IEEE Geoscience and Remote Sensing Magazine - June 2023 - 52

Compared with onboard storage and computing capabilities,
the ground processing system can make full use
of massive historical Earth observation satellite data for
learning and has a more complete knowledge base and
model base as learning support. Simultaneously, with the
gradual transfer of satellite authority to end users, users can
directly uplink observation task requests to satellites and
receive data products at different levels in real time. However,
terminal users, particularly highly effective users, pay
great attention to high-level situation intelligence. Taking
moving-target monitoring as an example, it is necessary
to assess the threat, intention, and trend of danger targets
and provide danger warnings and other information. It is
difficult for satellite onboard processing systems to realize
high-level situational awareness; therefore, they must
interact with ground processing systems to obtain ground
knowledge and intelligence support.
Establish a satellite-ground-joint intelligent learning
mechanism, learn massive satellite spatial-temporal remote
sensing big data on the ground, generate a lightweight
learning network model, and uplink it to the satellite in real-time
for software updating and reconstruction to achieve
onboard intelligent processing and continuous online
learning. For emergencies and abnormal events, learning
their recognition characteristics and behavior rules on the
ground based on satellite spatial-temporal remote sensing
big data, forming a knowledge base, and uplinking it to
the satellite for storage and updating in real-time to realize
the autonomous perception of emergencies and abnormal
events in orbit. Ground processing can combine the information
obtained by space-, sea-, and land-based sensors to
carry out collaborative reasoning, overcome the problems of
incomplete and discontinuous satellite sparse observation
data, and uplink the generated situational intelligence to the
satellite, providing support for onboard high-level situation
awareness and collaborative observation task planning.
FUTURE DEVELOPMENT TREND
Networking and intelligence are development directions
for future Earth observation satellite systems. Networking
includes the networking of multiple platforms and sensors
as well as observation, communication, and computing
space resources. Intelligence includes intelligent cooperative
observations and intelligent multisource data fusion.
The future key development directions of multisatellite
onboard intelligent observation and data fusion include
implementation of the following methods:
◗ Building elastic and expandable multiorbit and multisensor
Earth observation satellite constellation systems.
LEO, MEO, and HEO satellites achieve cooperative observation
through satellite formation flying, constellation
group networking, and other technologies. Realtime
dynamic observations of hotspots and time-critical
targets can be realized by deploying a small number of
high- and medium-orbit satellites. Combined with highdensity
LEO satellite constellation groups, the capability
52
of global near real-time, high spatial-temporal resolution,
and coverage can be greatly improved. Integrated
satellites and constellations can realize visible, infrared,
SAR, microwave, spectrum, SIGINT, ELINT, and other
multispectral bands in active and passive modes, and
they can provide multimodality heterogeneous data for
onboard information fusion processing.
◗ Establishing an onboard autonomous planning and
scheduling mechanism integrating satellite resources of
observation, communication, and computing; designing
a satellite virtual resource pool under a unified time-
space framework; establishing a new mode of task-driven
software-defined intelligent cooperative observation; dynamically
allocating cooperative observation tasks and
data processing tasks under a dynamic high-speed reconstruction
environment to different satellites in near realtime;
and improving the utilization efficiency of satellite
resources and the collaboration efficiency between multiple
types of satellites, providing highly timely data for
onboard information fusion processing.
◗ Designing an onboard hybrid heterogeneous intelligent
computing architecture for multisatellite data fusion processing,
combining the performance and characteristics
advantage of FPGA, DSP, CPU, GPU, and other hardware;
designing a reconfigurable, scalable and sustainable onboard
intelligent fusion processing model for multisatellite
data; establishing a integration of satellite-ground
cooperative learning and uplink mechanism; learning
knowledge, regularities and model on the ground based
on massive satellite observation data and updating the
intelligent fusion processing system on the satellite in
real-time; adapting to the requirements of onboard multitask
processing; realizing onboard autonomous awareness
of emergencies; and providing users with near realtime
multidimensional and multilevel information.
◗ Promoting the transfer of satellite task control authority
to end users, such that users can directly control satellites
in orbit, uplink instructions, and acquire satellite
data; substantially shorten the satellite task planning,
data processing, and information transmission chain
from the satellite to end users; and improve the rapid
response capability to hot events.
CONCLUSION
Modern Earth observation satellites are capable of high
spatial-temporal-spectral resolutions, multiple working
modes, high agility, and networking collaboration for Earth
observation. The onboard information fusion of multisatellites
can further improve the capability of large-scale
observation, accurate interpretation, and rapid response
for wide-area surveillance, emergency rescue, and other
application scenarios. In this study, the key technologies
of onboard collaborative observation and information fusion
of multisatellites are analyzed and studied, and the
development and suggestions for onboard information
fusion of multisatellite in the future are proposed and
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE JUNE 2023
IEEE Geoscience and Remote Sensing Magazine - June 2023

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2023
