Distributed machine learning systems train models via iterative updates between parallel workers and the parameter server.To expedite the transmissions,in-network aggregation of updates along with the packet forwardin...Distributed machine learning systems train models via iterative updates between parallel workers and the parameter server.To expedite the transmissions,in-network aggregation of updates along with the packet forwarding at those programmable switches decreases the network traffic over these bottleneck links.However,existing in-network aggregation schemas do not adequately prepare the most suitable switches for various worker distributions and fail to capture the dynamic network status.Based on the status derived from in-band network telemetry,we aim to select the best switches upon the optimization we formulate with the objective of minimum transmission latency.Although the problem is actually a non-linear integer program,by adopting delicate transformations,a substitute with totally unimodular constraints and separable convex objective is then solved to obtain the integral optimum.We implement our in-network aggregation protocol and reconstruct in-band network telemetry protocol upon real devices,i.e.,Barefoot WedgelooBF and Dell servers.Then,we evaluate the performance of our proposed AGG algorithm and the results indicate that the completion of related coflows decreases 40%on average compared with other strategies,improving at least 30%performance,compared with the state-of-the-art.展开更多
In the Ethernet lossless Data Center Networks (DCNs) deployedwith Priority-based Flow Control (PFC), the head-of-line blocking problemis still difficult to prevent due to PFC triggering under burst trafficscenarios ev...In the Ethernet lossless Data Center Networks (DCNs) deployedwith Priority-based Flow Control (PFC), the head-of-line blocking problemis still difficult to prevent due to PFC triggering under burst trafficscenarios even with the existing congestion control solutions. To addressthe head-of-line blocking problem of PFC, we propose a new congestioncontrol mechanism. The key point of Congestion Control Using In-NetworkTelemetry for Lossless Datacenters (ICC) is to use In-Network Telemetry(INT) technology to obtain comprehensive congestion information, which isthen fed back to the sender to adjust the sending rate timely and accurately.It is possible to control congestion in time, converge to the target rate quickly,and maintain a near-zero queue length at the switch when using ICC. Weconducted Network Simulator-3 (NS-3) simulation experiments to test theICC’s performance. When compared to Congestion Control for Large-ScaleRDMA Deployments (DCQCN), TIMELY: RTT-based Congestion Controlfor the Datacenter (TIMELY), and Re-architecting Congestion Managementin Lossless Ethernet (PCN), ICC effectively reduces PFC pause messages andFlow Completion Time (FCT) by 47%, 56%, 34%, and 15.3×, 14.8×, and11.2×, respectively.展开更多
This article introduces a cable-free real-time telemetry seismic acquisition system(hereinafter referred to as the cable-free real-time telemetry system)that utilizes 4G/5G technology.This system facilitates the real-...This article introduces a cable-free real-time telemetry seismic acquisition system(hereinafter referred to as the cable-free real-time telemetry system)that utilizes 4G/5G technology.This system facilitates the real-time acquisition and quality control of seismic data,the real-time monitoring of equipment location and health status,the synchronous transmission of collected data between the cloud and client,and the real-time issuance of operational instructions.It addresses the critical limitation of existing seismic node equipment,which is often restricted to mining and blind storage due to the absence of a wired or wireless communication link between the acquisition node device and the central control unit.This limitation necessitates local data storage and rendering real-time quality control unfeasible.Typically,quality control is conducted post-task completion,requiring the overall retrieval and downloading of data.If data issues are identifi ed,it becomes necessary to eliminate faulty tracks and determine the need for supplementary acquisition,which can lead to delays in the acquisition process.The implementation of real-time monitoring and early warning systems for equipment health status aims to mitigate the risk of poor data quality resulting from equipment anomalies.Furthermore,the real-time synchronous transmission between the cloud and server addresses the bottleneck of slow download speeds associated with the centralized retrieval of data from multiple node devices during blind acquisition and storage.A real-time microseismic data acquisition test and verifi cation were conducted at a fracturing site in an eastern oil and gas fi eld.Analysis of the test data indicates that the overall performance indicators of the system are comparable to those of existing mainstream system equipment,demonstrating stability and reliability.The performance parameters fully satisfy the technical requirements for oilfield fracturing monitoring scenarios,suggesting promising prospects for further promotion and application.展开更多
Underwater acoustic modem technology has attained a level of maturity to support underwater acoustic sensor networks （UASNs） which are generally formed by acoustically connected sensor nodes and a surface station pr...Underwater acoustic modem technology has attained a level of maturity to support underwater acoustic sensor networks （UASNs） which are generally formed by acoustically connected sensor nodes and a surface station providing a link to an on-shore control center. While many applications require long-term monitoring of the deployment area, the batterypowered network nodes limit the lifetime of UASNs. Therefore, designing a UASN that minimizes the power consumption while maximizing lifetime becomes a very difficult task. In this paper, a method is proposed to determine the optimum number of clnstens throngh combining an application-specific protocol architecture and underwater acoustic communication model so as to reduce the energy dissipation of UASNs. Deploying more sensor nodes which work alternately is another way to prolong the lifetime of UASNs, An algorithm is presented for selecting sensor nodes and putting them into operation in each round, ensuring the monitoring to the whole given area. The present results show that the algorithm can help prolong system lifetime remarkably when it is applied to other conventional approaches for sensor networks under the condition that the sensor node density is high.展开更多
针对飞行器飞行试验中外测级间段数据缺失和精度不高的问题,提出了基于长短期记忆(long-short term memory,LSTM)网络的外测级间段数据预测方法。利用遥测视速度数据和外测融合数据建立LSTM网络回归模型,将外测级间段数据作为缺失数据...针对飞行器飞行试验中外测级间段数据缺失和精度不高的问题,提出了基于长短期记忆(long-short term memory,LSTM)网络的外测级间段数据预测方法。利用遥测视速度数据和外测融合数据建立LSTM网络回归模型,将外测级间段数据作为缺失数据进行预测插值,可将制导工具系统误差以及飞行器初始误差,包括遥外测时间对不准误差,一并利用回归网络表示,从而将遥测视速度数据作为网络输入,得到外测级间段的预测数据。试验数据处理结果证明,基于LSTM网络获得的外测级间段预测数据满足精度要求,所提方法具有实际应用价值。展开更多
现阶段SRv6(Segment Routing over IPv6)中,段标签(Segment Identifier,SID)设计为流量工程、安全认证等网络功能提供了可编程性。这些功能的实现依赖于数据平面中流表的精确匹配与执行,但当流表被恶意篡改或错误配置时,容易导致功能一...现阶段SRv6(Segment Routing over IPv6)中,段标签(Segment Identifier,SID)设计为流量工程、安全认证等网络功能提供了可编程性。这些功能的实现依赖于数据平面中流表的精确匹配与执行,但当流表被恶意篡改或错误配置时,容易导致功能一致性问题的出现。而带内网络遥测(In-band Network Telemetry,INT)技术作为SDN场景中经典的具有可编程性的校验工具,可将二者天然结合。为此,提出了基于可编程数据平面的SRv6功能一致性验证机制(Programmable Data Plane Based Consistency Verification Mechanism for SRv6 Functional,SRv6FCV)。SRv6FCV采用数据平面可编程技术为探针包插入认证标识,首先依据监控需求动态地将SID转换为特定的INT元数据结构,然后构造探针报文注入并逐跳收集特定网络功能的流表执行信息,最后对遥测信息进行解析并基于符号执行算法完成功能一致性验证。仿真结果表明,SRv6FCV能够保证满足流表规则以及业务功能执行策略的一致性。相较于之前的研究,SRv6FCV在完成对网络功能一致性校验的基础上,拥有更低的运行开销,同时其校验时长也有显著缩短。展开更多
基金supported by the National Natural Science Foundation of China under Grant No.61832005Jiangsu Collaborative Innovation Center of Novel Software Technology and Industrialization.
文摘Distributed machine learning systems train models via iterative updates between parallel workers and the parameter server.To expedite the transmissions,in-network aggregation of updates along with the packet forwarding at those programmable switches decreases the network traffic over these bottleneck links.However,existing in-network aggregation schemas do not adequately prepare the most suitable switches for various worker distributions and fail to capture the dynamic network status.Based on the status derived from in-band network telemetry,we aim to select the best switches upon the optimization we formulate with the objective of minimum transmission latency.Although the problem is actually a non-linear integer program,by adopting delicate transformations,a substitute with totally unimodular constraints and separable convex objective is then solved to obtain the integral optimum.We implement our in-network aggregation protocol and reconstruct in-band network telemetry protocol upon real devices,i.e.,Barefoot WedgelooBF and Dell servers.Then,we evaluate the performance of our proposed AGG algorithm and the results indicate that the completion of related coflows decreases 40%on average compared with other strategies,improving at least 30%performance,compared with the state-of-the-art.
基金supported by the National Natural Science Foundation of China (No.62102046,62072249,62072056)JinWang,YongjunRen,and Jinbin Hu receive the grant,and the URLs to the sponsors’websites are https://www.nsfc.gov.cn/.This work is also funded by the National Science Foundation of Hunan Province (No.2022JJ30618,2020JJ2029).
文摘In the Ethernet lossless Data Center Networks (DCNs) deployedwith Priority-based Flow Control (PFC), the head-of-line blocking problemis still difficult to prevent due to PFC triggering under burst trafficscenarios even with the existing congestion control solutions. To addressthe head-of-line blocking problem of PFC, we propose a new congestioncontrol mechanism. The key point of Congestion Control Using In-NetworkTelemetry for Lossless Datacenters (ICC) is to use In-Network Telemetry(INT) technology to obtain comprehensive congestion information, which isthen fed back to the sender to adjust the sending rate timely and accurately.It is possible to control congestion in time, converge to the target rate quickly,and maintain a near-zero queue length at the switch when using ICC. Weconducted Network Simulator-3 (NS-3) simulation experiments to test theICC’s performance. When compared to Congestion Control for Large-ScaleRDMA Deployments (DCQCN), TIMELY: RTT-based Congestion Controlfor the Datacenter (TIMELY), and Re-architecting Congestion Managementin Lossless Ethernet (PCN), ICC effectively reduces PFC pause messages andFlow Completion Time (FCT) by 47%, 56%, 34%, and 15.3×, 14.8×, and11.2×, respectively.
基金funded by the National Natural Science Foundation of China (42074127)the Key Program of National Natural Science Foundation of China (41930425)Research on Key Technologies for the Production, Exploration, and Development of Continental Shale Oil (2023ZZ15YJ02)。
文摘This article introduces a cable-free real-time telemetry seismic acquisition system(hereinafter referred to as the cable-free real-time telemetry system)that utilizes 4G/5G technology.This system facilitates the real-time acquisition and quality control of seismic data,the real-time monitoring of equipment location and health status,the synchronous transmission of collected data between the cloud and client,and the real-time issuance of operational instructions.It addresses the critical limitation of existing seismic node equipment,which is often restricted to mining and blind storage due to the absence of a wired or wireless communication link between the acquisition node device and the central control unit.This limitation necessitates local data storage and rendering real-time quality control unfeasible.Typically,quality control is conducted post-task completion,requiring the overall retrieval and downloading of data.If data issues are identifi ed,it becomes necessary to eliminate faulty tracks and determine the need for supplementary acquisition,which can lead to delays in the acquisition process.The implementation of real-time monitoring and early warning systems for equipment health status aims to mitigate the risk of poor data quality resulting from equipment anomalies.Furthermore,the real-time synchronous transmission between the cloud and server addresses the bottleneck of slow download speeds associated with the centralized retrieval of data from multiple node devices during blind acquisition and storage.A real-time microseismic data acquisition test and verifi cation were conducted at a fracturing site in an eastern oil and gas fi eld.Analysis of the test data indicates that the overall performance indicators of the system are comparable to those of existing mainstream system equipment,demonstrating stability and reliability.The performance parameters fully satisfy the technical requirements for oilfield fracturing monitoring scenarios,suggesting promising prospects for further promotion and application.
文摘Underwater acoustic modem technology has attained a level of maturity to support underwater acoustic sensor networks （UASNs） which are generally formed by acoustically connected sensor nodes and a surface station providing a link to an on-shore control center. While many applications require long-term monitoring of the deployment area, the batterypowered network nodes limit the lifetime of UASNs. Therefore, designing a UASN that minimizes the power consumption while maximizing lifetime becomes a very difficult task. In this paper, a method is proposed to determine the optimum number of clnstens throngh combining an application-specific protocol architecture and underwater acoustic communication model so as to reduce the energy dissipation of UASNs. Deploying more sensor nodes which work alternately is another way to prolong the lifetime of UASNs, An algorithm is presented for selecting sensor nodes and putting them into operation in each round, ensuring the monitoring to the whole given area. The present results show that the algorithm can help prolong system lifetime remarkably when it is applied to other conventional approaches for sensor networks under the condition that the sensor node density is high.
文摘针对飞行器飞行试验中外测级间段数据缺失和精度不高的问题,提出了基于长短期记忆(long-short term memory,LSTM)网络的外测级间段数据预测方法。利用遥测视速度数据和外测融合数据建立LSTM网络回归模型,将外测级间段数据作为缺失数据进行预测插值,可将制导工具系统误差以及飞行器初始误差,包括遥外测时间对不准误差,一并利用回归网络表示,从而将遥测视速度数据作为网络输入,得到外测级间段的预测数据。试验数据处理结果证明,基于LSTM网络获得的外测级间段预测数据满足精度要求,所提方法具有实际应用价值。
文摘现阶段SRv6(Segment Routing over IPv6)中,段标签(Segment Identifier,SID)设计为流量工程、安全认证等网络功能提供了可编程性。这些功能的实现依赖于数据平面中流表的精确匹配与执行,但当流表被恶意篡改或错误配置时,容易导致功能一致性问题的出现。而带内网络遥测(In-band Network Telemetry,INT)技术作为SDN场景中经典的具有可编程性的校验工具,可将二者天然结合。为此,提出了基于可编程数据平面的SRv6功能一致性验证机制(Programmable Data Plane Based Consistency Verification Mechanism for SRv6 Functional,SRv6FCV)。SRv6FCV采用数据平面可编程技术为探针包插入认证标识,首先依据监控需求动态地将SID转换为特定的INT元数据结构,然后构造探针报文注入并逐跳收集特定网络功能的流表执行信息,最后对遥测信息进行解析并基于符号执行算法完成功能一致性验证。仿真结果表明,SRv6FCV能够保证满足流表规则以及业务功能执行策略的一致性。相较于之前的研究,SRv6FCV在完成对网络功能一致性校验的基础上,拥有更低的运行开销,同时其校验时长也有显著缩短。
