容错一直是高性能计算领域的热点和难点问题。检查点是解决容错问题的一种常用技术手段,它能够将运行进程的状态转储成文件并恢复。容器具有较强的资源隔离能力,可以为检查点技术提供更理想的运行环境与载体,避免迁移后任务在节点变更...容错一直是高性能计算领域的热点和难点问题。检查点是解决容错问题的一种常用技术手段,它能够将运行进程的状态转储成文件并恢复。容器具有较强的资源隔离能力,可以为检查点技术提供更理想的运行环境与载体,避免迁移后任务在节点变更的情况下由于环境与资源变化而出现异常。因此,容器和检查点相结合能够更好地支撑任务迁移的研究与实现。文中围绕基于CRIU(Checkpoint/Restore In Userspace)的Singularity容器检查点方案的设计和优化展开,根据检查点技术在高性能计算容器应用中的特点,在CRIU安全使用、迁移性能优化、保持网络状态方面给出了有效的解决方案,基于这些方案拓展了Singularity容器检查点功能,并且实现了原型工具Migrator来验证容器迁移性能。期望本工作能为后续实现高性能计算任务迁移提供有效的支撑。展开更多
The Turpan-Hami(Tuha)Basin in Xinjiang Uygur Autonomous Region of China,holds significant strategic importance as a key economic artery of the ancient Silk Road and the Belt and Road Initiative,necessitating a holisti...The Turpan-Hami(Tuha)Basin in Xinjiang Uygur Autonomous Region of China,holds significant strategic importance as a key economic artery of the ancient Silk Road and the Belt and Road Initiative,necessitating a holistic understanding of the spatiotemporal evolution of land use/land cover(LULC)to foster sustainable planning that is tailored to the region's unique resource endowments.However,existing LULC classification methods demonstrate inadequate accuracy,hindering effective regional planning.In this study,we established a two-level LULC classification system(8 primary types and 22 secondary types)for the Tuha Basin.By employing Landsat 5/7/8 imagery at 5-a intervals,we developed the LULC dataset of the Tuha Basin from 1990 to 2020,conducted the accuracy assessment and spatiotemporal evolution analysis,and simulated the future LULC under various scenarios via the Markov-Future Land Use Simulation(Markov-FLUS)model.The results revealed that the average overall accuracy values of our LULC dataset were 0.917 and 0.864 for the primary types and secondary types,respectively.Compared with the seven mainstream LULC products(GlobeLand30,Global 30-meter Land Cover with Fine Classification System(GLC_FCS30),Finer Resolution Observation and Monitoring of Global Land Cover PLUS(FROM_GLC PLUS),ESA Global Land Cover(ESA_LC),Esri Land Cover(ESRI_LC),China Multi-Period Land Use Land Cover Change Remote Sensing Monitoring Dataset(CNLUCC),and China Annual Land Cover Dataset(CLCD))in 2020,our LULC data exhibited dramatically elevated overall accuracy and provided more precise delineations for land features,thereby yielding high-quality data backups for land resource analyses within the basin.In 2020,unused land(78.0%of the study area)and grassland(18.6%)were the dominant LULC types of the basin;although cropland and construction land constituted less than 1.0%of the total area,they played a vital role in arid land development and primarily situated within oases that form the urban cores of the cities of Turpan and Hami.Between 1990 and 2020,cropland and construction land exhibited a rapid expansion,and the total area of water body decreased yet resurging after 2015 due to an increase in areas of reservoir and pond.In future scenario simulations,significant increases in areas of construction land and cropland are anticipated under the business-as-usual scenario,whereas the wetland area will decrease,suggesting the need for ecological attention under this development pathway.In contrast,the economic development scenario underscores the fast-paced expansion of construction land,primarily from the conversion of unused land,highlighting the significant developmental potential of unused land with a slowing increase in cropland.Special attention should thus be directed toward ecological and cropland protection during development.This study provides data supports and policy recommendations for the sustainable development goals of Tuha Basin and other similar arid areas.展开更多
认知无线电(cognitive radio,CR)被认为是解决频谱资源稀缺的核心技术,频谱感知作为CR技术中最为核心的一环,其安全性也就显得尤为重要。分布式协作频谱感知是真实环境中十分常见的一种认知无线电网络(cognitive radio network,CRN)下...认知无线电(cognitive radio,CR)被认为是解决频谱资源稀缺的核心技术,频谱感知作为CR技术中最为核心的一环,其安全性也就显得尤为重要。分布式协作频谱感知是真实环境中十分常见的一种认知无线电网络(cognitive radio network,CRN)下的协作模型。考虑了分布式网络中SU在软判决情况下的变化,考虑完美的信号传输,利用K-means算法对分布式网络中的攻击者进行提前数据划分,集群分类后剔除攻击者,完成进一步的频谱感知,并在Python仿真平台验证,相比传统的信誉度防御算法效果更好。展开更多
文摘容错一直是高性能计算领域的热点和难点问题。检查点是解决容错问题的一种常用技术手段,它能够将运行进程的状态转储成文件并恢复。容器具有较强的资源隔离能力,可以为检查点技术提供更理想的运行环境与载体,避免迁移后任务在节点变更的情况下由于环境与资源变化而出现异常。因此,容器和检查点相结合能够更好地支撑任务迁移的研究与实现。文中围绕基于CRIU(Checkpoint/Restore In Userspace)的Singularity容器检查点方案的设计和优化展开,根据检查点技术在高性能计算容器应用中的特点,在CRIU安全使用、迁移性能优化、保持网络状态方面给出了有效的解决方案,基于这些方案拓展了Singularity容器检查点功能,并且实现了原型工具Migrator来验证容器迁移性能。期望本工作能为后续实现高性能计算任务迁移提供有效的支撑。
基金supported by the Third Xinjiang Scientific Expedition Program (2022xjkk1100)the Tianchi Talent Project
文摘The Turpan-Hami(Tuha)Basin in Xinjiang Uygur Autonomous Region of China,holds significant strategic importance as a key economic artery of the ancient Silk Road and the Belt and Road Initiative,necessitating a holistic understanding of the spatiotemporal evolution of land use/land cover(LULC)to foster sustainable planning that is tailored to the region's unique resource endowments.However,existing LULC classification methods demonstrate inadequate accuracy,hindering effective regional planning.In this study,we established a two-level LULC classification system(8 primary types and 22 secondary types)for the Tuha Basin.By employing Landsat 5/7/8 imagery at 5-a intervals,we developed the LULC dataset of the Tuha Basin from 1990 to 2020,conducted the accuracy assessment and spatiotemporal evolution analysis,and simulated the future LULC under various scenarios via the Markov-Future Land Use Simulation(Markov-FLUS)model.The results revealed that the average overall accuracy values of our LULC dataset were 0.917 and 0.864 for the primary types and secondary types,respectively.Compared with the seven mainstream LULC products(GlobeLand30,Global 30-meter Land Cover with Fine Classification System(GLC_FCS30),Finer Resolution Observation and Monitoring of Global Land Cover PLUS(FROM_GLC PLUS),ESA Global Land Cover(ESA_LC),Esri Land Cover(ESRI_LC),China Multi-Period Land Use Land Cover Change Remote Sensing Monitoring Dataset(CNLUCC),and China Annual Land Cover Dataset(CLCD))in 2020,our LULC data exhibited dramatically elevated overall accuracy and provided more precise delineations for land features,thereby yielding high-quality data backups for land resource analyses within the basin.In 2020,unused land(78.0%of the study area)and grassland(18.6%)were the dominant LULC types of the basin;although cropland and construction land constituted less than 1.0%of the total area,they played a vital role in arid land development and primarily situated within oases that form the urban cores of the cities of Turpan and Hami.Between 1990 and 2020,cropland and construction land exhibited a rapid expansion,and the total area of water body decreased yet resurging after 2015 due to an increase in areas of reservoir and pond.In future scenario simulations,significant increases in areas of construction land and cropland are anticipated under the business-as-usual scenario,whereas the wetland area will decrease,suggesting the need for ecological attention under this development pathway.In contrast,the economic development scenario underscores the fast-paced expansion of construction land,primarily from the conversion of unused land,highlighting the significant developmental potential of unused land with a slowing increase in cropland.Special attention should thus be directed toward ecological and cropland protection during development.This study provides data supports and policy recommendations for the sustainable development goals of Tuha Basin and other similar arid areas.
文摘认知无线电(cognitive radio,CR)被认为是解决频谱资源稀缺的核心技术,频谱感知作为CR技术中最为核心的一环,其安全性也就显得尤为重要。分布式协作频谱感知是真实环境中十分常见的一种认知无线电网络(cognitive radio network,CRN)下的协作模型。考虑了分布式网络中SU在软判决情况下的变化,考虑完美的信号传输,利用K-means算法对分布式网络中的攻击者进行提前数据划分,集群分类后剔除攻击者,完成进一步的频谱感知,并在Python仿真平台验证,相比传统的信誉度防御算法效果更好。
