multidatabase systems and interoperability issues
TRANSCRIPT
소형 위성군 연구 동향 및 연구결과 소개
2017. 12. 01
한상혁 Ph.D/ITPE
Korea Aerospace Research Institute
2017년국내초소형위성개발자워크샵
목차TABLE OF CONTENTS
위성군사례및연구동향
관련연구결과
Page 2
Ⅰ.위성군 사례 및 연구동향
1. 위성군 사례 및 연구동향
❖ 초소형 위성군 서비스 분류
Earth Observation : Visible IR, SAR Traffic Monitoring Service : AIS ADB-S
Weather and Environment Monitoring
Service
Communication : Relay
Satellite
Page 4
❖ 초소형 위성군
Aerial & Maritime : Monitoring aircraft(ADB-S) and Vessels(AIS)($ 7.2M)
AIS Tech : Constellation of 25 Sat / Thermal Imaging and Aviation Tracking
Analytical Space : In-Orbit relay communication and laser communication
Astro Digital : Earth Observation with 6U and 12U($ 16.7M+)
ASTROCAST : 64EA 3U Constellation($ 5.85M)
Blink Astro : Small Ground transmitter and cloud based data analytics service
Blue Field : Methane Tracking with 16U
Capella Space : SAR Constellation($ 12+M)
Douria AeroSpace : Earth Observation with 6U and 16U
Deep Space Industries : Asteroid Prospecting 6U
Dunvegan Space Systems : 24EA BitSat / Communication
EarthCube : Infrared Imaging constellation
Fleet Space : 12U CubeSat
1. 위성군 사례 및 연구동향
Page 5
❖ 초소형 위성군
GeoOptics : Using GPS Occultation for weather Data($ 5.15M)
GHSat : Greenhouse gas and air quality gas emissions monitoring
Harris Corporation : 12EA 6U for Weather wind data
HawkEye 360 : Communication
Helios Wire : 16U Cubesat
HyperCubes : Pollution tracking
Iceye : SAR Constellation
ISIS : AIS Data service
Karten Space : Earth Observation and AIS Data service with 6U
Koolock : Earth Observation with IR
Magnitude Space :
NSLCOmm : Communication network with 60 EA with 1Gbits 3U
OQ Technology
Orbital Micro Systems : weather constellation
Planet : Earth Observation($ 183M)
1. 위성군 사례 및 연구동향
Page 6
❖ 초소형 위성군
Planetary Resource : Earth Observation with midwave infrared($ 50+M)
PlanetiQ : Weather data based on GPS radio occulation($ 5+M)
Promethean Labs : Measuring Greenhouse gas
Rangnarok Industries : Polar Broadband Service
Reaktor Space : Hyperspectral Image Constellation
Sky and Space Global : Communication Service($ 11.5M)
SpacePharma : Microgravity Service
Spire : Weather Monitoring, AIS, ADB-S($ 69.5M)
Transcelestial Technology : Laser Network
Tyvak
406 Aerospace : Weather Data
4Skies
Macro AeroSpace : Space Data Storage
Outernet : Broadcast service
Xoterra AeroSpace : 88EA CubeSats for Earth Observation
1. 위성군 사례 및 연구동향
Page 7
❖ Planet Labs
서비스 내용
➢ Planet Monitoring
➢ Hi Res Monitoring
➢ Open Water Monitoring
➢ Planet Basemaps
➢ Planet Imagery
➢ Planet Archive
➢ Platform
➢ Planet Explorer Beta
➢ Open California
➢ Application Developer Program
➢ Education and Research Program
1. 위성군 사례 및 연구동향
Page 8
❖ Planet Labs
위성군 구성 및 위성 스펙
1. 위성군 사례 및 연구동향
Page 9
❖ Spire
서비스 내용
➢ Maritime
➢ Weather
➢ Aviation
➢ Custom
위성군 구성 및 위성 스펙
➢ Lemur-1, Lemur-2
➢ 125 satellites
➢ Automatic Identification System
➢ weather payloads measure temperature, pressure and precipitation.
1. 위성군 사례 및 연구동향
Page 10
❖ Planetary Resource
서비스 내용
➢ Planetary Resource
➢ Earth Observation with IR
위성군 구성 및 위성 스펙
➢ Arkyd-6
➢ mid-wave infrared sensor
➢ Second-generation System
➢ broadband imager spanning 3 to 5 microns within the infrared region of the electromagnetic spectrum.
▪ This region is sensitive to the presence of water
▪ hydrated minerals
▪ thermal energy
1. 위성군 사례 및 연구동향
Page 11
❖ Astro Digital
서비스 내용➢ 6U and 16U CubeSat➢ fresh images of the planet every
day➢ AGRICULTURE, DISASTER
MANAGEMENT, FOREST MANAGEMENTURBAN DEVELOPMENT, BUSINESS INTELLIGENCE
위성군 구성 및 위성 스펙➢ Landmapper-HD, 20EA
▪ Imaging the world at 2.5 meters resolution, the spacecraft weighs 20kg and is about the size of a small microwave
➢ Landmapper-BC▪ 22 meters resolution, the
spacecraft weighs 10kg and is about the size of a shoebox.
▪ 1TB of imagery.
1. 위성군 사례 및 연구동향
Page 12
❖ Sky and Space Global
서비스 내용➢ Band width satellite
communication providers➢ Machine-to-Machine➢ Real-time tracking for airliners
and shipping companies➢ Complementary service for
cellular networks in Latin Americas, Asian and African regions with poor coverage
➢ Ad-hoc disaster and crisis areas solutions
➢ Premium secure services
위성군 구성 및 위성 스펙➢ Goal : 200 EA➢ 4-way inter-satellites links➢ Onboard orbit control➢ Autonomous optimal network
management➢ Fully independent space segment
1. 위성군 사례 및 연구동향
Page 13
❖ Capella Space
서비스 내용
➢ Maritime Domain Awareness
➢ City Planning and Mapping
➢ Oil and Energy
➢ Patterns of Life
➢ Business Intelligence
➢ Agriculture
위성군 구성 및 위성 스펙
➢ SAR
➢ Goal : 40 EA
➢ 2017년 Q4 발사 예정
1. 위성군 사례 및 연구동향
Page 14
❖ ICEYE
서비스 내용
➢ Maritime Domain Awareness
➢ City Planning and Mapping
➢ Oil and Energy
➢ Patterns of Life
➢ Business Intelligence
➢ Agriculture
위성군 구성 및 위성 스펙
➢ SAR
1. 위성군 사례 및 연구동향
Page 15
Ⅱ.연구결과
과제명 : “중소기업상용화지원사업”
Techniques for Nanosatellite Constellation
Space Segment
• Payload
– High Resolution Camera
– Radar or None Visible Payload
• BUS
– Inter-Satellite Communication
– Relative Orbit control and Determination System
– Mass Data Processing On Board Computer
– High Frequency Transceiver
Ground Segment
• Big Data Processing & Analytic Technique
• Contact Schedule Managing
위성간 데이터 동기화 기술
빅데이터, 딥러닝 기술
2. 연구결과 위성간데이터동기화기술
Page 17
❖ Research Objective
Development of data synchronization module for small satellites constellation
❖ Research SignificanceDevelopment of technique for small satellites constellation mission
Various applications
➢ Ex) Drone constellation / landing, vehicle-to-vehicle communication
2. 연구결과 위성간데이터동기화기술
Page 18
❖ Research Reference
Data synchronization module for small satellite constellation is developed based on patent “APPARATUS AND METHOD FOR DATA SYNCHRONIZATION IN MULTIPLE SATELLITES” (patent number 10-16202330000)
(Han, 2013)
2. 연구결과 위성간데이터동기화기술
Page 19
❖ Data Synchronization Module Overview Data synchronization process is as follows
➢ Determine synchronization target
➢ Determine synchronization data
➢ Form synchronization log
➢ Send and receive synchronization data
2. 연구결과 위성간데이터동기화기술
Page 20
❖ Software Simulation
Scenario : Sync position and velocity data
→ Check whether TX satellite sends its sync data according to its configuration file
→ Check RX satellite receives sync data
RX
Satellite
Onboard
Propagator
Data Sync.
통신
Socket
통신
TX
Satellite
Onboard
Propagator
Data Sync.
Socket
통신
2. 연구결과 위성간데이터동기화기술
Page 21
❖ Software Simulation
Performed hardware simulation with two Raspberry Pis
→ Socket is used for internal communication
→ VNC Viewer is used for control of Raspberry Pis
2. 연구결과 위성간데이터동기화기술
Page 22
❖ Software Verification
TX satellite
RX satellite
2. 연구결과 위성간데이터동기화기술
Page 23
Satellite
Image
GDAL
TranslateSciDB
GDAL
TranslateVisualize
GDAL
read
Satellite
Image
Tiff to Scidb Scidb to Tiff
2001-04-01, NDVI, H12V9 2004-04-01, NDVI, H12V9
Dev. Language : R
2. 연구결과 빅데이터처리기술
Page 24
Research Steps
1. Study Bigdata processing techniques for Satellite Imagery
• Reference [1]
• => it’s multi-dimensional array database technology; SciDB
2. Study image processing techniques based SciDB
• Reference [2-4]
• Image processing technique : GDAL
• Image processing with SciDB : GDAL4SciDB
3. Visualize of MODIS satellite data by SciDB & GDAL
• Reference [2-3]
MODIS; Moderate Resolution Imaging Spectroradiometer, NASA Terra
Swath 2,330 km, every 1-2 day in the world, 1 km spatial resolution
2. 연구결과 빅데이터처리기술
Page 25
Bigdata processing technique for Satellite Imagery
HadoopNoSQL
/MongoDBRDBMS
Data
Manage
Simple File
Distribution
based HDFS
No schemaRelational
model
Query
ProcessMapReduce Key-value SQL
Constraint
Difficult to
process mass
science data
Difficult to
process
mathematic
al analysis
Difficult to
process
multi-
dimensional
data
Limit of Data Processing Technique
For Satellite Data
SciDB
▪ Open source high
performance DB
▪ Support over a number of
petabytes data
▪ Array data model
▪ Support Multi-Dimensional
data
▪ Support Data version
▪ Support R, Python lan.
2. 연구결과 빅데이터처리기술
Page 26
Visualize MODIS by SciDB & GDAL
H08v05-the western United States
Test Data(MODIS) : MOD13A3
EVI : Enhanced Vegetation Index
NDVI; Normalized Difference Vegetation Index
H8V5
2. 연구결과 빅데이터처리기술
Page 27
Visualize MODIS by SciDB & GDAL
2001-04-01, NDVI, H12V9 2004-04-01, NDVI, H12V9
2. 연구결과 빅데이터처리기술
Page 28
2. 연구결과 빅데이터분석기술
∙ Step1. target 지역 선정 : 물동량이 많으며, 다양한 환경의 도시의 최근영상을 획득
∙ Step2. Labeling : 동일 지역의 서로 다른 날짜를 비교하여 선박으로 추정되는부분을 labeling하여 좌상, 우하의 위치 정보(+margin) 추출
∙ Step3. Answer sheet 제작 : 육지 및 바다와 같이 선박으로 분류될 수 없도록오답지 정보를 추출
∙ Step4. Training : 3,4 번 데이터를 바탕으로 훈련 데이터로 입력하여 모델 학습
∙ Step5. Prediction: 선박으로 추정되는 부분을 전체 지역에서 구분하여 표시
『목표 : SAR 영상 내에서 선박 추정』
Page 29
Training set (True) Training set (False)
∙ Develop Second Segmentation Algorithm
Item Content
Segmentation 1 type(Ship in the water)
Data SourceSentinel-1 SAR Image
https://earthdata.nasa.gov
Resolution5 x 20 m Spatial Resolution
(Swath 250 km)
Training Set Annotated in 10,000 x 10,000 image
Training Time About 4 hours
Model U-net (10 layers)
Develop Env.
Lan. : Python
FW : Keras over Tensorflow
HW : 4 x GPU
(NVIDIA GeForce GTX TITAN X)
Ship+15(margin)
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
Page 30
2017.6.3( Shanghai, China )
∙ Result – Original Data
2017.6.27( Shanghai, China )
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
Page 31
배경흰색으로
∙ Result – Predicted Image
2017.6.3( Shanghai, China ) 2017.6.27( Shanghai, China )
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
Page 32
∙ Result – Merge Original & Segmentation Image
2017.6.3( Shanghai, China ) 2017.6.27( Shanghai, China )
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
Page 33
∙ 노이즈감소와데이터부족해결을위해 data augmentation 수행
- Keras에서 지원하는 ImageDataGenerator 함수 사용
- 다양한 방식의 augmentation이 가능하나 최대한 배의 형상 유지 위해최소한의 변형만 부여
[ ImageDataGenerator 환경설정 ]
[ Original Img. (좌) Augmented Img. (우) ]
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
Page 34
∙ 노이즈개선및정확도증가
[노란색 (augmentation 전 ) -> 초록색 (후)]
2. 연구결과 빅데이터분석기술
■ SAR 영상에서배세그멘테이션
[ Panama (6/11) ] [ Hong Kong (6/4) ]
Page 35
For short-term
Enhance Training model with updating training set
• And, check the reliability of our algorithm.
Integrate Deep Learning Segmentation Algorithm in SciDB
• Develop Keras-R interface module
• Import SAR image in SciDB by GDAL interface
• Visualize in R
Develop Object Detection Algorithm and integrate in our system
Distribute SciDB Data in Multi-node
For long-term
Survey target Application for verifying Bigdata Framework
Revise the Bigdata Framework for Nanosatellite Constellation
3. 향후 계획
Page 36
[1] 박재필, “위성군 사례 및 연구동향”, 중소기업 상용화 지원사업과제회의, 2017.10.18
[2] 김민식, “소형 위성군 운용을 위한 데이터 동기화 모듈개발”, KSAS 2017
[3] 최연주, “Detection of ships in SAR image by using deep learning”,KSAS 2017
[4] 한상혁, “Bigdata Framework Design for Nanosatellite Constellation Using Multi-Dimensional Array Database”, COSPAR 2017
4. Reference
Page 37
Contact : [email protected]
Page 38