With the rapid development of Cloud-Edge-End(CEE)computing,the demand for secure and lightweight communication protocols is increasingly critical,particularly for latency-sensitive applications such as smart manufactu...With the rapid development of Cloud-Edge-End(CEE)computing,the demand for secure and lightweight communication protocols is increasingly critical,particularly for latency-sensitive applications such as smart manufacturing,healthcare,and real-time monitoring.While traditional cryptographic schemes offer robust protection,they often impose excessive computational and energy overhead,rendering them unsuitable for use in resource-constrained edge and end devices.To address these challenges,in this paper,we propose a novel lightweight encryption framework,namely Dynamic Session Key Allocation with Time-Indexed Ascon(DSKA-TIA).Built upon the NIST-endorsed Ascon algorithm,the DSKA-TIA introduces a time-indexed session key generation mechanism that derives unique,ephemeral keys for each communication round.The scheme supports bidirectional key separation to isolate uplink and downlink data,thereby minimizing the risk of key reuse and compromise.Additionally,mutual authentication is integrated through nonce-based validation and one-time token exchanges,ensuring entity legitimacy and protection against impersonation and replay attacks.We validate the performance of DSKA-TIA through implementation on a resource-constrained microcontroller platform.Results show that our scheme achieves significantly lower latency and computational cost compared to baseline schemes such as AES and standard Ascon.Security analysis demonstrates high entropy in key generation,resistance to brute-force and replay attacks,and robustness against eavesdropping and key compromise.The protocol also exhibits resilience to quantum computing threats by relying on symmetric encryption principles and randomized key selection.Given its efficiency,scalability,and temporal security enhancements,DSKA-TIA is well-suited for real-time,secure communication in heterogeneous CEE environments.Future work will explore post-quantum extensions and deployment in domains such as smart agriculture and edge-based healthcare.展开更多
Predicting user states in future and rendering visual feedbacks accordingly can effectively reduce the visual experienced delay in the tactile Internet(TI). However, most works omit the fact that different parts in an...Predicting user states in future and rendering visual feedbacks accordingly can effectively reduce the visual experienced delay in the tactile Internet(TI). However, most works omit the fact that different parts in an image may have distinct prediction requirements, based on which different prediction models can be used in the predicting process, and then it can further improve predicting quality especially under resources-limited environment. In this paper, a hybrid prediction scheme is proposed for the visual feedbacks in a typical TI scenario with mixed visuo-haptic interactions, in which haptic traffic needs sufficient wireless resources to meet its stringent communication requirement, leaving less radio resources for the visual feedback. First, the minimum required number of radio resources for haptic traffic is derived based on the haptic communication requirements, and wireless resources are allocated to the haptic and visual traffics afterwards. Then, a grouping strategy is designed based on the deep neural network(DNN) to allocate different parts from an image feedback into two groups to use different prediction models, which jointly considers the prediction deviation thresholds, latency and reliability requirements, and the bit sizes of different image parts. Simulations show that, the hybrid prediction scheme can further reduce the visual experienced delay under haptic traffic requirements compared with existing strategies.展开更多
Over the past few years,video live streaming has gained immense popularity as a leading internet application.In current solutions offered by cloud service providers,the Group of Pictures(GOP)length of the video source...Over the past few years,video live streaming has gained immense popularity as a leading internet application.In current solutions offered by cloud service providers,the Group of Pictures(GOP)length of the video source often significantly impacts end-to-end(E2E)latency.However,designing an optimized GOP structure to reduce this effect remains a significant challenge.This paper presents two key contributions.First,it explores how the GOP length at the video source influences E2E latency in mainstream cloud streaming services.Experimental results reveal that the mean E2E latency increases linearly with longer GOP lengths.Second,this paper proposes EGOP(an Enhanced GOP structure)that can be implemented in streaming media servers.Experiments demonstrate that EGOP maintains a consistent E2E latency,unaffected by the GOP length of the video source.Specifically,even with a GOP length of 10 s,the E2E latency remains at 1.35 s,achieving a reduction of 6.98 s compared to Volcano-Engine(the live streaming service provider for TikTok).This makes EGOP a promising solution for low-latency live streaming.展开更多
The 3GPP standard defines the requirements for next-generation wireless networks,with particular attention to Ultra-Reliable Low-Latency Communications(URLLC),critical for applications such as Unmanned Aerial Vehicles...The 3GPP standard defines the requirements for next-generation wireless networks,with particular attention to Ultra-Reliable Low-Latency Communications(URLLC),critical for applications such as Unmanned Aerial Vehicles(UAVs).In this context,Non-Orthogonal Multiple Access(NOMA)has emerged as a promising technique to improve spectrum efficiency and user fairness by allowing multiple users to share the same frequency resources.However,optimizing key parameters–such as beamforming,rate allocation,and UAV trajectory–presents significant challenges due to the nonconvex nature of the problem,especially under stringent URLLC constraints.This paper proposes an advanced deep learning-driven approach to address the resulting complex optimization challenges.We formulate a downlink multiuser UAV,Rate-Splitting Multiple Access(RSMA),and Multiple Input Multiple Output(MIMO)system aimed at maximizing the achievable rate under stringent constraints,including URLLC quality-of-service(QoS),power budgets,rate allocations,and UAV trajectory limitations.Due to the highly nonconvex nature of the optimization problem,we introduce a novel distributed deep reinforcement learning(DRL)framework based on dual-agent deep deterministic policy gradient(DA-DDPG).The proposed framework leverages inception-inspired and deep unfolding architectures to improve feature extraction and convergence in beamforming and rate allocation.For UAV trajectory optimization,we design a dedicated actor-critic agent using a fully connected deep neural network(DNN),further enhanced through incremental learning.Simulation results validate the effectiveness of our approach,demonstrating significant performance gains over existing methods and confirming its potential for real-time URLLC in next-generation UAV communication networks.展开更多
With the emerging diverse applications in data centers,the demands on quality of service in data centers also become diverse,such as high throughput of elephant flows and low latency of deadline-sensitive flows.Howeve...With the emerging diverse applications in data centers,the demands on quality of service in data centers also become diverse,such as high throughput of elephant flows and low latency of deadline-sensitive flows.However,traditional TCPs are ill-suited to such situations and always result in the inefficiency(e.g.missing the flow deadline,inevitable throughput collapse)of data transfers.This further degrades the user-perceived quality of service(QoS)in data centers.To reduce the flow completion time of mice and deadline-sensitive flows along with promoting the throughput of elephant flows,an efficient and deadline-aware priority-driven congestion control(PCC)protocol,which grants mice and deadline-sensitive flows the highest priority,is proposed in this paper.Specifically,PCC computes the priority of different flows according to the size of transmitted data,the remaining data volume,and the flows’deadline.Then PCC adjusts the congestion window according to the flow priority and the degree of network congestion.Furthermore,switches in data centers control the input/output of packets based on the flow priority and the queue length.Different from existing TCPs,to speed up the data transfers of mice and deadline-sensitive flows,PCC provides an effective method to compute and encode the flow priority explicitly.According to the flow priority,switches can manage packets efficiently and ensure the data transfers of high priority flows through a weighted priority scheduling with minor modification.The experimental results prove that PCC can improve the data transfer performance of mice and deadline-sensitive flows while guaranting the throughput of elephant flows.展开更多
In response to the requirements for large-scale device access and ultra-reliable and low-latency communication in the power internet of things,unmanned aerial vehicle-assisted multi-access edge computing can be used t...In response to the requirements for large-scale device access and ultra-reliable and low-latency communication in the power internet of things,unmanned aerial vehicle-assisted multi-access edge computing can be used to realize flexible access to power services and update large amounts of information in a timely manner.By considering factors such as machine communication traffic,MAC competition access,and information freshness,this paper develops a cross-layer computing framework in which the peak Age of Information(Ao I)provides a statistical delay boundary in the finite blocklength regime.We also propose a deep machine learning-based multi-access edge computing offloading algorithm.First,a traffic arrival model is established in which the time interval follows the Beta distribution,and then a business service model is proposed based on the carrier sense multiple access with collision avoidance algorithm.The peak Ao I boundary performance of multiple access is evaluated according to stochastic network calculus theory.Finally,an unmanned aerial vehicle-assisted multilevel offloading model with cache is designed,in which the peak Ao I violation probability and energy consumption provide the optimization goals.The optimal offloading strategy is obtained using deep reinforcement learning.Compared with baseline schemes based on non-cooperative game theory with stochastic learning automata and random edge unloading,the proposed algorithm improves the overall performance by approximately 3.52%and 20.73%,respectively,and provides superior deterministic offloading performance by using the peak Ao I boundary.展开更多
Enhanced mobile broadband(eMBB)and ultra-reliable low-latency communication(URLLC)are two critical services in 5G mobile networks.While there has been extensive research on their coexistence,few studies have considere...Enhanced mobile broadband(eMBB)and ultra-reliable low-latency communication(URLLC)are two critical services in 5G mobile networks.While there has been extensive research on their coexistence,few studies have considered the impact of bursty URLLC on their coexistence performance.In this paper,we propose a method to allocate computing and radio resources for coexisting eMBB and bursty URLLC services by preempting both computing queues in the base station(BS)and time-frequency resources at the air interface.Specifically,we first divide the computing resources at the BS into a shared part for both URLLC and eMBB users and an exclusive part only for eMBB users,and propose a queuing mechanism with preemptive-resume priority for accessing the shared computing resources.Furthermore,we propose a preemptive puncturing method and a threshold-based queuing mechanism in the air interface to enable the multiplexing of eMBB and URLLC on shared time-frequency resources.We analytically derive the average queuing delay,average computation delay,and average transmission delay of eMBB and URLLC packets.Based on this analysis,we formulate a mixed-integer nonlinear programming problem to minimize the average delay of URLLC packets while satisfying the average delay and throughput requirements of eMBB by jointly optimizing the eMBB subcarrier allocation,the URLLC subcarrier scheduling and the computing resource allocation.We decompose this problem into three subproblems and solve them alternately using a block coordinate descent algorithm.Numerical results show that our proposed method reduces the outage probability and average delay of URLLC compared to the existing works.展开更多
The coexistence of ultra-reliable low-latency communication(URLLC)and enhanced mobile broadband(eMBB)services in 5G-based industrial wireless networks(IWNs)poses significant resource slicing challenges due to their in...The coexistence of ultra-reliable low-latency communication(URLLC)and enhanced mobile broadband(eMBB)services in 5G-based industrial wireless networks(IWNs)poses significant resource slicing challenges due to their inherent performance requirement conflicts.To address this challenge,this paper proposes a puncturing method that uses a model-aided deep reinforcement learning(DRL)algorithm for URLLC over eMBB services in uplink 5G networks.First,a puncturing-based optimization problem is formulated to maximize the eMBB accumulated rate under strict URLLC latency and reliability constraints.Next,we design a random repetition coding-based contention(RRCC)scheme for sporadic URLLC traffic and derive its analytical reliability model.To jointly optimize the scheduling parameters of URLLC and eMBB,a DRL solution based on the reliability model is developed,which is capable of dynamically adapting to changing environments.The accelerated convergence of the model-aided DRL algorithm is demonstrated using simulations,and the superiority in resource efficiency of the proposed method over existing approaches is validated.展开更多
Wirreless sensor networks are being widely researched and are expected to be used in several scenarios. On the leading edge of treads, on-demand, high-reliability, and low-latency routing protocol is desirable for ind...Wirreless sensor networks are being widely researched and are expected to be used in several scenarios. On the leading edge of treads, on-demand, high-reliability, and low-latency routing protocol is desirable for indoor environment applications. This article proposes a routing scheme called robust multi-path routing that establishes and uses multiple node-disjoint routes. Providing multiple routes helps to reduce the route recovery process and control the message overhead. The performance comparison of this protocol with dynamic source routing （DSR） by OPNET simulations shows that this protocol is able to achieve a remarkable improvement in the packet delivery ratio and average end-to-end delay.展开更多
In data center, applications of big data analytics pose a big challenge to massive storage systems. It is signif- icant to achieve high availability, high performance and high scalability for PB-scale or EB-scale stor...In data center, applications of big data analytics pose a big challenge to massive storage systems. It is signif- icant to achieve high availability, high performance and high scalability for PB-scale or EB-scale storage systems. Meta- data server （MDS） cluster architecture is one of the most effective solutions to meet the requirements of applications in data center. Workload migration can achieve load balance and energy saving of duster systems. In this paper, a hybrid workload migration mechanism of MDS cluster is proposed and named as HWM. In HWM, workload of MDS is classi- fied into two categories： metadata service and state service, and they can be migrated rapidly from a source MDS to a target MDS in different ways. Firstly, in metadata service migration, all the dirty metadata of one sub file system is flushed to a shared storage pool by the source MDS, and then is loaded by the target MDS. Secondly, in state service mi- gration, all the states of that sub file system are migrated from source MDS to target MDS through network at file granular- ity, and then all of the related structures of these states are reconstructed in target MDS. Thirdly, in the process of work- load migration, instead of blocking client requests, the source MDS can decide which MDS will respond to each request according to the operation type and the migration stage. The proposed mechanism is implemented in the Blue Whale MDS cluster. The performance measurements show that the HWM mechanism is efficient to migrate the workload of a MDS cluster system and provides low-latency access to metadata and states.展开更多
文摘With the rapid development of Cloud-Edge-End(CEE)computing,the demand for secure and lightweight communication protocols is increasingly critical,particularly for latency-sensitive applications such as smart manufacturing,healthcare,and real-time monitoring.While traditional cryptographic schemes offer robust protection,they often impose excessive computational and energy overhead,rendering them unsuitable for use in resource-constrained edge and end devices.To address these challenges,in this paper,we propose a novel lightweight encryption framework,namely Dynamic Session Key Allocation with Time-Indexed Ascon(DSKA-TIA).Built upon the NIST-endorsed Ascon algorithm,the DSKA-TIA introduces a time-indexed session key generation mechanism that derives unique,ephemeral keys for each communication round.The scheme supports bidirectional key separation to isolate uplink and downlink data,thereby minimizing the risk of key reuse and compromise.Additionally,mutual authentication is integrated through nonce-based validation and one-time token exchanges,ensuring entity legitimacy and protection against impersonation and replay attacks.We validate the performance of DSKA-TIA through implementation on a resource-constrained microcontroller platform.Results show that our scheme achieves significantly lower latency and computational cost compared to baseline schemes such as AES and standard Ascon.Security analysis demonstrates high entropy in key generation,resistance to brute-force and replay attacks,and robustness against eavesdropping and key compromise.The protocol also exhibits resilience to quantum computing threats by relying on symmetric encryption principles and randomized key selection.Given its efficiency,scalability,and temporal security enhancements,DSKA-TIA is well-suited for real-time,secure communication in heterogeneous CEE environments.Future work will explore post-quantum extensions and deployment in domains such as smart agriculture and edge-based healthcare.
基金supported by the National Natural Science Foundation of China (61771070)。
文摘Predicting user states in future and rendering visual feedbacks accordingly can effectively reduce the visual experienced delay in the tactile Internet(TI). However, most works omit the fact that different parts in an image may have distinct prediction requirements, based on which different prediction models can be used in the predicting process, and then it can further improve predicting quality especially under resources-limited environment. In this paper, a hybrid prediction scheme is proposed for the visual feedbacks in a typical TI scenario with mixed visuo-haptic interactions, in which haptic traffic needs sufficient wireless resources to meet its stringent communication requirement, leaving less radio resources for the visual feedback. First, the minimum required number of radio resources for haptic traffic is derived based on the haptic communication requirements, and wireless resources are allocated to the haptic and visual traffics afterwards. Then, a grouping strategy is designed based on the deep neural network(DNN) to allocate different parts from an image feedback into two groups to use different prediction models, which jointly considers the prediction deviation thresholds, latency and reliability requirements, and the bit sizes of different image parts. Simulations show that, the hybrid prediction scheme can further reduce the visual experienced delay under haptic traffic requirements compared with existing strategies.
基金supported by Henan Province Major Science and Technology Project(241100210100).
文摘Over the past few years,video live streaming has gained immense popularity as a leading internet application.In current solutions offered by cloud service providers,the Group of Pictures(GOP)length of the video source often significantly impacts end-to-end(E2E)latency.However,designing an optimized GOP structure to reduce this effect remains a significant challenge.This paper presents two key contributions.First,it explores how the GOP length at the video source influences E2E latency in mainstream cloud streaming services.Experimental results reveal that the mean E2E latency increases linearly with longer GOP lengths.Second,this paper proposes EGOP(an Enhanced GOP structure)that can be implemented in streaming media servers.Experiments demonstrate that EGOP maintains a consistent E2E latency,unaffected by the GOP length of the video source.Specifically,even with a GOP length of 10 s,the E2E latency remains at 1.35 s,achieving a reduction of 6.98 s compared to Volcano-Engine(the live streaming service provider for TikTok).This makes EGOP a promising solution for low-latency live streaming.
基金supported by the Deputyship of Research&Innovation,Ministry of Education in Saudi Arabia for funding this research work through the project number RI-44-0291.
文摘The 3GPP standard defines the requirements for next-generation wireless networks,with particular attention to Ultra-Reliable Low-Latency Communications(URLLC),critical for applications such as Unmanned Aerial Vehicles(UAVs).In this context,Non-Orthogonal Multiple Access(NOMA)has emerged as a promising technique to improve spectrum efficiency and user fairness by allowing multiple users to share the same frequency resources.However,optimizing key parameters–such as beamforming,rate allocation,and UAV trajectory–presents significant challenges due to the nonconvex nature of the problem,especially under stringent URLLC constraints.This paper proposes an advanced deep learning-driven approach to address the resulting complex optimization challenges.We formulate a downlink multiuser UAV,Rate-Splitting Multiple Access(RSMA),and Multiple Input Multiple Output(MIMO)system aimed at maximizing the achievable rate under stringent constraints,including URLLC quality-of-service(QoS),power budgets,rate allocations,and UAV trajectory limitations.Due to the highly nonconvex nature of the optimization problem,we introduce a novel distributed deep reinforcement learning(DRL)framework based on dual-agent deep deterministic policy gradient(DA-DDPG).The proposed framework leverages inception-inspired and deep unfolding architectures to improve feature extraction and convergence in beamforming and rate allocation.For UAV trajectory optimization,we design a dedicated actor-critic agent using a fully connected deep neural network(DNN),further enhanced through incremental learning.Simulation results validate the effectiveness of our approach,demonstrating significant performance gains over existing methods and confirming its potential for real-time URLLC in next-generation UAV communication networks.
基金supported part by the National Natural Science Foundation of China(61601252,61801254)Public Technology Projects of Zhejiang Province(LG-G18F020007)+1 种基金Zhejiang Provincial Natural Science Foundation of China(LY20F020008,LY18F020011,LY20F010004)K.C.Wong Magna Fund in Ningbo University。
文摘With the emerging diverse applications in data centers,the demands on quality of service in data centers also become diverse,such as high throughput of elephant flows and low latency of deadline-sensitive flows.However,traditional TCPs are ill-suited to such situations and always result in the inefficiency(e.g.missing the flow deadline,inevitable throughput collapse)of data transfers.This further degrades the user-perceived quality of service(QoS)in data centers.To reduce the flow completion time of mice and deadline-sensitive flows along with promoting the throughput of elephant flows,an efficient and deadline-aware priority-driven congestion control(PCC)protocol,which grants mice and deadline-sensitive flows the highest priority,is proposed in this paper.Specifically,PCC computes the priority of different flows according to the size of transmitted data,the remaining data volume,and the flows’deadline.Then PCC adjusts the congestion window according to the flow priority and the degree of network congestion.Furthermore,switches in data centers control the input/output of packets based on the flow priority and the queue length.Different from existing TCPs,to speed up the data transfers of mice and deadline-sensitive flows,PCC provides an effective method to compute and encode the flow priority explicitly.According to the flow priority,switches can manage packets efficiently and ensure the data transfers of high priority flows through a weighted priority scheduling with minor modification.The experimental results prove that PCC can improve the data transfer performance of mice and deadline-sensitive flows while guaranting the throughput of elephant flows.
基金supported in part by the National Natural Science Foundation of China(Nos.61601182)in part by the Fundamental Research Funds for the Central Universities under Grant 2023MS113。
文摘In response to the requirements for large-scale device access and ultra-reliable and low-latency communication in the power internet of things,unmanned aerial vehicle-assisted multi-access edge computing can be used to realize flexible access to power services and update large amounts of information in a timely manner.By considering factors such as machine communication traffic,MAC competition access,and information freshness,this paper develops a cross-layer computing framework in which the peak Age of Information(Ao I)provides a statistical delay boundary in the finite blocklength regime.We also propose a deep machine learning-based multi-access edge computing offloading algorithm.First,a traffic arrival model is established in which the time interval follows the Beta distribution,and then a business service model is proposed based on the carrier sense multiple access with collision avoidance algorithm.The peak Ao I boundary performance of multiple access is evaluated according to stochastic network calculus theory.Finally,an unmanned aerial vehicle-assisted multilevel offloading model with cache is designed,in which the peak Ao I violation probability and energy consumption provide the optimization goals.The optimal offloading strategy is obtained using deep reinforcement learning.Compared with baseline schemes based on non-cooperative game theory with stochastic learning automata and random edge unloading,the proposed algorithm improves the overall performance by approximately 3.52%and 20.73%,respectively,and provides superior deterministic offloading performance by using the peak Ao I boundary.
基金supported in part by the Key Research and Development Program of Shaanxi(2024GX-YBXM-019)in part by Open Fund of Anhui Province Key Laboratory of Cyberspace Security Situation Awareness and Evaluation(CSSAE-2023-007)in part by the UKRI EPSRC(EP/X038971/1).
文摘Enhanced mobile broadband(eMBB)and ultra-reliable low-latency communication(URLLC)are two critical services in 5G mobile networks.While there has been extensive research on their coexistence,few studies have considered the impact of bursty URLLC on their coexistence performance.In this paper,we propose a method to allocate computing and radio resources for coexisting eMBB and bursty URLLC services by preempting both computing queues in the base station(BS)and time-frequency resources at the air interface.Specifically,we first divide the computing resources at the BS into a shared part for both URLLC and eMBB users and an exclusive part only for eMBB users,and propose a queuing mechanism with preemptive-resume priority for accessing the shared computing resources.Furthermore,we propose a preemptive puncturing method and a threshold-based queuing mechanism in the air interface to enable the multiplexing of eMBB and URLLC on shared time-frequency resources.We analytically derive the average queuing delay,average computation delay,and average transmission delay of eMBB and URLLC packets.Based on this analysis,we formulate a mixed-integer nonlinear programming problem to minimize the average delay of URLLC packets while satisfying the average delay and throughput requirements of eMBB by jointly optimizing the eMBB subcarrier allocation,the URLLC subcarrier scheduling and the computing resource allocation.We decompose this problem into three subproblems and solve them alternately using a block coordinate descent algorithm.Numerical results show that our proposed method reduces the outage probability and average delay of URLLC compared to the existing works.
基金Project supported by the Liaoning Revitalization Talents Program(Nos.XLYC2203148 and XLYC2403062)the National Natural Science Foundation of China(Nos.92267108 and 62173322)。
文摘The coexistence of ultra-reliable low-latency communication(URLLC)and enhanced mobile broadband(eMBB)services in 5G-based industrial wireless networks(IWNs)poses significant resource slicing challenges due to their inherent performance requirement conflicts.To address this challenge,this paper proposes a puncturing method that uses a model-aided deep reinforcement learning(DRL)algorithm for URLLC over eMBB services in uplink 5G networks.First,a puncturing-based optimization problem is formulated to maximize the eMBB accumulated rate under strict URLLC latency and reliability constraints.Next,we design a random repetition coding-based contention(RRCC)scheme for sporadic URLLC traffic and derive its analytical reliability model.To jointly optimize the scheduling parameters of URLLC and eMBB,a DRL solution based on the reliability model is developed,which is capable of dynamically adapting to changing environments.The accelerated convergence of the model-aided DRL algorithm is demonstrated using simulations,and the superiority in resource efficiency of the proposed method over existing approaches is validated.
文摘Wirreless sensor networks are being widely researched and are expected to be used in several scenarios. On the leading edge of treads, on-demand, high-reliability, and low-latency routing protocol is desirable for indoor environment applications. This article proposes a routing scheme called robust multi-path routing that establishes and uses multiple node-disjoint routes. Providing multiple routes helps to reduce the route recovery process and control the message overhead. The performance comparison of this protocol with dynamic source routing （DSR） by OPNET simulations shows that this protocol is able to achieve a remarkable improvement in the packet delivery ratio and average end-to-end delay.
文摘In data center, applications of big data analytics pose a big challenge to massive storage systems. It is signif- icant to achieve high availability, high performance and high scalability for PB-scale or EB-scale storage systems. Meta- data server （MDS） cluster architecture is one of the most effective solutions to meet the requirements of applications in data center. Workload migration can achieve load balance and energy saving of duster systems. In this paper, a hybrid workload migration mechanism of MDS cluster is proposed and named as HWM. In HWM, workload of MDS is classi- fied into two categories： metadata service and state service, and they can be migrated rapidly from a source MDS to a target MDS in different ways. Firstly, in metadata service migration, all the dirty metadata of one sub file system is flushed to a shared storage pool by the source MDS, and then is loaded by the target MDS. Secondly, in state service mi- gration, all the states of that sub file system are migrated from source MDS to target MDS through network at file granular- ity, and then all of the related structures of these states are reconstructed in target MDS. Thirdly, in the process of work- load migration, instead of blocking client requests, the source MDS can decide which MDS will respond to each request according to the operation type and the migration stage. The proposed mechanism is implemented in the Blue Whale MDS cluster. The performance measurements show that the HWM mechanism is efficient to migrate the workload of a MDS cluster system and provides low-latency access to metadata and states.
