Due to the restricted satellite payloads in LEO mega-constellation networks(LMCNs),remote sensing image analysis,online learning and other big data services desirably need onboard distributed processing(OBDP).In exist...Due to the restricted satellite payloads in LEO mega-constellation networks(LMCNs),remote sensing image analysis,online learning and other big data services desirably need onboard distributed processing(OBDP).In existing technologies,the efficiency of big data applications(BDAs)in distributed systems hinges on the stable-state and low-latency links between worker nodes.However,LMCNs with high-dynamic nodes and long-distance links can not provide the above conditions,which makes the performance of OBDP hard to be intuitively measured.To bridge this gap,a multidimensional simulation platform is indispensable that can simulate the network environment of LMCNs and put BDAs in it for performance testing.Using STK's APIs and parallel computing framework,we achieve real-time simulation for thousands of satellite nodes,which are mapped as application nodes through software defined network(SDN)and container technologies.We elaborate the architecture and mechanism of the simulation platform,and take the Starlink and Hadoop as realistic examples for simulations.The results indicate that LMCNs have dynamic end-to-end latency which fluctuates periodically with the constellation movement.Compared to ground data center networks(GDCNs),LMCNs deteriorate the computing and storage job throughput,which can be alleviated by the utilization of erasure codes and data flow scheduling of worker nodes.展开更多
Nowadays,the use of Machine Learning(ML)onboard Earth Observation(EO)satellites has been investigated for a plethora of applications relying on multispectral and hyperspectral imaging.Traditionally,these studies have ...Nowadays,the use of Machine Learning(ML)onboard Earth Observation(EO)satellites has been investigated for a plethora of applications relying on multispectral and hyperspectral imaging.Traditionally,these studies have heavily relied on high-end data products,subjected to extensive pre-processing chains natively designed to be executed on the ground.However,replicating such algorithms onboard EO satellites poses significant challenges due to their computational intensity and need for additional metadata,which are typically unavailable on board.Because of that,current missions exploring onboard ML models implement simplified but still complex processing chains that imitate their on-ground counterparts.Despite these advancements,the potential of ML models to process raw satellite data directly remains largely unexplored.To fill this gap,this paper investigates the feasibility of applying ML models directly to Sentinel-2 raw data to perform thermal hotspot classification.This approach significantly limits the processing steps to simple and lightweight algorithms to achieve real-time processing of data with low power consumption.To this aim,we present an end-to-end(E2E)pipeline to create a binary classification map of Sentinel-2 raw granules,where each point suggests the absence/presence of a thermal anomaly in a square area of 2.5 km.To this aim,lightweight coarse spatial registration is applied to register three different bands,and an EfficientNetlite0 model is used to perform the classification of the various bands.The trained models achieve an average Matthew’s correlation coefficient(MCC)score of 0.854(on 5 seeds)and a maximum MCC of 0.90 on a geographically tripartite dataset of cropped images from the THRawS dataset.The proposed E2E pipeline is capable of processing a Sentinel-2 granule in 1.8 s and within 6.4 W peak power on a combination of Raspberry PI 4 and CogniSat-XE2 board,demonstrating real-time performance.展开更多
基金supported by National Natural Sciences Foundation of China(No.62271165,62027802,62201307)the Guangdong Basic and Applied Basic Research Foundation(No.2023A1515030297)+2 种基金the Shenzhen Science and Technology Program ZDSYS20210623091808025Stable Support Plan Program GXWD20231129102638002the Major Key Project of PCL(No.PCL2024A01)。
文摘Due to the restricted satellite payloads in LEO mega-constellation networks(LMCNs),remote sensing image analysis,online learning and other big data services desirably need onboard distributed processing(OBDP).In existing technologies,the efficiency of big data applications(BDAs)in distributed systems hinges on the stable-state and low-latency links between worker nodes.However,LMCNs with high-dynamic nodes and long-distance links can not provide the above conditions,which makes the performance of OBDP hard to be intuitively measured.To bridge this gap,a multidimensional simulation platform is indispensable that can simulate the network environment of LMCNs and put BDAs in it for performance testing.Using STK's APIs and parallel computing framework,we achieve real-time simulation for thousands of satellite nodes,which are mapped as application nodes through software defined network(SDN)and container technologies.We elaborate the architecture and mechanism of the simulation platform,and take the Starlink and Hadoop as realistic examples for simulations.The results indicate that LMCNs have dynamic end-to-end latency which fluctuates periodically with the constellation movement.Compared to ground data center networks(GDCNs),LMCNs deteriorate the computing and storage job throughput,which can be alleviated by the utilization of erasure codes and data flow scheduling of worker nodes.
文摘Nowadays,the use of Machine Learning(ML)onboard Earth Observation(EO)satellites has been investigated for a plethora of applications relying on multispectral and hyperspectral imaging.Traditionally,these studies have heavily relied on high-end data products,subjected to extensive pre-processing chains natively designed to be executed on the ground.However,replicating such algorithms onboard EO satellites poses significant challenges due to their computational intensity and need for additional metadata,which are typically unavailable on board.Because of that,current missions exploring onboard ML models implement simplified but still complex processing chains that imitate their on-ground counterparts.Despite these advancements,the potential of ML models to process raw satellite data directly remains largely unexplored.To fill this gap,this paper investigates the feasibility of applying ML models directly to Sentinel-2 raw data to perform thermal hotspot classification.This approach significantly limits the processing steps to simple and lightweight algorithms to achieve real-time processing of data with low power consumption.To this aim,we present an end-to-end(E2E)pipeline to create a binary classification map of Sentinel-2 raw granules,where each point suggests the absence/presence of a thermal anomaly in a square area of 2.5 km.To this aim,lightweight coarse spatial registration is applied to register three different bands,and an EfficientNetlite0 model is used to perform the classification of the various bands.The trained models achieve an average Matthew’s correlation coefficient(MCC)score of 0.854(on 5 seeds)and a maximum MCC of 0.90 on a geographically tripartite dataset of cropped images from the THRawS dataset.The proposed E2E pipeline is capable of processing a Sentinel-2 granule in 1.8 s and within 6.4 W peak power on a combination of Raspberry PI 4 and CogniSat-XE2 board,demonstrating real-time performance.
