Email alert
2022 Vol. 44, No. 9
column
Display Method:
2022, 44(9): 2977-2985.
doi: 10.11999/JEIT211364
Abstract:
This paper dedicates to the design of heterogeneous multi-tier multi-cell communication networks and its multi-level Quality-of-Service (QoS) implementation in the future massive connectivity scenarios. To meet the requirement of massive connectivity in heterogeneous multi-tier networks, a Frequency-Hopping (FH) based Sparse Code Multiple Access (FH/SCMA) transmission is proposed for heterogeneous multi-tier infrastructure in this paper. In such a network infrastructure, all users’ transmitted data are multiplexed by SCMA technique and all intra- and inter- cells located in multi-tier are distinguished by FH multi-access technique. Since the traditional FH technique just provides a single-level frequency-hit rate (i.e., a single-level error-rate), it can not offer the multi-QoS for heterogeneous multi-tier networks. Thus, a novel type of FH sequence set with two level Hamming-correlations is proposed in this paper, and its construction algorithm is designed via the interleaving technique. The strictly theoretical analysis and extensively simulated analysis are carried out to verify the multi-QoS performance of the proposed heterogeneous FH/SCMA networks. The results of this paper are shown that, by employing the new FH into SCMA, the heterogeneous multi-tier networks possesses the massive connectivity, and the strongly anti- multi-tier interference and anti-fading capability; Meanwhile, it achieves successfully the multi-QoS target (i.e., multi-level Bit-Error Ratio(BER)). The proposed FH/SCMA provides a valuable solution to heterogeneous multi-tier networks and multi-QoS requirement from the viewpoint of signal and transmission.
This paper dedicates to the design of heterogeneous multi-tier multi-cell communication networks and its multi-level Quality-of-Service (QoS) implementation in the future massive connectivity scenarios. To meet the requirement of massive connectivity in heterogeneous multi-tier networks, a Frequency-Hopping (FH) based Sparse Code Multiple Access (FH/SCMA) transmission is proposed for heterogeneous multi-tier infrastructure in this paper. In such a network infrastructure, all users’ transmitted data are multiplexed by SCMA technique and all intra- and inter- cells located in multi-tier are distinguished by FH multi-access technique. Since the traditional FH technique just provides a single-level frequency-hit rate (i.e., a single-level error-rate), it can not offer the multi-QoS for heterogeneous multi-tier networks. Thus, a novel type of FH sequence set with two level Hamming-correlations is proposed in this paper, and its construction algorithm is designed via the interleaving technique. The strictly theoretical analysis and extensively simulated analysis are carried out to verify the multi-QoS performance of the proposed heterogeneous FH/SCMA networks. The results of this paper are shown that, by employing the new FH into SCMA, the heterogeneous multi-tier networks possesses the massive connectivity, and the strongly anti- multi-tier interference and anti-fading capability; Meanwhile, it achieves successfully the multi-QoS target (i.e., multi-level Bit-Error Ratio(BER)). The proposed FH/SCMA provides a valuable solution to heterogeneous multi-tier networks and multi-QoS requirement from the viewpoint of signal and transmission.
2022, 44(9): 2986-2993.
doi: 10.11999/JEIT211365
Abstract:
The heterogeneous and intensive deployment of wireless network improves greatly the system capacity, which can meet the increasing data traffic demand of users. However, the complex network structure and almost random base station distribution are not conducive to the performance evaluation and parameter design of systems. Considering this problem, a performance analysis framework for multi-antenna dense heterogeneous networks is proposed. Firstly, resorting to stochastic geometry model, the closed-form expression of coverage probability is derived, and the optimization scheme is proposed. In order to observe intuitively the effects of key system parameters on coverage probability, an asymptotic expression is also given. Secondly, the integral expression of Area Spectral Efficiency (ASE) is derived. In order to reduce the computational complexity, an upper bound of ASE is provided. Finally, an effective algorithm is proposed to design the optimal active Base Stations (BSs) densities, maximizing the ASE with appropriate requirements of coverage probability. The simulation results verify the correctness of the theoretical analysis and the effectiveness of the proposed optimization algorithm. The research results of this paper can not only provide theoretical basis for the performance analysis of complex networks, but also provide feasible schemes for the optimization and design of systems.
The heterogeneous and intensive deployment of wireless network improves greatly the system capacity, which can meet the increasing data traffic demand of users. However, the complex network structure and almost random base station distribution are not conducive to the performance evaluation and parameter design of systems. Considering this problem, a performance analysis framework for multi-antenna dense heterogeneous networks is proposed. Firstly, resorting to stochastic geometry model, the closed-form expression of coverage probability is derived, and the optimization scheme is proposed. In order to observe intuitively the effects of key system parameters on coverage probability, an asymptotic expression is also given. Secondly, the integral expression of Area Spectral Efficiency (ASE) is derived. In order to reduce the computational complexity, an upper bound of ASE is provided. Finally, an effective algorithm is proposed to design the optimal active Base Stations (BSs) densities, maximizing the ASE with appropriate requirements of coverage probability. The simulation results verify the correctness of the theoretical analysis and the effectiveness of the proposed optimization algorithm. The research results of this paper can not only provide theoretical basis for the performance analysis of complex networks, but also provide feasible schemes for the optimization and design of systems.
A Research on Collaborative UAVs Intelligent Decision Optimization for AoI-driven Federated Learning
2022, 44(9): 2994-3003.
doi: 10.11999/JEIT211406
Abstract:
Federated learning is one of the key technologies of 6G, which can use cross-device data to train a usable and safe sharing model on the premise of protecting data privacy. However, most end devices have limited processing capabilities and can not support complex machine learning model training processes. In the framework of Mobile Edge Computing (MEC) in a heterogeneous network convergence environment, multiple Unmanned Aerial Vehicles (UAVs) are used as aerial edge servers to move flexibly within the target area in a collaborative manner, and collect fresh data in time for federated learning and local training to ensure real-time data learning. Multiple factors, such as data freshness, communication cost and model quality, are considered, and the flight trajectories of UAVs, the communication decisions with the user equipment, and the collaborative work between UAVs are comprehensively optimized. Moreover, a priority-based decomposable multi-agent deep reinforcement learning algorithm is used to solve the continuous online decision-making problem of multiple UAVs federated learning to achieve effective collaboration and control. By using multiple real data sets for simulation experiments, simulation results verify that the proposed algorithm can achieve superior performance under different data distributions and in rapidly changing complex dynamic environments.
Federated learning is one of the key technologies of 6G, which can use cross-device data to train a usable and safe sharing model on the premise of protecting data privacy. However, most end devices have limited processing capabilities and can not support complex machine learning model training processes. In the framework of Mobile Edge Computing (MEC) in a heterogeneous network convergence environment, multiple Unmanned Aerial Vehicles (UAVs) are used as aerial edge servers to move flexibly within the target area in a collaborative manner, and collect fresh data in time for federated learning and local training to ensure real-time data learning. Multiple factors, such as data freshness, communication cost and model quality, are considered, and the flight trajectories of UAVs, the communication decisions with the user equipment, and the collaborative work between UAVs are comprehensively optimized. Moreover, a priority-based decomposable multi-agent deep reinforcement learning algorithm is used to solve the continuous online decision-making problem of multiple UAVs federated learning to achieve effective collaboration and control. By using multiple real data sets for simulation experiments, simulation results verify that the proposed algorithm can achieve superior performance under different data distributions and in rapidly changing complex dynamic environments.
2022, 44(9): 3004-3013.
doi: 10.11999/JEIT220370
Abstract:
6G networks will bring the new paradigm of ubiquitous intelligence and on-demand services in all scenarios, in which trustworthy and reliable network services become key technical indicators. Facing the communication requirements in the 6G zero-trust environment, the blockchain is used as a "trust bridge", and the trusted and reliable access management scheme of the 6G in-vehicle network is studied in this paper. Firstly, a zero-knowledge authentication algorithm based on quadratic residual is used to complete the mutual authentication and authorization between the base station and the vehicle without exposing the privacy of the vehicle. Then, in order to improve the verification efficiency and save the energy consumption of the base station, a road redundant computing power incentive model based on the contract theory is established, and part of the verification tasks of the base station are allocated to edge servers or parked vehicles and given corresponding rewards. Finally, a 6G zero-trust network architecture of vehicle based on two-tiered blockchain is established. The primary chain maintained by the base stations and the secondary chain maintained by the edge computing nodes are used to record the important parameters of the identity verification of the vehicular networks, so as to meet the requirement of trustworthy access in the zero-trust environment. Compared to the existing baseline approaches, the proposed scheme improves significantly the vehicle verification efficiency and reduces the energy consumption of the base station without revealing the privacy of the vehicles.
6G networks will bring the new paradigm of ubiquitous intelligence and on-demand services in all scenarios, in which trustworthy and reliable network services become key technical indicators. Facing the communication requirements in the 6G zero-trust environment, the blockchain is used as a "trust bridge", and the trusted and reliable access management scheme of the 6G in-vehicle network is studied in this paper. Firstly, a zero-knowledge authentication algorithm based on quadratic residual is used to complete the mutual authentication and authorization between the base station and the vehicle without exposing the privacy of the vehicle. Then, in order to improve the verification efficiency and save the energy consumption of the base station, a road redundant computing power incentive model based on the contract theory is established, and part of the verification tasks of the base station are allocated to edge servers or parked vehicles and given corresponding rewards. Finally, a 6G zero-trust network architecture of vehicle based on two-tiered blockchain is established. The primary chain maintained by the base stations and the secondary chain maintained by the edge computing nodes are used to record the important parameters of the identity verification of the vehicular networks, so as to meet the requirement of trustworthy access in the zero-trust environment. Compared to the existing baseline approaches, the proposed scheme improves significantly the vehicle verification efficiency and reduces the energy consumption of the base station without revealing the privacy of the vehicles.
2022, 44(9): 3014-3023.
doi: 10.11999/JEIT220306
Abstract:
To balance the network loads and utilize fully the network resources, joint cooperative computation offloading and wireless resource management is considered for ultra-dense heterogeneous edge computing networks with multiple users and multiple tasks, which minimizes the system energy consumption under the constraints of users’ delay. During the problem modeling, a frequency spectrum partitioning mechanism is introduced to tackle serious network interference caused by ultra-dense deployment of base stations, and Non-Orthogonal Multiple Access (NOMA) technology is introduced to improve the uplink frequency spectrum efficiency. Considering that the optimization problem is a nonlinear mixed-integer form, according to Adaptive Genetic Algorithm with Diversity-Guided Mutation (AGADGM), an effective algorithm used for cooperative computation offloading and resource allocation is designed. The simulation results show that proposed algorithm could achieve lower system energy consumption than other existing algorithms under strict constraints of users’ delay.
To balance the network loads and utilize fully the network resources, joint cooperative computation offloading and wireless resource management is considered for ultra-dense heterogeneous edge computing networks with multiple users and multiple tasks, which minimizes the system energy consumption under the constraints of users’ delay. During the problem modeling, a frequency spectrum partitioning mechanism is introduced to tackle serious network interference caused by ultra-dense deployment of base stations, and Non-Orthogonal Multiple Access (NOMA) technology is introduced to improve the uplink frequency spectrum efficiency. Considering that the optimization problem is a nonlinear mixed-integer form, according to Adaptive Genetic Algorithm with Diversity-Guided Mutation (AGADGM), an effective algorithm used for cooperative computation offloading and resource allocation is designed. The simulation results show that proposed algorithm could achieve lower system energy consumption than other existing algorithms under strict constraints of users’ delay.
2022, 44(9): 3024-3032.
doi: 10.11999/JEIT220409
Abstract:
To balance the spectrum efficiency and energy efficiency of multi-beam satellite communication system, and guarantee the fairness of user service in multiusers scenarios, a fairness utility function-based BeamForming (BF) scheme is proposed. Specifically, considering the satellite transmit power minimization criteria and the system sum rate maximization criteria, a multi-objective optimization problem is first formulated, which adopts the\begin{document}$\alpha $\end{document} ![]()
![]()
To balance the spectrum efficiency and energy efficiency of multi-beam satellite communication system, and guarantee the fairness of user service in multiusers scenarios, a fairness utility function-based BeamForming (BF) scheme is proposed. Specifically, considering the satellite transmit power minimization criteria and the system sum rate maximization criteria, a multi-objective optimization problem is first formulated, which adopts the
2022, 44(9): 3033-3042.
doi: 10.11999/JEIT211289
Abstract:
Multi-user transmission system combined multi-antenna beamforming with cooperative Non-Orthogonal Multiple Access (NOMA) technique is investigated for Unmanned Aerial Vehicle (UAV) assisted satellite communication system. Firstly, in case that UAV employs multi-antenna and NOMA technique to serve multiple users, a beamforming scheme is proposed to maximize the average signal-to-interference-plus-noise ratio by using angle information of users. Secondly, under the condition that the satellite - UAV link follows the correlated Shadowed-Rician fading while the UAV - terrestrial link follows Nakagami-m fading, the closed-form expressions of outage probability for system are derived. Furthermore, the asymptotic outage probability formulas in the high signal-to-noise ratio regime are also developed to analyze the system performance. Finally, computer simulations are provided to validate the correctness of the theoretical analysis and the superiority of the proposed scheme.
Multi-user transmission system combined multi-antenna beamforming with cooperative Non-Orthogonal Multiple Access (NOMA) technique is investigated for Unmanned Aerial Vehicle (UAV) assisted satellite communication system. Firstly, in case that UAV employs multi-antenna and NOMA technique to serve multiple users, a beamforming scheme is proposed to maximize the average signal-to-interference-plus-noise ratio by using angle information of users. Secondly, under the condition that the satellite - UAV link follows the correlated Shadowed-Rician fading while the UAV - terrestrial link follows Nakagami-m fading, the closed-form expressions of outage probability for system are derived. Furthermore, the asymptotic outage probability formulas in the high signal-to-noise ratio regime are also developed to analyze the system performance. Finally, computer simulations are provided to validate the correctness of the theoretical analysis and the superiority of the proposed scheme.
2022, 44(9): 3043-3050.
doi: 10.11999/JEIT220374
Abstract:
The edge caching based on blockchain will achieve a wider range of content sharing, and enhance the efficiency of caching contents. However, different operators build their own edge devices and the cached contents are isolated and have difficulty in sharing information. In this paper, a blockchain-based edge caching system framework and a content sharing and transaction process is proposed, which can realize content sharing between different operators. In addition, a partial Practical Byzantine Fault Tolerant (pPBFT) consensus mechanism based on content caching is designed to reduce the consensus cost of high-dimensional caching nodes, in which only the consortium nodes that cache the relevant content can be selected as execution nodes for validating smart contracts. Finally, through quantifying the benefit obtained by operators' content sharing, the closed-form optimal solution is derived with the aim to maximize the profit by adopting the proposed content caching strategy, and the optimal caching strategy related to the popularity of the content is further developed. Simulation results show that the proposed consensus mechanism and caching strategy based on this framework can effectively increase the operator's caching revenue.
The edge caching based on blockchain will achieve a wider range of content sharing, and enhance the efficiency of caching contents. However, different operators build their own edge devices and the cached contents are isolated and have difficulty in sharing information. In this paper, a blockchain-based edge caching system framework and a content sharing and transaction process is proposed, which can realize content sharing between different operators. In addition, a partial Practical Byzantine Fault Tolerant (pPBFT) consensus mechanism based on content caching is designed to reduce the consensus cost of high-dimensional caching nodes, in which only the consortium nodes that cache the relevant content can be selected as execution nodes for validating smart contracts. Finally, through quantifying the benefit obtained by operators' content sharing, the closed-form optimal solution is derived with the aim to maximize the profit by adopting the proposed content caching strategy, and the optimal caching strategy related to the popularity of the content is further developed. Simulation results show that the proposed consensus mechanism and caching strategy based on this framework can effectively increase the operator's caching revenue.
2022, 44(9): 3051-3057.
doi: 10.11999/JEIT211300
Abstract:
The existing literature adopt ellipse models to describe the communication scenarios of the power Internet of Things (IoT) for sixth Generation (6G), which neglect the impacts of the elevation angles of the paths on the propagation characteristics. To solve this issue, a three-dimensional semi-ellipsoid model to describe the communication scenarios of the power IoT for 6G is proposed, which improves the matching accuracy of the propagation model with the practical communication scenarios of the power Internet of Things. In the proposed algorithm, the transmission characteristics of physical layer data are revealed by deriving expressions for the complex impulse response functions of different transmission paths in the wireless transmission channel of power IoT communication. Simulation results analyze the cross-correlation characteristics between different transmission paths, explore the multi-node channel transmission characteristics of the power IoT, and Verify the correctness of the above data transmission analysis algorithm. The aforementioned research is meaningful for the design of the power internet of things communications systems.
The existing literature adopt ellipse models to describe the communication scenarios of the power Internet of Things (IoT) for sixth Generation (6G), which neglect the impacts of the elevation angles of the paths on the propagation characteristics. To solve this issue, a three-dimensional semi-ellipsoid model to describe the communication scenarios of the power IoT for 6G is proposed, which improves the matching accuracy of the propagation model with the practical communication scenarios of the power Internet of Things. In the proposed algorithm, the transmission characteristics of physical layer data are revealed by deriving expressions for the complex impulse response functions of different transmission paths in the wireless transmission channel of power IoT communication. Simulation results analyze the cross-correlation characteristics between different transmission paths, explore the multi-node channel transmission characteristics of the power IoT, and Verify the correctness of the above data transmission analysis algorithm. The aforementioned research is meaningful for the design of the power internet of things communications systems.
2022, 44(9): 3058-3064.
doi: 10.11999/JEIT220212
Abstract:
The entangled photon positioning method is one of the current research hotspots in the field of navigation and positioning. However, the existing methods consider rarely the influence of the dynamic changes of the scattering environment on the performance of propagation distance estimation in different light sources, resulting in the problem of low positioning accuracy and poor robustness. To solve this problem, an entangled light quantum positioning method is proposed based on adaptive light source selection. First of all, the mathematical relationship between the interference of different scattering environments and the light propagation distance in each light source is established, the average photon loss rate of each light source signal light time pulse sequence is calculated, and the time pulse sequence of each light source is dynamically grouped. Then, the optical time pulse sequence is matched, and the propagation distance of each light source in each group is obtained according to the second-order correlation curve of light. Finally, based on the relative error of each light source in each group, the light source with the smaller relative error in each group is dynamically selected for positioning. The experimental results show that the proposed method has high positioning accuracy and strong positioning robustness.
The entangled photon positioning method is one of the current research hotspots in the field of navigation and positioning. However, the existing methods consider rarely the influence of the dynamic changes of the scattering environment on the performance of propagation distance estimation in different light sources, resulting in the problem of low positioning accuracy and poor robustness. To solve this problem, an entangled light quantum positioning method is proposed based on adaptive light source selection. First of all, the mathematical relationship between the interference of different scattering environments and the light propagation distance in each light source is established, the average photon loss rate of each light source signal light time pulse sequence is calculated, and the time pulse sequence of each light source is dynamically grouped. Then, the optical time pulse sequence is matched, and the propagation distance of each light source in each group is obtained according to the second-order correlation curve of light. Finally, based on the relative error of each light source in each group, the light source with the smaller relative error in each group is dynamically selected for positioning. The experimental results show that the proposed method has high positioning accuracy and strong positioning robustness.
2022, 44(9): 3065-3074.
doi: 10.11999/JEIT211390
Abstract:
In order to meet the ever-increasing computation-intensive and delay-sensitive service requirements of users, as well as minimizing the processing cost of computation tasks, an optimization problem of joint task offloading, wireless resource management, and computation resource block allocation are formulated for ultra-dense heterogeneous edge computing networks under users’ delay constraints. Such a formulated problem is in a nonlinear and mixed-integer form. In order to meet the constraints and improve the convergence speed of algorithm, a Hybrid Particle Swarm Optimization (HPSO) algorithm is developed by improving Hierarchical Adaptive Search (HAS) algorithm. The simulation results show that HPSO algorithm is superior to other benchmark algorithms under users’ delay constraints, and can reduce the task processing cost effectively.
In order to meet the ever-increasing computation-intensive and delay-sensitive service requirements of users, as well as minimizing the processing cost of computation tasks, an optimization problem of joint task offloading, wireless resource management, and computation resource block allocation are formulated for ultra-dense heterogeneous edge computing networks under users’ delay constraints. Such a formulated problem is in a nonlinear and mixed-integer form. In order to meet the constraints and improve the convergence speed of algorithm, a Hybrid Particle Swarm Optimization (HPSO) algorithm is developed by improving Hierarchical Adaptive Search (HAS) algorithm. The simulation results show that HPSO algorithm is superior to other benchmark algorithms under users’ delay constraints, and can reduce the task processing cost effectively.
2022, 44(9): 3075-3082.
doi: 10.11999/JEIT220390
Abstract:
In order to meet the requirements of ultra-high reliability and ultra-low latency in future cellular Internet-of-Things (IoT), an algorithm for such networks, which ensures the fairness-aware energy efficiency maximization in short packet domain, is developed. Firstly, a nonlinear fractional programming resource allocation model, which optimize the beamforming vector, is constructed with several constraints such as minimum user transmission rate and maximum power of per transmitter. Subsequently, the original non-convex optimization problem is transformed into a standard convex problem by exploiting the variable substitution and continuous convex approximation. Furthermore, an iterative energy efficiency optimization algorithm is developed in short packet regime. Finally, the numerical simulation results verify that the proposed algorithm has good energy efficiency performance in the short packet domain.
In order to meet the requirements of ultra-high reliability and ultra-low latency in future cellular Internet-of-Things (IoT), an algorithm for such networks, which ensures the fairness-aware energy efficiency maximization in short packet domain, is developed. Firstly, a nonlinear fractional programming resource allocation model, which optimize the beamforming vector, is constructed with several constraints such as minimum user transmission rate and maximum power of per transmitter. Subsequently, the original non-convex optimization problem is transformed into a standard convex problem by exploiting the variable substitution and continuous convex approximation. Furthermore, an iterative energy efficiency optimization algorithm is developed in short packet regime. Finally, the numerical simulation results verify that the proposed algorithm has good energy efficiency performance in the short packet domain.
2022, 44(9): 3083-3090.
doi: 10.11999/JEIT220343
Abstract:
The massive data generated by the sixth generation mobile communication technology (6G) network brings new challenges to data storage, which promotes further the rapid development of storage technology. Not AND (NAND) flash memory has the advantages of fast reading/writing speed and high reliability, and hence it possesses a wide application prospect in the 6G network. To improve the reliability of NAND flash memory, according to the error characteristics of two different bit-line structures, an all-bit-line-structure-aided equal-precision remapping scheme and an odd-even-bit-line-structure-aided unequal-precision remapping scheme are proposed. Simulation results show that the two new remapping schemes improve effectively the bit error performance of flash memory. Therefore, the remapping technology proposed in this paper can be regarded as a reliable and efficient storage optimization technology for 6G network.
The massive data generated by the sixth generation mobile communication technology (6G) network brings new challenges to data storage, which promotes further the rapid development of storage technology. Not AND (NAND) flash memory has the advantages of fast reading/writing speed and high reliability, and hence it possesses a wide application prospect in the 6G network. To improve the reliability of NAND flash memory, according to the error characteristics of two different bit-line structures, an all-bit-line-structure-aided equal-precision remapping scheme and an odd-even-bit-line-structure-aided unequal-precision remapping scheme are proposed. Simulation results show that the two new remapping schemes improve effectively the bit error performance of flash memory. Therefore, the remapping technology proposed in this paper can be regarded as a reliable and efficient storage optimization technology for 6G network.
2022, 44(9): 3091-3098.
doi: 10.11999/JEIT220272
Abstract:
As a new type of network architecture, Space-Air-Ground Integrated Network (SAGIN) is the key support for 6G to realize ubiquitous connection in the future. In this paper, a partial task offloading approach for Mobile Edge Computing (MEC) in SAGIN is proposed. Firstly, the coverage time of Low Earth Orbit (LEO) satellite is analyzed. Then, a non-convex and multivariable coupling problem for minimization of total energy consumption of all Unmanned Aerial Vehicles (UAVs) is formulated by the joint design of the association control, computation task allocation, bandwidth allocation, UAV computation resource, and UAV trajectory. To solve this problem, the alternation optimization technique is invoked to decouple the original non-convex problem into three subproblems which are solved by the successive convex approximation method. Numerical results demonstrate that the proposed algorithm has good convergence performance and reduces effectively the system energy consumption.
As a new type of network architecture, Space-Air-Ground Integrated Network (SAGIN) is the key support for 6G to realize ubiquitous connection in the future. In this paper, a partial task offloading approach for Mobile Edge Computing (MEC) in SAGIN is proposed. Firstly, the coverage time of Low Earth Orbit (LEO) satellite is analyzed. Then, a non-convex and multivariable coupling problem for minimization of total energy consumption of all Unmanned Aerial Vehicles (UAVs) is formulated by the joint design of the association control, computation task allocation, bandwidth allocation, UAV computation resource, and UAV trajectory. To solve this problem, the alternation optimization technique is invoked to decouple the original non-convex problem into three subproblems which are solved by the successive convex approximation method. Numerical results demonstrate that the proposed algorithm has good convergence performance and reduces effectively the system energy consumption.
2022, 44(9): 3099-3107.
doi: 10.11999/JEIT210838
Abstract:
The Satellite-Terrestrial Integrated Networks (STIN) is able to provide an effective framework to solve the limitation of coverage and spectrum shortage of terrestrial base station for the next generation wireless communication. Considering the performance limitation of downlink broadcast and unicast standalone transmission in STIN, an optimization problem that minimize the transmit power based on Quality of Service (QoS) constraints is established, and a robust joint beamforming transmission scheme based on Layer Division Multiplexing (LDM) is proposed. According to the worst-case criterion, the robust optimization problem with infinite-dimension constraints is transformed into a deterministic optimization form with Linear Matrix Inequality (LMI) by employing S-procedure and Semi-Definite Relaxation (SDR) methods. And then, an iterative algorithm based on penalty function is also proposed to address the original problem. Simulation results show that, the power consumption of the proposed algorithm is about 6 dBm lower than traditional orthogonal Time Division Multiplexing (TDM) transmission scheme, while the cooperative transmission scheme outperforms significantly the non-cooperative ones in terms of the average user rate.
The Satellite-Terrestrial Integrated Networks (STIN) is able to provide an effective framework to solve the limitation of coverage and spectrum shortage of terrestrial base station for the next generation wireless communication. Considering the performance limitation of downlink broadcast and unicast standalone transmission in STIN, an optimization problem that minimize the transmit power based on Quality of Service (QoS) constraints is established, and a robust joint beamforming transmission scheme based on Layer Division Multiplexing (LDM) is proposed. According to the worst-case criterion, the robust optimization problem with infinite-dimension constraints is transformed into a deterministic optimization form with Linear Matrix Inequality (LMI) by employing S-procedure and Semi-Definite Relaxation (SDR) methods. And then, an iterative algorithm based on penalty function is also proposed to address the original problem. Simulation results show that, the power consumption of the proposed algorithm is about 6 dBm lower than traditional orthogonal Time Division Multiplexing (TDM) transmission scheme, while the cooperative transmission scheme outperforms significantly the non-cooperative ones in terms of the average user rate.
2022, 44(9): 3108-3116.
doi: 10.11999/JEIT220373
Abstract:
Introducing edge caching into fog radio access networks can effectively reduce the redundancy of content transmission. However, the existing content caching strategies consider rarely the dynamic nature of already cached content. A caching update algorithm based on content popularity and information freshness is proposed. The proposed algorithm considers fully the mobility of users and the temporal and spatial dynamics of content popularity. Furthermore, the Age of Information (AoI) is introduced to achieve a dynamic content update procedure. More specifically, the proposed algorithm adopts initially a Bidirectional Long Short-Term Memory network (Bi-LSTM) to predict the user's location in the next period according to the user's historical location information. Secondly, according to the acquired user location, combined with the user's preference model, the content popularity of each location area is obtained accordingly, and the most popular content will be cached at the Fog Access Points(F-APs). Finally, concerning AoI requirements of the already cached content, the caching update window can be dynamically adjusted to achieve a high-efficient and low-latency caching process. Simulation results demonstrate that the proposed algorithm improves effectively the content cache hit rate, and also minimizes the average delay of content transmission while ensuring the timeliness of the information.
Introducing edge caching into fog radio access networks can effectively reduce the redundancy of content transmission. However, the existing content caching strategies consider rarely the dynamic nature of already cached content. A caching update algorithm based on content popularity and information freshness is proposed. The proposed algorithm considers fully the mobility of users and the temporal and spatial dynamics of content popularity. Furthermore, the Age of Information (AoI) is introduced to achieve a dynamic content update procedure. More specifically, the proposed algorithm adopts initially a Bidirectional Long Short-Term Memory network (Bi-LSTM) to predict the user's location in the next period according to the user's historical location information. Secondly, according to the acquired user location, combined with the user's preference model, the content popularity of each location area is obtained accordingly, and the most popular content will be cached at the Fog Access Points(F-APs). Finally, concerning AoI requirements of the already cached content, the caching update window can be dynamically adjusted to achieve a high-efficient and low-latency caching process. Simulation results demonstrate that the proposed algorithm improves effectively the content cache hit rate, and also minimizes the average delay of content transmission while ensuring the timeliness of the information.
2022, 44(9): 3117-3125.
doi: 10.11999/JEIT210877
Abstract:
In view of the problem of resource allocation in the Unmanned Aerial Vehicle (UAV) enabled air-ground network with massive MIMO, a K-armed bandit-based reinforcement learning algorithm is proposed to jointly optimize the user selection and power allocation to maximize the total throughput of ground users. Firstly, users are clustered according to their geographic location, and the cluster center nodes are used to plan the trajectory of UAVs. Secondly, without considering the UAV-UAV communication links, the problem of multi-UAV resource allocation is transformed into a mutually independent multi-agent reinforcement learning problem. Finally, an episode-based K-armed bandit algorithm with multi-agent and multi-state is proposed to realize the joint optimization of user selection and power allocation, so that the UAV can dynamically adapt to the changes of its position and channel state by defining the position index of the UAV as the state space. Simulation results verify that the proposed algorithm can adaptively adjust the resource allocation strategy according to the channel conditions, which can effectively improve the total system throughput compared with the existing schemes.
In view of the problem of resource allocation in the Unmanned Aerial Vehicle (UAV) enabled air-ground network with massive MIMO, a K-armed bandit-based reinforcement learning algorithm is proposed to jointly optimize the user selection and power allocation to maximize the total throughput of ground users. Firstly, users are clustered according to their geographic location, and the cluster center nodes are used to plan the trajectory of UAVs. Secondly, without considering the UAV-UAV communication links, the problem of multi-UAV resource allocation is transformed into a mutually independent multi-agent reinforcement learning problem. Finally, an episode-based K-armed bandit algorithm with multi-agent and multi-state is proposed to realize the joint optimization of user selection and power allocation, so that the UAV can dynamically adapt to the changes of its position and channel state by defining the position index of the UAV as the state space. Simulation results verify that the proposed algorithm can adaptively adjust the resource allocation strategy according to the channel conditions, which can effectively improve the total system throughput compared with the existing schemes.
2022, 44(9): 3126-3135.
doi: 10.11999/JEIT220383
Abstract:
With the rapid development of the Sixth Generation (6G) mobile communications and Unmanned Aerial Vehicle (UAV) technology, UAV communication networks become the key part of intelligent space-air-ground integration networks in 6G, which play an important role in battlefield reconnaissance, field rescue, information transmission of Internet of things and other military and civilian fields. Considering the characteristics of UAV networks such as large-scale, high-dynamic and self-organization, mission-driven UAV networks model based on coalitions for 6G is proposed. The model is discussed in three aspects: coalition formations, mission executions, and resource management. Combined with game theory, machine learning and online decisions, the optimization methods and simulation examples of UAV coalition networks are given. Finally, the application prospect of 6G UAV communication networks and the problems to be solved are discussed.
With the rapid development of the Sixth Generation (6G) mobile communications and Unmanned Aerial Vehicle (UAV) technology, UAV communication networks become the key part of intelligent space-air-ground integration networks in 6G, which play an important role in battlefield reconnaissance, field rescue, information transmission of Internet of things and other military and civilian fields. Considering the characteristics of UAV networks such as large-scale, high-dynamic and self-organization, mission-driven UAV networks model based on coalitions for 6G is proposed. The model is discussed in three aspects: coalition formations, mission executions, and resource management. Combined with game theory, machine learning and online decisions, the optimization methods and simulation examples of UAV coalition networks are given. Finally, the application prospect of 6G UAV communication networks and the problems to be solved are discussed.
2022, 44(9): 3136-3143.
doi: 10.11999/JEIT210560
Abstract:
In the Geosynchronous-Earth-Orbit (GEO) SAR imaging, the extremely large swath width causes the imaging plane to no longer satisfy the flat plane approximation, which makes the accelerated BP algorithms based on the flat plane grid invalid. In this paper, an accelerated BP algorithm based on ground grid is proposed to process accurately and efficiently the GEO SAR signals. The imaging grids are arranged on the ground surface to correct the complex space variance of the signal caused by orbit and ground surface curvature. To solve the spectrum aliasing of sub-aperture images, a two-step spectrum compressing method is proposed to achieve the sub-aperture spectrum de-aliasing before fusion. And a multi-stage sub-aperture image fusion method is adopted to improve the imaging efficiency. Finally, simulation results are shown to verify the accuracy and efficiency of the proposed focusing approaches.
In the Geosynchronous-Earth-Orbit (GEO) SAR imaging, the extremely large swath width causes the imaging plane to no longer satisfy the flat plane approximation, which makes the accelerated BP algorithms based on the flat plane grid invalid. In this paper, an accelerated BP algorithm based on ground grid is proposed to process accurately and efficiently the GEO SAR signals. The imaging grids are arranged on the ground surface to correct the complex space variance of the signal caused by orbit and ground surface curvature. To solve the spectrum aliasing of sub-aperture images, a two-step spectrum compressing method is proposed to achieve the sub-aperture spectrum de-aliasing before fusion. And a multi-stage sub-aperture image fusion method is adopted to improve the imaging efficiency. Finally, simulation results are shown to verify the accuracy and efficiency of the proposed focusing approaches.
2022, 44(9): 3144-3154.
doi: 10.11999/JEIT210610
Abstract:
Considering the problem that the output of three antennas Synthetic Aperture Radar-Ground Moving Target Indication (SAR-GMTI) under single jammer is periodically cancelled, causing the moving targets exposed due to the blind jamming areas. According to the noise multiplication, a countermeasure against the jamming cancellation is proposed, which is based on multi-jammer cooperation in space-frequency domain. First, the noise multiplication jamming is supplemented with frequency shift and the signal model of multi-jammer is established. Then, the analytical expression of SAR-GMTI jamming output is deduced and the essential reason why jamming is cancelled is analyzed, based on which the cooperation condition in space-frequency domain is obtained. Afterwards, the performance of the proposed countermeasure is analyzed in detail. Simulation results demonstrate that the proposed cooperation countermeasure can effectively offset the SAR-GMTI jamming cancellation, achieving suppression and deception jamming flexibly in the appointed location and scope without blind jamming area, which will provide some reference values for engineering applications.
Considering the problem that the output of three antennas Synthetic Aperture Radar-Ground Moving Target Indication (SAR-GMTI) under single jammer is periodically cancelled, causing the moving targets exposed due to the blind jamming areas. According to the noise multiplication, a countermeasure against the jamming cancellation is proposed, which is based on multi-jammer cooperation in space-frequency domain. First, the noise multiplication jamming is supplemented with frequency shift and the signal model of multi-jammer is established. Then, the analytical expression of SAR-GMTI jamming output is deduced and the essential reason why jamming is cancelled is analyzed, based on which the cooperation condition in space-frequency domain is obtained. Afterwards, the performance of the proposed countermeasure is analyzed in detail. Simulation results demonstrate that the proposed cooperation countermeasure can effectively offset the SAR-GMTI jamming cancellation, achieving suppression and deception jamming flexibly in the appointed location and scope without blind jamming area, which will provide some reference values for engineering applications.
2022, 44(9): 3155-3162.
doi: 10.11999/JEIT210524
Abstract:
A long synthetic aperture is required to obtain two-dimensional high resolution in synthetic aperture passive positioning. However, the polynomial approximation of the slant range causes range positioning errors. To resolve the problem, the influencing factors of the range positioning error are analyzed in this paper and an approximate expression of the positioning error is given. The influencing factors of range resolution and azimuth resolution are also analyzed, and the resolution analysis method is studied. On this basis, the optimization method of synthetic aperture length is given by combining the constraints of positioning accuracy and resolution. The simulation results verify the effectiveness of the proposed method.
A long synthetic aperture is required to obtain two-dimensional high resolution in synthetic aperture passive positioning. However, the polynomial approximation of the slant range causes range positioning errors. To resolve the problem, the influencing factors of the range positioning error are analyzed in this paper and an approximate expression of the positioning error is given. The influencing factors of range resolution and azimuth resolution are also analyzed, and the resolution analysis method is studied. On this basis, the optimization method of synthetic aperture length is given by combining the constraints of positioning accuracy and resolution. The simulation results verify the effectiveness of the proposed method.
2022, 44(9): 3163-3169.
doi: 10.11999/JEIT210690
Abstract:
In this paper, target detection in the presence of jamming and compound Gaussian clutter is studied. For improving the better detection performance under a low number of Independent Identically Distributed (IID) training samples, the persymmetric structure of the receive antenna and the priori information of clutter covariance matrix are employed. Based on two step Generalized Likelihood Ratio Test (GLRT), the Bayesian detector is derived under this background. Simulation results show that the proposed detector achieves better performance when the number of training samples is small.
In this paper, target detection in the presence of jamming and compound Gaussian clutter is studied. For improving the better detection performance under a low number of Independent Identically Distributed (IID) training samples, the persymmetric structure of the receive antenna and the priori information of clutter covariance matrix are employed. Based on two step Generalized Likelihood Ratio Test (GLRT), the Bayesian detector is derived under this background. Simulation results show that the proposed detector achieves better performance when the number of training samples is small.
2022, 44(9): 3170-3177.
doi: 10.11999/JEIT210695
Abstract:
In order to solve the problem that traditional radar signal recognition methods can not effectively expand the recognition types, a radar signal recognition method based on knowledge distillation and attention map is proposed. Firstly, the Smooth Pseudo Wigner-Ville Distribution (SPWVD) of the radar signal is used as input; Then, the incremental learning network structure based on residual network is designed, and the loss function based on knowledge distillation and attention map is used to alleviate the catastrophic forgetting in the process of category increment; Finally, a method based on the mean distance of sample features is used to manage the data set, which reduces effectively the occupied storage resources. Experiments show that this method can quickly complete the training of the extended classification signal under the condition of limited storage resources, and has good recognition accuracy for the original classification and the extended classification signal.
In order to solve the problem that traditional radar signal recognition methods can not effectively expand the recognition types, a radar signal recognition method based on knowledge distillation and attention map is proposed. Firstly, the Smooth Pseudo Wigner-Ville Distribution (SPWVD) of the radar signal is used as input; Then, the incremental learning network structure based on residual network is designed, and the loss function based on knowledge distillation and attention map is used to alleviate the catastrophic forgetting in the process of category increment; Finally, a method based on the mean distance of sample features is used to manage the data set, which reduces effectively the occupied storage resources. Experiments show that this method can quickly complete the training of the extended classification signal under the condition of limited storage resources, and has good recognition accuracy for the original classification and the extended classification signal.
2022, 44(9): 3178-3185.
doi: 10.11999/JEIT210607
Abstract:
Considering the problem of parasitic loop oscillation caused by pointing error slope of phased array seeker for missile guidance system, an estimation algorithm of pointing error slope is proposed, and target state can be estimated synchronously. Based on the Rao-Blackwellised Particle Filters (RBPF), the simultaneous estimation of pointing error slope and target state is decomposed into two problems: one is the posterior estimation of pointing error slope, the other is the target state estimation conditional on the estimation of pointing error slope. The derivation process of the algorithm is given and the numerical simulation is carried out. The simulation results show that the algorithm proposed has better performance in estimating the pointing error slope of phased array seeker, and the target state information can be estimated accurately at the same time. Using this information to form the guidance command can eliminate the adverse effects of pointing error slope on the guidance system, and improve system stability and guidance accuracy.
Considering the problem of parasitic loop oscillation caused by pointing error slope of phased array seeker for missile guidance system, an estimation algorithm of pointing error slope is proposed, and target state can be estimated synchronously. Based on the Rao-Blackwellised Particle Filters (RBPF), the simultaneous estimation of pointing error slope and target state is decomposed into two problems: one is the posterior estimation of pointing error slope, the other is the target state estimation conditional on the estimation of pointing error slope. The derivation process of the algorithm is given and the numerical simulation is carried out. The simulation results show that the algorithm proposed has better performance in estimating the pointing error slope of phased array seeker, and the target state information can be estimated accurately at the same time. Using this information to form the guidance command can eliminate the adverse effects of pointing error slope on the guidance system, and improve system stability and guidance accuracy.
2022, 44(9): 3186-3194.
doi: 10.11999/JEIT210994
Abstract:
In order to solve the problems of complex tracking loop structure of traditional GPS receivers, and poor tracking performance in the environment of low Signal-to-Noise Ratio (SNR) and high dynamic, a new GPS signal tracking algorithm based on phase slope stripe detection is proposed, in which the pseudo-code delay can be accurately estimated by the Frequency domain phase of Time Difference Of Arrival (TDOA), and the carrier Doppler frequency offset can be accurately estimated by the Time domain phase of Frequency Difference Of Arrival (FDOA).The loop structure is simplified and the tracking accuracy is improved, compared with the traditional method, the code phase measurement accuracy is improved by 60% and the carrier Doppler measurement accuracy is improved by 31% when the SCNR is 32 dB-Hz. In addition, accurate tracking can be achieved in high dynamic environment, which has research significance for improving the performance of GPS receiver.
In order to solve the problems of complex tracking loop structure of traditional GPS receivers, and poor tracking performance in the environment of low Signal-to-Noise Ratio (SNR) and high dynamic, a new GPS signal tracking algorithm based on phase slope stripe detection is proposed, in which the pseudo-code delay can be accurately estimated by the Frequency domain phase of Time Difference Of Arrival (TDOA), and the carrier Doppler frequency offset can be accurately estimated by the Time domain phase of Frequency Difference Of Arrival (FDOA).The loop structure is simplified and the tracking accuracy is improved, compared with the traditional method, the code phase measurement accuracy is improved by 60% and the carrier Doppler measurement accuracy is improved by 31% when the SCNR is 32 dB-Hz. In addition, accurate tracking can be achieved in high dynamic environment, which has research significance for improving the performance of GPS receiver.
2022, 44(9): 3195-3202.
doi: 10.11999/JEIT210640
Abstract:
Global Navigation Satellite System (GNSS) spoofing causes the target receiver to generate incorrect positioning results. Using Inertial Navigation System (INS)-aided, constructing chi-squared test statistics based on the Kalman filter innovation sequence is an effective means to detect airborne GNSS spoofing. However, the algorithm cannot give the spoofing duration, causing the INS/GNSS system could not to determine whether the calculated positioning information is correct based on the algorithm. In this paper, a limited-memory chi-squared detection based on the reconstructed innovation sequence is proposed by using the Distance Measuring Equipment (DME) system. The algorithm uses the existing INS, GNSS and DME data to construct an innovation sequence that does not participate in Kalman filter, and then constructs the innovation sequence into a limited memory chi-square test statistic to detect spoofing interference. When the airborne GNSS spoofing causes a position deviation of 250 m and above, the designed algorithm can obtain a more accurate spoofing duration in simulation part. Finally, this paper gives the correct positioning information of the INS/GNSS/DME system based on the detection result of the proposed algorithm.
Global Navigation Satellite System (GNSS) spoofing causes the target receiver to generate incorrect positioning results. Using Inertial Navigation System (INS)-aided, constructing chi-squared test statistics based on the Kalman filter innovation sequence is an effective means to detect airborne GNSS spoofing. However, the algorithm cannot give the spoofing duration, causing the INS/GNSS system could not to determine whether the calculated positioning information is correct based on the algorithm. In this paper, a limited-memory chi-squared detection based on the reconstructed innovation sequence is proposed by using the Distance Measuring Equipment (DME) system. The algorithm uses the existing INS, GNSS and DME data to construct an innovation sequence that does not participate in Kalman filter, and then constructs the innovation sequence into a limited memory chi-square test statistic to detect spoofing interference. When the airborne GNSS spoofing causes a position deviation of 250 m and above, the designed algorithm can obtain a more accurate spoofing duration in simulation part. Finally, this paper gives the correct positioning information of the INS/GNSS/DME system based on the detection result of the proposed algorithm.
2022, 44(9): 3203-3211.
doi: 10.11999/JEIT210629
Abstract:
Precise Point Positioning (PPP) of Global Navigation Satellite System (GNSS) has the advantages of simple operation, low cost, and high positioning accuracy, and has been widely used in precise timing. Considering the problems of high combined noise of ionosphere-free PPP model, major navigation satellite system being Global Positioning System (GPS), and fewer real-time dynamic scenes in the existing research, the uncombined PPP model is used to study the accuracy of timing in BeiDou navigation Satellite system (BDS). Static and dynamic models of Kalman filter is used in parameter estimation. PPP processing strategies for static and real-time dynamic scenarios are proposed, and ultra-fast forecast ephemeris is used to ensure real time. The results show that: Under static and real-time dynamic conditions, the accuracy of precise timing with the uncombined PPP model is better than that of the ionosphere-free PPP model; Under static condition, considering that the positioning accuracy of BDS and GPS is equivalent, and the positioning accuracy of BDS in the Asia Pacific Region is even higher, the timing accuracy of BDS is slightly higher than that of GPS; Under real-time dynamic condition, the precise timing accuracy of BDS and GPS is both within 2 ns. But owing to Position Dilution Of Precision (PDOP), sudden changes of carrier speed, and environmental factors, the timing accuracy of BDS is slightly lower. And when the convergence is completed, the timing accuracy of the two systems is equivalent, and the clock biases are all within 0.3 m.
Precise Point Positioning (PPP) of Global Navigation Satellite System (GNSS) has the advantages of simple operation, low cost, and high positioning accuracy, and has been widely used in precise timing. Considering the problems of high combined noise of ionosphere-free PPP model, major navigation satellite system being Global Positioning System (GPS), and fewer real-time dynamic scenes in the existing research, the uncombined PPP model is used to study the accuracy of timing in BeiDou navigation Satellite system (BDS). Static and dynamic models of Kalman filter is used in parameter estimation. PPP processing strategies for static and real-time dynamic scenarios are proposed, and ultra-fast forecast ephemeris is used to ensure real time. The results show that: Under static and real-time dynamic conditions, the accuracy of precise timing with the uncombined PPP model is better than that of the ionosphere-free PPP model; Under static condition, considering that the positioning accuracy of BDS and GPS is equivalent, and the positioning accuracy of BDS in the Asia Pacific Region is even higher, the timing accuracy of BDS is slightly higher than that of GPS; Under real-time dynamic condition, the precise timing accuracy of BDS and GPS is both within 2 ns. But owing to Position Dilution Of Precision (PDOP), sudden changes of carrier speed, and environmental factors, the timing accuracy of BDS is slightly lower. And when the convergence is completed, the timing accuracy of the two systems is equivalent, and the clock biases are all within 0.3 m.
2022, 44(9): 3212-3219.
doi: 10.11999/JEIT210626
Abstract:
The Orbital Angular Momentum (OAM) satisfies orthogonality between each mode, which provides a new multiplexing dimension for wireless communication systems. At present, OAM communication still focuses on the Line of Sight (LoS) scenarios. The OAM Multiple Input Multiple Output (OAM-MIMO) communication system performance can be deteriorated by the non-ideal transmission conditions such as multipath and misalignment effects in the real scenarios. In order to improve the performance of the OAM-MIMO communication system, a millimeter-wave OAM-MIMO ten-rays channel in the actual transmission scenario is modelled in this paper; Then, the performance loss caused by multipath and misalignment effects are evaluated; Finally, a low-complexity Average Phase Compensation and Iterative Power Allocation (APC-IPA) joint optimization scheme is proposed to eliminate the phase deviation from the misalignment and multipath effects, and improve the capacity. The simulation results show that the proposed APC-IPA joint scheme increase effectively the channel capacity of the system when suffering from misalignment and multipath effects.
The Orbital Angular Momentum (OAM) satisfies orthogonality between each mode, which provides a new multiplexing dimension for wireless communication systems. At present, OAM communication still focuses on the Line of Sight (LoS) scenarios. The OAM Multiple Input Multiple Output (OAM-MIMO) communication system performance can be deteriorated by the non-ideal transmission conditions such as multipath and misalignment effects in the real scenarios. In order to improve the performance of the OAM-MIMO communication system, a millimeter-wave OAM-MIMO ten-rays channel in the actual transmission scenario is modelled in this paper; Then, the performance loss caused by multipath and misalignment effects are evaluated; Finally, a low-complexity Average Phase Compensation and Iterative Power Allocation (APC-IPA) joint optimization scheme is proposed to eliminate the phase deviation from the misalignment and multipath effects, and improve the capacity. The simulation results show that the proposed APC-IPA joint scheme increase effectively the channel capacity of the system when suffering from misalignment and multipath effects.
2022, 44(9): 3220-3230.
doi: 10.11999/JEIT210102
Abstract:
For Internet Of Vehicles(IOV), if all the computing tasks of vehicles are placed on the cloud platform, it can not meet the real-time requirement of information processing. Considering the mobile edge computing technology and task offloading method, the computing tasks are offloaded to the server near the edge of the device. However, in a dense environment, if all the tasks are offloaded, it would also bring large pressure to the edge server. A new method for offloading mobile edge computing tasks for vehicle users based on simulated annealing mechanism is proposed in this paper. By defining the user's task to calculate the offloading utility, comprehensively considering the time consumption and energy consumption, combining with simulated annealing, the utility of system offloading is optimized according to the current road density, and the user's offloading decision is changed. The offloading is executed locally or on the edge server, so that all users in a given environment can get high-quality service with low delay. The simulation results show that the algorithm can reduce the user task computing time and energy consumption at the same time.
For Internet Of Vehicles(IOV), if all the computing tasks of vehicles are placed on the cloud platform, it can not meet the real-time requirement of information processing. Considering the mobile edge computing technology and task offloading method, the computing tasks are offloaded to the server near the edge of the device. However, in a dense environment, if all the tasks are offloaded, it would also bring large pressure to the edge server. A new method for offloading mobile edge computing tasks for vehicle users based on simulated annealing mechanism is proposed in this paper. By defining the user's task to calculate the offloading utility, comprehensively considering the time consumption and energy consumption, combining with simulated annealing, the utility of system offloading is optimized according to the current road density, and the user's offloading decision is changed. The offloading is executed locally or on the edge server, so that all users in a given environment can get high-quality service with low delay. The simulation results show that the algorithm can reduce the user task computing time and energy consumption at the same time.
2022, 44(9): 3231-3240.
doi: 10.11999/JEIT210240
Abstract:
In order to solve the dynamic and unstable situation of each node in Unmanned Aerial Vehicle(UAV)cluster task scheduling, a multi computing node oriented task scheduling method is proposed, which can avoid task interruption as much as possible and has fault tolerance. Firstly, a task allocation strategy based on multi computing nodes is constructed to minimize the average completion time of tasks. Secondly, based on the probability distribution of the completion time of tasks and the retention time of edge computing nodes, the execution risk on task computing nodes is quantified as the extra overhead time. Finally, the sum of the completion time and the extra overhead time of tasks is calculated instead of the original completion time, a risk aware task allocation strategy is designed. In the simulation environment, the proposed task scheduling method is compared with three benchmark scheduling methods. The experimental results show that the proposed method can effectively reduce the average response time, the average execution times and the miss rate of task deadline. It is proved that the proposed method can reduce the additional cost of task rescheduling and re-execution, realize the task scheduling of distributed collaborative computing, and provide new technical support for UAV cluster network in complex scenarios.
In order to solve the dynamic and unstable situation of each node in Unmanned Aerial Vehicle(UAV)cluster task scheduling, a multi computing node oriented task scheduling method is proposed, which can avoid task interruption as much as possible and has fault tolerance. Firstly, a task allocation strategy based on multi computing nodes is constructed to minimize the average completion time of tasks. Secondly, based on the probability distribution of the completion time of tasks and the retention time of edge computing nodes, the execution risk on task computing nodes is quantified as the extra overhead time. Finally, the sum of the completion time and the extra overhead time of tasks is calculated instead of the original completion time, a risk aware task allocation strategy is designed. In the simulation environment, the proposed task scheduling method is compared with three benchmark scheduling methods. The experimental results show that the proposed method can effectively reduce the average response time, the average execution times and the miss rate of task deadline. It is proved that the proposed method can reduce the additional cost of task rescheduling and re-execution, realize the task scheduling of distributed collaborative computing, and provide new technical support for UAV cluster network in complex scenarios.
2022, 44(9): 3241-3248.
doi: 10.11999/JEIT210675
Abstract:
Deep Neural Network (DNN) is the major method in current Chinese word segmentation. However, its performance is significantly degraded when the network trained for one domain is used in other domains due to the Out Of Vocabulary (OOV) words and expression gaps. In this paper, a cross domain Chinese word segmentation system based on new word discovery is built to handle the OOV word and expression gap problems. An unsupervised new word discovery algorithm based on vector enhanced mutual information and weighted adjacency entropy, and a Chinese word segmentation model based on adversarial training are also proposed to improve the performance of the baseline system. Experimental results show that the proposed method is superior to the conventional methods in the OOV rates, precisions, recalls and F-scores.
Deep Neural Network (DNN) is the major method in current Chinese word segmentation. However, its performance is significantly degraded when the network trained for one domain is used in other domains due to the Out Of Vocabulary (OOV) words and expression gaps. In this paper, a cross domain Chinese word segmentation system based on new word discovery is built to handle the OOV word and expression gap problems. An unsupervised new word discovery algorithm based on vector enhanced mutual information and weighted adjacency entropy, and a Chinese word segmentation model based on adversarial training are also proposed to improve the performance of the baseline system. Experimental results show that the proposed method is superior to the conventional methods in the OOV rates, precisions, recalls and F-scores.
2022, 44(9): 3249-3257.
doi: 10.11999/JEIT210664
Abstract:
Focusing on the problem that the object features are not obvious in complex urban surveillance scenes due to large object size changes, object occlusion and weather influence, a multi-scale full-scene surveillance object detection method based on attention mechanism is proposed. In this paper, a multi-scale detection network structure based on Yolov5s model is designed to improve the adaptability of the network to the changes of object size. Meanwhile, a feature extraction module based on attention mechanism is constructed to obtain channel level weight of features through network learning, which enhances the object features, suppresses the background features, and improves the network extraction capability of features. The initial anchor frame size of the full-scene surveillance dataset is calculated by the K-means clustering algorithm to accelerate the model convergence while improving the detection accuracy. On the COCO dataset, the mean Average Precision (mAP) is improved by 3.7%, and the mAP50 is improved by 4.7% compared with the basic network, and the model inference time is only 3.8 ms. In the full-scene surveillance dataset, the mAP50 reaches 89.6% and the fps is 154 frames per second when processing the surveillance video, which meets the real-time detection requirements of the surveillance scene.
Focusing on the problem that the object features are not obvious in complex urban surveillance scenes due to large object size changes, object occlusion and weather influence, a multi-scale full-scene surveillance object detection method based on attention mechanism is proposed. In this paper, a multi-scale detection network structure based on Yolov5s model is designed to improve the adaptability of the network to the changes of object size. Meanwhile, a feature extraction module based on attention mechanism is constructed to obtain channel level weight of features through network learning, which enhances the object features, suppresses the background features, and improves the network extraction capability of features. The initial anchor frame size of the full-scene surveillance dataset is calculated by the K-means clustering algorithm to accelerate the model convergence while improving the detection accuracy. On the COCO dataset, the mean Average Precision (mAP) is improved by 3.7%, and the mAP50 is improved by 4.7% compared with the basic network, and the model inference time is only 3.8 ms. In the full-scene surveillance dataset, the mAP50 reaches 89.6% and the fps is 154 frames per second when processing the surveillance video, which meets the real-time detection requirements of the surveillance scene.
2022, 44(9): 3258-3265.
doi: 10.11999/JEIT210693
Abstract:
In view of the problems of slow calculation speed of digital holographic reconstruction algorithm, weak real-time application ability and poor cross-platform portability of existing GPU acceleration strategies, a scheme is proposed based on Open Computing Language (OpenCL) architecture to improve the execution efficiency of digital holographic reconstruction algorithm. In more details, the heterogeneous collaborative computing capabilities of the OpenCL architecture is fully used to design a CPU+GPU heterogeneous operation for the digital holographic convolutional reconstruction algorithm, which is programmed in the data parallel mode. The tests are carried out on the digital holograms in various image resolutions and on the different GPU acceleration platforms. The results indicate that the average execution time of this acceleration strategy is approximately an order of magnitude lower than that of the CPU, the highest total acceleration ratio is 54.2, and the parallel computing acceleration ratio even reaches up to 94.7. Characterized by a scale growth, good cross-platform portability and significant acceleration efficiency, it is more suitable for the engineering realization of digital holographic technology, especially in the real-time applications.
In view of the problems of slow calculation speed of digital holographic reconstruction algorithm, weak real-time application ability and poor cross-platform portability of existing GPU acceleration strategies, a scheme is proposed based on Open Computing Language (OpenCL) architecture to improve the execution efficiency of digital holographic reconstruction algorithm. In more details, the heterogeneous collaborative computing capabilities of the OpenCL architecture is fully used to design a CPU+GPU heterogeneous operation for the digital holographic convolutional reconstruction algorithm, which is programmed in the data parallel mode. The tests are carried out on the digital holograms in various image resolutions and on the different GPU acceleration platforms. The results indicate that the average execution time of this acceleration strategy is approximately an order of magnitude lower than that of the CPU, the highest total acceleration ratio is 54.2, and the parallel computing acceleration ratio even reaches up to 94.7. Characterized by a scale growth, good cross-platform portability and significant acceleration efficiency, it is more suitable for the engineering realization of digital holographic technology, especially in the real-time applications.
2022, 44(9): 3266-3276.
doi: 10.11999/JEIT210606
Abstract:
To improve the performance of monaural speech separation, a monaural speech separation method based on deep learning feature fusion and joint constraints is proposed. The loss function of the traditional separation algorithm based on deep learning only considers the error between the predicted value and the true one, which makes the error between the separated speech and the pure speech larger. To combat it, a new joint constrained loss function is proposed, which not only constrains the error between the predicted value and the true one of ideal ratio mask, but also penalizes the error of the corresponding amplitude spectrum. In addition, to make full use of the complementarity of multiple features, a Convolutional Neural Network (CNN) structure with feature fusion layer is proposed, which extracts the depth feature of the multi-channel input feature, and then fuses the depth feature and the acoustic feature in the fusion layer to train the separation model. The fused separation feature contains abundant acoustic information and has a strong acoustic representative ability, which makes the mask predicted by the separation model more accurate. The experimental results show that from Signal Distortion Ratio (SDR), Perceptual Evaluation of Speech Quality (PESQ) and Short-Time Objective Intelligibility (STOI), compared with other excellent speech separation methods based on deep learning, the proposed method can separate the mixed speech more effectively.
To improve the performance of monaural speech separation, a monaural speech separation method based on deep learning feature fusion and joint constraints is proposed. The loss function of the traditional separation algorithm based on deep learning only considers the error between the predicted value and the true one, which makes the error between the separated speech and the pure speech larger. To combat it, a new joint constrained loss function is proposed, which not only constrains the error between the predicted value and the true one of ideal ratio mask, but also penalizes the error of the corresponding amplitude spectrum. In addition, to make full use of the complementarity of multiple features, a Convolutional Neural Network (CNN) structure with feature fusion layer is proposed, which extracts the depth feature of the multi-channel input feature, and then fuses the depth feature and the acoustic feature in the fusion layer to train the separation model. The fused separation feature contains abundant acoustic information and has a strong acoustic representative ability, which makes the mask predicted by the separation model more accurate. The experimental results show that from Signal Distortion Ratio (SDR), Perceptual Evaluation of Speech Quality (PESQ) and Short-Time Objective Intelligibility (STOI), compared with other excellent speech separation methods based on deep learning, the proposed method can separate the mixed speech more effectively.
2022, 44(9): 3277-3288.
doi: 10.11999/JEIT210654
Abstract:
In supervised speech enhancement, contextual information has an important influence on the estimation of target speech. In order to obtain richer global related features of speech, a new convolution network for speech enhancement on the premise of the smallest possible parameters is designed in this paper. The proposed network contains three parts: encode layer, transfer layer and decode layer. The encode and decode part propose a Two-Dimensional Asymmetric Dilated Residual (2D-ADR) module, which can significantly reduce training parameters and expand the receptive field, and improve the model’s ability to obtain contextual information. The transfer layer proposes a One-Dimensional Gating Dilated Residual (1D-GDR) module, which combines dilated convolution, residual learning and gating mechanism to transfer selectively features and obtain more time-related information. Moreover, the eight 1D-GDR modules are stacked by a dense skip-connection way to enhance the information flow between layers and provide more gradient propagation path. Finally, the corresponding encode and decode layer is connected by skip-connection and attention mechanism is introduced to make the decoding process obtain more robust underlying features. In the experimental part, different parameter settings and comparison methods are used to verify the effectiveness and robustness of the network. By training and testing under 28 kinds of noise, compared with other methods, the proposed method has achieved better objective and subjective metrics with 1.25 million parameters, and has better enhancement effect and generalization ability.
In supervised speech enhancement, contextual information has an important influence on the estimation of target speech. In order to obtain richer global related features of speech, a new convolution network for speech enhancement on the premise of the smallest possible parameters is designed in this paper. The proposed network contains three parts: encode layer, transfer layer and decode layer. The encode and decode part propose a Two-Dimensional Asymmetric Dilated Residual (2D-ADR) module, which can significantly reduce training parameters and expand the receptive field, and improve the model’s ability to obtain contextual information. The transfer layer proposes a One-Dimensional Gating Dilated Residual (1D-GDR) module, which combines dilated convolution, residual learning and gating mechanism to transfer selectively features and obtain more time-related information. Moreover, the eight 1D-GDR modules are stacked by a dense skip-connection way to enhance the information flow between layers and provide more gradient propagation path. Finally, the corresponding encode and decode layer is connected by skip-connection and attention mechanism is introduced to make the decoding process obtain more robust underlying features. In the experimental part, different parameter settings and comparison methods are used to verify the effectiveness and robustness of the network. By training and testing under 28 kinds of noise, compared with other methods, the proposed method has achieved better objective and subjective metrics with 1.25 million parameters, and has better enhancement effect and generalization ability.
2022, 44(9): 3289-3298.
doi: 10.11999/JEIT210605
Abstract:
Objective reduction is an important research direction in many-objective optimization. Through proper algorithm design, it can eliminate some redundant objectives to achieve the effect of greatly simplifying an optimization problem. Among the many-objective optimization problems with redundant objectives, the problems with nonlinear Pareto-Front are the most common and most difficult to tackle. In this paper, an algorithm based on Decomposition and Hyperplane Approximation (DHA) is proposed to deal with objective reduction problems with nonlinear Pareto-Front. The proposed algorithm decomposes a population with nonlinear geometric distribution into several subsets with approximate linear distribution in the process of evolution, and uses a hyperplane with sparse coefficients combined with some perturbation terms to fit these subsets, and then it extractes an essential objective set of original problem based on the coefficients of the fitting hyperplane. In order to test the performance of the proposed algorithm, this study compares it with some state-of-the-art algorithms in the benchmark DTLZ5(I, m), WFG3(I, m) and MAOP(I, m). The experimental results show that the proposed algorithm has good performance both in the problems with linear and nonlinear Pareto-Front.
Objective reduction is an important research direction in many-objective optimization. Through proper algorithm design, it can eliminate some redundant objectives to achieve the effect of greatly simplifying an optimization problem. Among the many-objective optimization problems with redundant objectives, the problems with nonlinear Pareto-Front are the most common and most difficult to tackle. In this paper, an algorithm based on Decomposition and Hyperplane Approximation (DHA) is proposed to deal with objective reduction problems with nonlinear Pareto-Front. The proposed algorithm decomposes a population with nonlinear geometric distribution into several subsets with approximate linear distribution in the process of evolution, and uses a hyperplane with sparse coefficients combined with some perturbation terms to fit these subsets, and then it extractes an essential objective set of original problem based on the coefficients of the fitting hyperplane. In order to test the performance of the proposed algorithm, this study compares it with some state-of-the-art algorithms in the benchmark DTLZ5(I, m), WFG3(I, m) and MAOP(I, m). The experimental results show that the proposed algorithm has good performance both in the problems with linear and nonlinear Pareto-Front.
2022, 44(9): 3299-3307.
doi: 10.11999/JEIT210634
Abstract:
According to the target identity switch and tracking trajectory interruption, a multi-pedestrian tracking algorithm based on Convolutional Block Attention Module (CBAM) and anchor-free detection network is proposed. Firstly, attention mechanism is introduced to HrnetV2′s stem stage to extract more expressive features, thus strengthening the training of re-recognition branch. Secondly, in order to improve the operation speed of algorithm, detection task and recognition one share feature weights and are carried out simultaneously. Meanwhile, the convolutional channel’s number and parameter amount are reduced in the head network. Finally, the network is fully trained with proper parameters, and the algorithm is validated by multiple test sets. Experimental results show that compared with FairMOT, the accuracy of the proposed algorithm on 2DMOT15, MOT17 and MOT20 data sets is improved by 1.1%, 1.1%, 0.2% respectively, and the speed is improved by 0.82, 0.88 and 0.41 fps respectively. Compared with other mainstream algorithms, the proposed algorithm has the least number of target identity switching. The proposed algorithm improves effectively real-time performance of network model, which could be better applied to the scenes with severe occlusion.
According to the target identity switch and tracking trajectory interruption, a multi-pedestrian tracking algorithm based on Convolutional Block Attention Module (CBAM) and anchor-free detection network is proposed. Firstly, attention mechanism is introduced to HrnetV2′s stem stage to extract more expressive features, thus strengthening the training of re-recognition branch. Secondly, in order to improve the operation speed of algorithm, detection task and recognition one share feature weights and are carried out simultaneously. Meanwhile, the convolutional channel’s number and parameter amount are reduced in the head network. Finally, the network is fully trained with proper parameters, and the algorithm is validated by multiple test sets. Experimental results show that compared with FairMOT, the accuracy of the proposed algorithm on 2DMOT15, MOT17 and MOT20 data sets is improved by 1.1%, 1.1%, 0.2% respectively, and the speed is improved by 0.82, 0.88 and 0.41 fps respectively. Compared with other mainstream algorithms, the proposed algorithm has the least number of target identity switching. The proposed algorithm improves effectively real-time performance of network model, which could be better applied to the scenes with severe occlusion.
2022, 44(9): 3308-3319.
doi: 10.11999/JEIT210623
Abstract:
In this paper, an improved and adaptive transfer learning algorithm is proposed for mobile application traffic recognition filed, which maps the sample features of source domain and target domain into high-dimensional feature space to minimize the marginal distribution and conditional distribution distances of the domains. A probabilistic model is presented to judge and calculate the difference between marginal distribution and conditional distribution between the domains. It can determine the degree of classification category and calculate quantitively the balance factor\begin{document}$ \mu $\end{document} ![]()
![]()
In this paper, an improved and adaptive transfer learning algorithm is proposed for mobile application traffic recognition filed, which maps the sample features of source domain and target domain into high-dimensional feature space to minimize the marginal distribution and conditional distribution distances of the domains. A probabilistic model is presented to judge and calculate the difference between marginal distribution and conditional distribution between the domains. It can determine the degree of classification category and calculate quantitively the balance factor
2022, 44(9): 3320-3327.
doi: 10.11999/JEIT210648
Abstract:
To solve the problem of reconstruction image artifacts caused by the problem of "underdetermined" and the “soft field“ effect in the visualization process of electrical impedance tomography, an unsupervised image quality evaluation index based on neighborhood information and fast Fuzzy C-Means clustering (fast FCM) is proposed. Based on this evaluation index and Tikhonov regularization algorithm, a reconstruction image artifact optimization algorithm TR-NC is proposed. Simulation results show that the proposed algorithm can effectively correct artifacts in the reconstructed image, and the correlation coefficient of the modified reconstructed image has increased by 18.45% on average, and the relative error has reduced by 22.2% on average. Simulation experimental results show that the proposed algorithm can accurately detect the target when the change rate of target conductivity is more than 30%. It is shown that compared with the traditional Tikhonov regularization algorithm, the proposed modified algorithm TR-NC has been significantly improved in the number and position accuracy of reconstruction image targets, which provides a new imaging theoretical basis and technical reference for the application of electrical tomography technology to medical and industrial fields.
To solve the problem of reconstruction image artifacts caused by the problem of "underdetermined" and the “soft field“ effect in the visualization process of electrical impedance tomography, an unsupervised image quality evaluation index based on neighborhood information and fast Fuzzy C-Means clustering (fast FCM) is proposed. Based on this evaluation index and Tikhonov regularization algorithm, a reconstruction image artifact optimization algorithm TR-NC is proposed. Simulation results show that the proposed algorithm can effectively correct artifacts in the reconstructed image, and the correlation coefficient of the modified reconstructed image has increased by 18.45% on average, and the relative error has reduced by 22.2% on average. Simulation experimental results show that the proposed algorithm can accurately detect the target when the change rate of target conductivity is more than 30%. It is shown that compared with the traditional Tikhonov regularization algorithm, the proposed modified algorithm TR-NC has been significantly improved in the number and position accuracy of reconstruction image targets, which provides a new imaging theoretical basis and technical reference for the application of electrical tomography technology to medical and industrial fields.
2022, 44(9): 3328-3334.
doi: 10.11999/JEIT210646
Abstract:
Graph coloring problem divides the vertices of the graph into disjoint sets and the vertices in each set are assigned by the same color under the constraints that adjacent vertices can not be assigned the same color and the number of colors is the smallest. Since graph coloring problem belongs to the class of NP-complete problems, the complexity for solving the graph coloring problem increases exponentially with the number of vertices. The performance of solving graph coloring problem by general-purpose processors decreases significantly when the number of vertices is large enough. This paper implements a dedicated hardware accelerator for solving graph coloring problem based on Field Programmable Gate Array (FPGA). Firstly, by utilizing the rule of FPGA modular design, the hardware architecture for solving graph coloring problem based on the backtracking is proposed and implemented. Secondly, the relationship between the resource consumption of FPGA and the number of vertices is analyzed. Finally, the general-purpose processor and FPGA can communicate through universal asynchronous transmitter-receiver protocol. The experimental results show that the running time of graph coloring algorithm based on FPGA is about an order of magnitude smaller than that of graph coloring algorithm based on software on general-purpose processors. Besides, the resource consumption of FPGA is linear with the number of vertices and the time consumption at each iteration is independent of the number of vertices.
Graph coloring problem divides the vertices of the graph into disjoint sets and the vertices in each set are assigned by the same color under the constraints that adjacent vertices can not be assigned the same color and the number of colors is the smallest. Since graph coloring problem belongs to the class of NP-complete problems, the complexity for solving the graph coloring problem increases exponentially with the number of vertices. The performance of solving graph coloring problem by general-purpose processors decreases significantly when the number of vertices is large enough. This paper implements a dedicated hardware accelerator for solving graph coloring problem based on Field Programmable Gate Array (FPGA). Firstly, by utilizing the rule of FPGA modular design, the hardware architecture for solving graph coloring problem based on the backtracking is proposed and implemented. Secondly, the relationship between the resource consumption of FPGA and the number of vertices is analyzed. Finally, the general-purpose processor and FPGA can communicate through universal asynchronous transmitter-receiver protocol. The experimental results show that the running time of graph coloring algorithm based on FPGA is about an order of magnitude smaller than that of graph coloring algorithm based on software on general-purpose processors. Besides, the resource consumption of FPGA is linear with the number of vertices and the time consumption at each iteration is independent of the number of vertices.
2022, 44(9): 3335-3342.
doi: 10.11999/JEIT210638
Abstract:
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
10.1007/978-981-15-0058-9-27 ). The cipher has 15 rounds in total and adopts TYPE-II generalized feistel network. The ability of the \begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}${2^{76}}$\end{document} ![]()
![]()
\begin{document}${2^{48}}$\end{document} ![]()
![]()
\begin{document}${2^{48}}$\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}${2^{49}}$\end{document} ![]()
![]()
\begin{document}${2^{64}}$\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}$ {\mu ^{\text{2}}} $\end{document} ![]()
![]()
\begin{document}${{\text{2}}^{{{ - 39}}}}$\end{document} ![]()
![]()
\begin{document}${2^{ - 3{\text{6}}}}$\end{document} ![]()
![]()
