution and graph convolution are introduced; Secondly, a 3D graph convolution method is proposed. It is based on the graph convolution kernel that can handle variable-length neighbor nodes in graph and 3D sampling space of 3D convolution is introduced to improve 2D graph convolution kernel to 3D graph convolution kernel with 3D sampling space. For neighbor nodes in 3D sampling space, this method realizes effective extraction of spatial-temporal information with a 3D graph convolution kernel; Thirdly, in order to enhance attention to specific joints and focus important action information, an attention enhanced structure is designed. Besides, through combining 3D graph convolution with attention enhanced structure, action recognition model based on 3D graph convolution and attention enhanced is proposed. Finally, the researches are carried on NTU-RGBD and MSR Action 3D skeleton action dataset. The results further verify the ability to extract spatial-temporal information of this model and its classification accuracy.
To solve the problems that current behavior recognition methods can not effectively extract the spatial-temporal information in non-European 3D skeleton sequence and lack attention for specific joints, an action recognition model based on 3D graph convolution and attention enhanced is proposed in this paper. Firstly, the specific working principles of the 3D convolution and graph convolution are introduced; Secondly, a 3D graph convolution method is proposed. It is based on the graph convolution kernel that can handle variable-length neighbor nodes in graph and 3D sampling space of 3D convolution is introduced to improve 2D graph convolution kernel to 3D graph convolution kernel with 3D sampling space. For neighbor nodes in 3D sampling space, this method realizes effective extraction of spatial-temporal information with a 3D graph convolution kernel; Thirdly, in order to enhance attention to specific joints and focus important action information, an attention enhanced structure is designed. Besides, through combining 3D graph convolution with attention enhanced structure, action recognition model based on 3D graph convolution and attention enhanced is proposed. Finally, the researches are carried on NTU-RGBD and MSR Action 3D skeleton action dataset. The results further verify the ability to extract spatial-temporal information of this model and its classification accuracy.
2021, 43(7): 2079-2086.
doi: 10.11999/JEIT200263
Abstract:
The correlation information between pixels is of great significance for image segmentation. The existing Fuzzy C-Means (FCM) clustering algorithm lacks sufficient consideration for it. Based on the reliability measure of spatial context, this paper proposes a Reliability-based Spatial context Fuzzy C-Means (RSFCM) clustering algorithm: The clustering algorithm anti-noise performance is improved by effectively modeling the spatial neighborhood; A new reliability fuzzy metric is proposed, which balances the relationship between detail retention and anti-noise, so that the clustering results are more accurate. A synthetic image, a traffic sign image and a remote sensing image are used to test the algorithms performance. The results show, compared with the existing FCM algorithm, RSFCM can effectively suppress heterogeneity of intra-class objects caused by Salt & Pepper noise and Gaussian noise for the image segmentation, improve pixels separability and preserve the edge details of the image greatly.
The correlation information between pixels is of great significance for image segmentation. The existing Fuzzy C-Means (FCM) clustering algorithm lacks sufficient consideration for it. Based on the reliability measure of spatial context, this paper proposes a Reliability-based Spatial context Fuzzy C-Means (RSFCM) clustering algorithm: The clustering algorithm anti-noise performance is improved by effectively modeling the spatial neighborhood; A new reliability fuzzy metric is proposed, which balances the relationship between detail retention and anti-noise, so that the clustering results are more accurate. A synthetic image, a traffic sign image and a remote sensing image are used to test the algorithms performance. The results show, compared with the existing FCM algorithm, RSFCM can effectively suppress heterogeneity of intra-class objects caused by Salt & Pepper noise and Gaussian noise for the image segmentation, improve pixels separability and preserve the edge details of the image greatly.
2021, 43(7): 2087-2095.
doi: 10.11999/JEIT200147
Abstract:
Current object detection algorithms have poor detection results on small targets and dense targets. To address this challenge, a Shallow Enhanced Feature Network (SEFN) is proposed in this paper, which is based on the fusion of multiple features and enhanced shallow feature characterization capabilities. Firstly, the features extracted from the Conv4_3 layer and Conv5_3 layer are combined to form basic fusion features. Then the basic fusion features are inputted into a small multi-scale semantic information fusion module to obtain semantic features of rich contextual information and spatial detail information. The semantic features are fused into the basics features by the feature reuse module to obtain shallow enhanced features. Finally, a series of convolutions are performed based on the shallow enhanced features to obtain multiple features with different scales. Multiple detection branches are then constructed based on the features of different scales. The non-maximum suppression algorithm is used to achieve the final detection. The average accuracy of the proposed model is 81.2% and 33.7% on the PASCAL VOC2007 and MS COCO2014 datasets respectively, which is 2.7% and 4.9% higher than the classic Single Shot multibox Detector (SSD) algorithm. In addition, on detecting small targets in dense target scenes, the detection accuracy and recall rate of the proposed method are significantly improved. The experimental results show that the feature pyramid structure can enhance the semantic information of shallow features, and the feature reuse module can effectively retain shallow detail information for detection, so the proposed method can get better detection performance on small targets and dense targets.
Current object detection algorithms have poor detection results on small targets and dense targets. To address this challenge, a Shallow Enhanced Feature Network (SEFN) is proposed in this paper, which is based on the fusion of multiple features and enhanced shallow feature characterization capabilities. Firstly, the features extracted from the Conv4_3 layer and Conv5_3 layer are combined to form basic fusion features. Then the basic fusion features are inputted into a small multi-scale semantic information fusion module to obtain semantic features of rich contextual information and spatial detail information. The semantic features are fused into the basics features by the feature reuse module to obtain shallow enhanced features. Finally, a series of convolutions are performed based on the shallow enhanced features to obtain multiple features with different scales. Multiple detection branches are then constructed based on the features of different scales. The non-maximum suppression algorithm is used to achieve the final detection. The average accuracy of the proposed model is 81.2% and 33.7% on the PASCAL VOC2007 and MS COCO2014 datasets respectively, which is 2.7% and 4.9% higher than the classic Single Shot multibox Detector (SSD) algorithm. In addition, on detecting small targets in dense target scenes, the detection accuracy and recall rate of the proposed method are significantly improved. The experimental results show that the feature pyramid structure can enhance the semantic information of shallow features, and the feature reuse module can effectively retain shallow detail information for detection, so the proposed method can get better detection performance on small targets and dense targets.
2021, 43(7): 2096-2104.
doi: 10.11999/JEIT200304
Abstract:
Single Shot multibox Detector (SSD) is a object detection algorithm that provides the optimal trade-off among simplicity, speed and accuracy. The single use of detection layers in SSD network structure makes the feature information not fully utilized, which will lead to the small object detection are not robust enough. In this paper, an Attention based Single Shot multibox Detector (ASSD) is proposed. The ASSD algorithm first uses the proposed two-way feature fusion module to fuse the feature information to obtain the feature layer which containing rich details and semantic information. Then, the proposed joint attention unit is used to mine further the key feature information to guide the model optimization. Finally, a series of experiments on the common data set show that the ASSD algorithm effectively improves the detection accuracy of conventional SSD algorithm, especially for small object detection.
Single Shot multibox Detector (SSD) is a object detection algorithm that provides the optimal trade-off among simplicity, speed and accuracy. The single use of detection layers in SSD network structure makes the feature information not fully utilized, which will lead to the small object detection are not robust enough. In this paper, an Attention based Single Shot multibox Detector (ASSD) is proposed. The ASSD algorithm first uses the proposed two-way feature fusion module to fuse the feature information to obtain the feature layer which containing rich details and semantic information. Then, the proposed joint attention unit is used to mine further the key feature information to guide the model optimization. Finally, a series of experiments on the common data set show that the ASSD algorithm effectively improves the detection accuracy of conventional SSD algorithm, especially for small object detection.
2021, 43(7): 2105-2112.
doi: 10.11999/JEIT200376
Abstract:
In view of the gradient effect caused by the gradient effect of the Total Variation (TV) algorithm and the environmental noise in the single pixel imaging system, an image reconstruction based on the Gaussian Smooth compressed sensing Fractional Order Total Variation algorithm (FOTVGS) is proposed. Fractional differential loss of low-frequency components of the image increases the high-frequency components of the image to achieve the purpose of enhancing image details. The Gaussian smoothing filter operator updates the Lagrangian gradient operator to filter out the additive white Gaussian noise caused by the differential operator. Simulation results show that, compared with other four similar algorithms, the algorithm can achieve the maximum Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity(SSIM) at the same sampling rate and noise level. When the sampling rate is 0.2, compared with the Fractional Order Total Variation (FOTV) algorithm, the maximum PSNR and SSIM increase by 1.39 dB (0.035) and 3.91 dB (0.098) respectively. It can be proved that this algorithm can improve the reconstruction quality of the image in the absence of noise and noise, especially in the case of noise, the quality of image reconstruction is greatly improved. The proposed algorithm provides a feasible solution for image reconstruction of noise caused by environment in single-pixel imaging and other computing imaging system.
In view of the gradient effect caused by the gradient effect of the Total Variation (TV) algorithm and the environmental noise in the single pixel imaging system, an image reconstruction based on the Gaussian Smooth compressed sensing Fractional Order Total Variation algorithm (FOTVGS) is proposed. Fractional differential loss of low-frequency components of the image increases the high-frequency components of the image to achieve the purpose of enhancing image details. The Gaussian smoothing filter operator updates the Lagrangian gradient operator to filter out the additive white Gaussian noise caused by the differential operator. Simulation results show that, compared with other four similar algorithms, the algorithm can achieve the maximum Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity(SSIM) at the same sampling rate and noise level. When the sampling rate is 0.2, compared with the Fractional Order Total Variation (FOTV) algorithm, the maximum PSNR and SSIM increase by 1.39 dB (0.035) and 3.91 dB (0.098) respectively. It can be proved that this algorithm can improve the reconstruction quality of the image in the absence of noise and noise, especially in the case of noise, the quality of image reconstruction is greatly improved. The proposed algorithm provides a feasible solution for image reconstruction of noise caused by environment in single-pixel imaging and other computing imaging system.
2021, 43(7): 2113-2120.
doi: 10.11999/JEIT200450
Abstract:
With the large increase of passengers in metro stations, precise and real-time monitoring of passenger flow in subway stations is of great significance for ensuring passenger safety. Based on the features of complicated subway scenes and small pedestrian targets, a Multi-scale Weighted Feature (MWF) fusion network to achieve accurate real-time monitoring of subway passengers is proposed. In the data preprocessing stage, an oversampling target enhancement algorithm is proposed to stitch the pictures with an insufficient proportion of small targets to increase the iteration frequency of small targets during training. Secondly, feature extraction layers based on the VGG16 network are added to the Single Shot multibox Detector (SSD) network. The feature layers of different scales are weighted and fused in different ways, and the optimal feature fusion method is selected. Finally, combined with the small target oversampling enhancement algorithm, a multi-scale weighted feature fusion model is obtained. Experiments show that the detection accuracy of this method has improved by 5.82 percent compared with the SSD network and doesn’t reduce the speed of detection.
With the large increase of passengers in metro stations, precise and real-time monitoring of passenger flow in subway stations is of great significance for ensuring passenger safety. Based on the features of complicated subway scenes and small pedestrian targets, a Multi-scale Weighted Feature (MWF) fusion network to achieve accurate real-time monitoring of subway passengers is proposed. In the data preprocessing stage, an oversampling target enhancement algorithm is proposed to stitch the pictures with an insufficient proportion of small targets to increase the iteration frequency of small targets during training. Secondly, feature extraction layers based on the VGG16 network are added to the Single Shot multibox Detector (SSD) network. The feature layers of different scales are weighted and fused in different ways, and the optimal feature fusion method is selected. Finally, combined with the small target oversampling enhancement algorithm, a multi-scale weighted feature fusion model is obtained. Experiments show that the detection accuracy of this method has improved by 5.82 percent compared with the SSD network and doesn’t reduce the speed of detection.
