Revisiting-Aware In-Orbit Edge Computing for Earth Observation Zehua Sun† , Tao Ni‡ , Kaiyan Cui∥ , Weitao Xu§∗ , Jingxian Wang†
(a) Orbital flight within a revisiting cycle (T)
Revisiting Point
t0
t0+1/8T
t0+2/8T
t0+3/8T
t0+4/8T
t0+5/8T
t0+6/8T
t0+7/8T
t0+T
(b) Imagery captured within a revisiting cycle
Revisiting image
(c) Stride: Revisiting-aware in-orbit edge computing Satellite side
t0+kT
Ground side
T0
T0
···
Abstract—Typically, Earth observation satellites follow a rule of revisiting cycle to periodically pass over the same area of the Earth at regular intervals, which is jointly determined by their orbital properties (e.g., eccentricity, inclination) and instrument characteristics (e.g., off-nadir pointing and swath capabilities). However, we have observed delays in perceived revisiting cycles where limited satellite downlink bandwidth allows only partial images to be delivered, pushing back the timeliness of the full set of data, which we term as revisiting cycle delay. In this paper, we present a revisiting-aware in-orbit edge computing framework for Earth observation termed Stride. Stride leverages the unique orbital revisiting properties to afford historical reference revisiting images onboard, and exploits the inherent temporal redundancy in the revisiting imagery to transmit only the Regions of Interest (RoIs). Specifically, Stride comprises a mono- and multi-temporal cloud indicator to alleviate cloud contamination, a coarse-to-fine reference selector for orbit deviation correction, and an ensemble-local change detector to address inter-band complexities and pixel-level perturbations. Experiments on a Flat-Sat testbed and a constellation simulator demonstrate Stride improves the Revisiting Imagery Delivery (RID) score by up to 4.55×, decreases the connectivity latency by 5.02×, and enlarges the mapping coverage by 2.56×, yielding state-of-the-art performance.
···
arXiv:2607.25813v1 [cs.NI] 28 Jul 2026
† National University of Singapore, Singapore ‡ King Abdullah University of Science and Technology, Saudi Arabia ∥ Nanjing University of Posts and Telecommunications, China § City University of Hong Kong, China
Revisiting Point
Imagery captured
Tn
Tn Revisiting imagery
Region of Interests (ROIs)
Imagery reconstructed
Fig. 1: (a) The revisiting cycle is a primary property of Earth observation satellites, whereby satellites repeatedly pass over the same regions of the Earth. (b) As a satellite revisits a region, the new image often shares substantial temporal redundancy with prior observations. (c) Stride exploits the inherent temporal redundancy among revisiting imagery to transmit only the RoIs, which significantly decreases connectivity latency (5.02×) and enlarges mapping coverage (2.56×).
I. I NTRODUCTION The deployment of spacecraft in Earth’s orbits has experienced a remarkable surge in recent years, resulting in an approximately 10× increase in the number of satellites currently in orbit [1], with well-known types as remote sensing (e.g., Landsat, Sentinel), communication (e.g., Starlink, OneWeb), and navigation (e.g., GPS, GLONASS). In remote sensing, Earth observation satellites are now equipped with state-ofthe-art sensor arrays and sophisticated computer systems, empowering them to support a diverse range of applications (e.g., ground mapping, climate monitoring, and precision agriculture [2], [3]), and possess in-orbit computing capabilities for data processing in space [4], [5]. Typically, Earth observation satellites follow a periodic path, repeatedly passing over specific regions to capture and transmit imagery to ground stations—a process termed revisiting cycle. Satellite revisiting cycles are jointly determined by orbital properties (e.g., eccentricity, inclination [6]) that dictate when the ground track recurs [7], and instrument characteristics (e.g., off-nadir pointing and swath capabilities) that can significantly outpace the orbital cycle by expanding the observable * Weitao Xu is the corresponding author.
area per overpass. For instance, NASA’s Terra and Aqua satellites, equipped with Moderate Resolution Imaging Spectroradiometer (MODIS), achieve 1–2 day global revisit [8]. However, we have observed a revisiting cycle delay, where limited satellite downlink bandwidth allows only partial images to be delivered, pushing back the timeliness of the full set of data. Specifically, despite the proliferation of observation satellites, it has been proved that the downlink capacity remains woefully inadequate to support massive satellite imagery data, rendering only 2% of the captured data being received, due to high orbital dynamics and sparse ground station distributions [4], [5]. The consequence is that, while a satellite can transmit the revisiting imagery from some specific nadir points to ground stations, it cannot transmit the full set of imagery of all nadir points along its track within the same revisiting cycle. This limitation results in a significant perceived revisiting cycle delay for some regions, potentially extending the revisiting cycle beyond the intended daily revisiting (e.g., from 1 to 1.5 days), which causes significant connectivity delays and limited mapping coverage in satellite observations. The impact is most pronounced in time-critical scenarios, such as wildfire outbreaks, sudden flooding, and impending hurricanes, where
TABLE I: Comparison with the existing works.
Multi-level
Mono-level
Method
Data completeness ❍ ❍ ❍
Data reliability ● ● ●
Uplink independence ● ● ❍
Orbital properties ❍ ❍ ❍
To resolve these challenges, we further propose the monoand multi-temporal cloud indicator, coarse-to-fine reference selector, and ensemble-local change detector. As illustrated in Figure 1, for each captured image, Stride first utilizes the cloud indicator to tackle cloud cover contamination, then employs the reference selector to frame the reference revisiting image, with which Stride can identify RoIs of the revisiting imagery based on the change detector. To design and implement Stride, we have overcome the following three main challenges: Challenge 1: Cloud cover contamination. Earth observation satellite imagery frequently suffers from cloud cover contamination [15], which severely degrades imagery quality. Although cloud detection methods are well studied, including spectral feature methods [16], [17] and deep learning-based ones [18], [19], these methods typically remain confined to independent analysis of single or multi-temporal imagery over fixed regions. However, they overlook the richer perspective of satellite orbit dynamics, that is, they rarely adopt the satellite’s own orbital perspective, relying instead on groundbased viewpoints. In contrast, both inherent spatial geographical continuity along the orbit, and their associated temporal continuity can be jointly exploited (see § IV-B). To this end, we propose a novel mono- and multi-temporal cloud indicator that explicitly leverages the inherent spatial-temporal geographical continuity of captured imagery along the satellite orbit. Stride initially processes the mono-temporal candidate image by exploiting the distinctive color and brightness characteristics of cloud cover in specific spectral bands for cloud detection. To further mitigate interference from other highly reflective objects (e.g., snowy mountains, ice sheets), Stride leverages two complementary principles: (1) Spatial continuity: snow and ice exhibit widespread, uniform aggregation (e.g., polar regions), unlike scattered clouds—Stride maintains a cloud cache tracking n recent labels for each location; (2) Temporal continuity: when the cache fills with consecutive “cloudy” labels, Stride triggers multi-temporal verification by comparing the candidate image against a temporally-distant reference—persistent high similarity indicates stationary terrain (false positive), while low similarity confirms transient cloud cover (true positive). Challenge 2: Orbit deviation. Deviations in satellite orbits inevitably occur (as elaborated in Figure 4), caused by various factors such as orbital decay, Earth’s gravitational perturbations, and solar radiation pressure [20]. The common solutions are orbit control (e.g., station-keeping using low-thrust maneuvers) and attitude control (e.g., hardware compensation and calibration) systems. However, long-duration invariance remains unattainable due to the intermittent controls and the saturation of momentum wheels (i.e., temporally limiting control) [21]. Thus, we propose a coarse-to-fine reference selector to achieve strict satellite imagery alignment at the software level. Stride first conducts a coarse-level query to identify potential reference image sets based on embedded geographical coordinate labels. However, fine alignment poses further challenges, primarily stemming from nonlinear changes
Practical deployability ● ● ●
Task-specific filtering
OEC [4] Kodan [5] Serval [9]
Conventional compression
JPEG-2000 [10] CCSDS [11]
● ●
● ●
● ●
❍ ❍
● ●
Deep learning compression
RTCS [12] DeepSpace [13]
● ●
❍ ❍
● ●
❍ ❍
❍ ❍
Reference UL & compression DL
Earth+ [14]
❍
●
❍
❍
❍
Orbital revisiting properties
Stride (ours)
●
●
●
●
●
(●–satisfactory, ❍–weak).
partial image delivery can severely compromise emergency response efforts. As illustrated in Table I, despite extensive efforts to alleviate data transmission bottlenecks, the existing solutions remain limited in data reliability, practical deployment, etc. Specifically, the existing solutions typically operate at the mono-imagery level (i.e., task-specific filtering [4], [5], and mono-imagery compression [10], [11], [12], [13]) and the multi-imagery level (i.e., reference uplinking-based compression [14]). (1) Task-specific filtering. The emerging satellite orbital edge computing systems (e.g., OEC [4], Kodan [5], and Serval [9]) perform task-specific and high-value data filtering or queries, but sacrifice data completeness by preserving only selected information. (2) Conventional compression. The conventional mono-imagery compression methods (e.g., JPEG-2000 [10], CCSDS 121.0-B-3 [11]) predominantly exploit spatial or spectral redundancy within individual images, but fail to exploit the substantial temporal redundancy inherent in satellite imagery, resulting in limited compression performance. (3) Deep learning compression. The recent deep learning-based ones (e.g., RTCS [12], DeepSpace [13]) typically train neural networks to learn a compact latent representation of imagery for compression, which is then deployed on satellites for encoding, and on the ground for reconstruction. However, they exhibit poor generalization, often failing to handle unseen object categories or novel scenes, rendering unreliable data for operators. (4) Reference uplink & compression downlink. A recent advancement is Earth+ [14], where the ground stations pick and then uplink the reference imagery to satellites, while the satellites use the references to compress and then downlink the compressed blocks in the images to the ground. Although effective in exploiting multitemporal redundancy, it requires frequent reference selection and uplinks for onboard comparison, rendering it impractical for satellite systems with highly asymmetric uplink and downlink capacities. As the first attempt from the perspective of celestial mechanics, we present a revisiting-aware in-orbit edge computing framework for Earth observation termed Stride. The key rationale of Stride is to leverage the unique orbital revisiting properties of the satellite to afford historical reference revisiting images, and exploit the inherent temporal redundancy in the revisiting imagery to transmit only the Regions of Interest (RoIs). However, to design and implement Stride, there are three main challenges: cloud cover contamination, orbit deviation, and inter-band complexities & pixel perturbations.
2
97.65
AOP RAAN Inclination
Platform
AOP/RAAN (°)
300
Payload
Satellites
MEO
250 200
97.55 97.50
150
Backhaul
Ground stations
Fig. 2: Satellite systems.
0.0026
50
97.40
545
(a) RAAN, AOP, and inclination.
GEO
0.0030 0.0028
550
0 2022-072022-102023-012023-042023-072023-102024-012024-042024-072024-10
0.0032
555
97.45
LEO
SMA Eccentricity
560
100
Time
Cloud servers
565
97.60
Eccentricity
350
Inclination (°) SMA (km)
PO / SSO
0.0024
2022-072022-102023-012023-042023-072023-102024-012024-042024-072024-10
Time
(b) SMA and eccentricity.
Fig. 4: Illustration of the orbital elements variation over time, indicating the presence of orbit deviations.
Fig. 3: Space orbits.
To the best of our knowledge, Stride is the first framework from the perspective of in-orbit revisiting properties in celestial mechanics. • Stride integrates a mono- and multi-temporal cloud indicator, a coarse-to-fine reference selector, and an ensemblelocal change detector for various challenges.
(e.g., time-induced alterations, scale distortions) in imagery content over time. The existing methods either fall short of efficient computation or struggle with matching distortions (see § IV-C2). To address these problems, we propose a localglobal consensus method. Specifically, we first derive a local geometric prior by extracting the local feature points from the image pairs for keypoint matching. Subsequently, we establish the global consensus by leveraging the relative positions of these matching keypoint groups in their respective images for voting to achieve fine alignment. Challenge 3: Inter-band complexities & pixel perturbations. With the reference revisiting images, change detection is performed to identify RoIs. Although the existing pixellevel change detection methods [22] can capture temporal variations, they face two key challenges: (1) the multi-channel characteristics of satellite imagery introduce complex interband correlations, and (2) pixel perturbations can still arise due to optical hardware imperfections and environmental variations (e.g., illumination and atmospheric disturbances), which unnecessarily inflate the range of RoIs. To address these challenges, we propose an ensemble-local change detector. Specifically, Stride first performs pixel-level ensemble differencing within a single channel to identify RoIs with temporal discrepancies, capitalizing on the “one-to-many” notion that pixel-point variations across layers are correlated. We subsequently design local compensation techniques to refine the RoIs by distinguishing them from Regions of Noise (RoNs) and sparse representation. We implement and evaluate Stride on a real-world Flat-Sat testbed and a constellation simulator. The Flat-Sat testbed integrates key flight subsystems in real satellites, where NVIDIA Jetson TX2 serves as the on-board computing (OBC) platform. The constellation simulator models orbital dynamics, data acquisition, and satellite communication behaviors. We evaluate Stride across five popular applications with three wellknown observation satellite Two-Line Element (TLE) orbit descriptors. Experimental results reveal that:
•
II. P RELIMINARY Earth observation satellite system. As shown in Figure 2, an Earth observation satellite system comprises satellites, ground stations, and the cloud. Satellites capture imagery and transmit it to ground stations via downlink, which then relay data to cloud platforms (e.g., Google Earth Engine, ArcGIS Online) for storage and application services. Uplink/downlink. Earth observation satellite transmission has an extremely asymmetric uplink and downlink. Specifically, the link frequency bands are primarily concentrated in the 7/8 GHz X band [5]. The narrowband uplink typically reaches tens to hundreds of kbps for telecommand, while the wideband downlink can achieve hundreds of Mbps to tens of Gbps [23] for payload data transmission. Satellite orbit. As shown in Figure 3, satellite orbits are categorized by altitude into three types: Low Earth Orbit (LEO), Medium Earth Orbit (MEO), and Geostationary Orbit (GEO). Additionally, two special orbits in LEO are critical for Earth observation: Polar Orbits (POs) and Sun-Synchronous Orbits (SSOs) [24]. POs pass over the poles to provide global coverage. SSOs are near-polar orbits, enabling satellites to revisit the same location at the same local solar time. Revisiting property. The revisiting property of satellites is jointly determined by their orbital properties and instrument characteristics [6]. As illustrated in Figure 4, the orbital properties determine the orbital repeat cycle, dictating when the satellite’s ground track recurs. Specifically, orbital elements consist of six parameters, including Semi-Major Axis (SMA) a, eccentricity e, inclination i, Argument of Perigee (AOP) ω, Right Ascension of the Ascending Node (RAAN) Ω, and mean anomaly M . These elements uniquely define a satellite’s Keplerian orbit, where a, e, and i determine its size, shape, and orientation, while ω, Ω, and M specify its position along the orbit. It is noted that the orbital elements are not stable and exhibit long-term fluctuations, leading to a progressive drift in the satellite’s orbit. On the basis, instrument characteristics, such as swath width and off-nadir pointing capability, expand the observable area per pass, enabling coverage from adjacent orbits and thus reducing revisit time beyond orbital constraints.
Stride improves the Revisiting Imagery Delivery (RID) score (the metric to measure the revisiting capability as illustrated in § III) by up to 4.55× while maintaining the imagery reconstruction quality. • Stride decreases the connectivity latency by 5.02× and enlarges the mapping coverage by 2.56×. • Stride yields state-of-the-art performance. •
Contributions. Stride makes the following contributions:
3
1.0
RID score
0.8 0.6
/ / / /
where cloud pixels exhibit high values in the channels of Red, Green, and Blue. We classify the pixel as part of a cloud if the values of all three channels exceed this threshold ratio τmono . The cloud masking process can be represented as:
Raw/Stride (GS#=1) Raw/Stride (GS#=4) Raw/Stride (GS#=7) Raw/Stride (GS#=10)
0.4
(
0.2 0.0 0
M(h,w) = 5
10
15
20
Time (hours) (a) RID score over ground stations.
1, 0,
if I(h,w,0:2) > max(I) × τmono otherwise.
(1)
By comparing pixel values against the threshold, we can quantify the cloud coverage score by generating a binary cloud mask, whichP indicates the spatial extent of cloud formations: Scloud ← M/|M |. The image with a cloud coverage score exceeding 0.25, is classified as “cloudy” and transmitted directly to ground stations. 2) Multi-Temporal Cloud Verification: Cloud cache. As mentioned previously, other highly reflective surfaces, particularly in snowy landscapes and ice sheets, may influence the cloud coverage score. To address this issue, we propose a multi-temporal cloud verification method as a supplement. We leverage the distinctive spatial continuity that expansive white snow and ice terrain typically exhibits extensive aggregation (e.g., in the Arctic and Antarctica regions), in contrast to the scattered nature of cloud cover. Subsequently, we devise a cache to store each cloud label of the captured revisiting image over time. Essentially, if the cache consistently records multiple “cloudy” labels, it signifies the potential presence of a prominent snowy or icy landscape rather than the cloud cover. Trigger & verification. The cloud cache records the cloud labels of the n most recent revisiting images. When the cloud cache is filled with “cloudy” labels, we perform multi-temporal cloud verification by considering the temporal continuity. Specifically, for such a candidate image, rather than direct transmission, it is processed by the reference selector (§ IV-C) and then the change detector (§ IV-D). The selected reference is the image of the same revisiting area but captured at different times with the candidate image. Thus, the similarity between these two multi-temporal images can be utilized to validate the label of “cloudy”, leveraging the dynamic nature of cloud formations. We indirectly represent this similarity τmulti through the differencing ratio in the change detector. In cases where the candidate image labeled as “cloudy” exhibits a relatively high similarity score, it suggests that other highly reflective objects are misidentified as clouds. Conversely, for an actual “cloudy” image, due to the drowning of useful features in the image, the reference will be selected incorrectly, or the similarity will be relatively low. This multi-temporal cloud verification aids in distinguishing between actual cloud cover and other objects present in the image, which ensures a more reliable cloud assessment in satellite imagery analysis.
(b) Downlink distribution.
Fig. 5: Quantification of revisiting cycle delay. III. R EVISITING C YCLE D ELAY P HENOMENON We quantify the scale of the revisiting cycle delay phenomenon. Due to the varying revisiting cycles across different satellites, we define a unified metric, the RID score, to enable fair comparison. Specifically, the RID score quantifies revisiting capability over a fixed time window (e.g., 24 hours) as the ratio of successfully delivered data to the total collected data along the satellite’s ground tracks. Figure 5a illustrates the RID score over the number of ground stations. It is evident that the volume of data transmitted by the satellite is significantly lower than the total volume of data collected. With ten ground stations, the RID score is only 0.2982 (in contrast to 0.8776 for Stride), which means that the revisiting cycles of 29.82% of the regions are not delayed. This phenomenon is further exacerbated by the restricted number of ground stations, as depicted in Figure 5b. IV. Stride D ESIGN A. Overview Figure 6 presents an overview of Stride. When the satellite captures a revisiting image, it is processed by Stride to identify RoIs for transmission, which are then reconstructed on the ground station side. Specifically, Stride consists of three main components: the cloud indicator (§IV-B), the reference selector (§ IV-C), and the change detector (§ IV-D). B. Cloud Indicator 1) Mono-Temporal Cloud Detection: When a satellite captures an image, it is typically stored in Tagged Image File Format (TIFF). In matrix representation, each element signifies a pixel value within the image. Thus, its three-dimensional matrix is denoted as I (H×W ×N ) , where H, W , and N represent the height, width, and channel number of the image, respectively. The element I(h,w,n) represents the pixel value at a specific height h, width w, and channel index n. For the captured revisiting image, we perform cloud detection on this mono-temporal candidate image I to determine its cloud coverage score. It is noted that cloud shadows are not considered in this process, because the texture features of objects obscured by the cloud shadow remain largely intact. Clouds typically render high reflectivity in some specific spectral bands [15]. For the most common visible spectrum, clouds generally appear as bright features in the image due to the high reflectivity. Therefore, we adopt a threshold-based cloud detection method to capitalize on this characteristic,
C. Reference Selector 1) Coarse Query: Satellite images are typically embedded with geographical information of latitude and longitude coordinates, which allows for a coarse query to filter potential images by imposing restrictions on these geographic labels. In this context, given a candidate image I with its nadir point coordinates (λI , ϕI ), we identify potential reference image
4
Candidate Image Capture
Cloud Indicator
Reference Selector
Change Detector
Mono-Temp. Cloud Detection
Coarse Query
Ensemble Differencing
Multi-Temp. Cloud Verif.
Fine Alignment
Local Compensation
Cloud Caching
Local Prior Deriving
RoN Identification
Trigger & Verification
Global Consensus Voting
Sparse Representation
Revisiting Image Reconstruction
Fig. 6: Overview of Stride (§ IV-A), consisting of a cloud indicator, a reference selector, and a change detector. groups by constraining its latitude and longitude coordinates within a predefined range (τλ , τϕ ). Among them, adjacent potential reference images are concatenated to form larger reference images for fine alignment. 2) Fine Alignment: Local prior deriving. We utilize Speeded Up Robust Features (SURF) [25] to detect points of interest (referred to as keypoints), by constructing a Hessian matrix [26], which typically identifies image edges or regions with rapid intensity changes. For the given candidate image frame I, we utilize its first channel for consideration of energy efficiency. We then construct its Hessian matrix H after applying Gaussian filters, which is represented as: H(w, σ) =
Lww (w, σ) Lhw (w, σ)
Lwh (w, σ) Lhh (w, σ)
(a) Multi-spectral data.
change detector. To address this issue, we propose a global consensus fine alignment method via origin voting. Leveraging the fact that the revisiting reference image shares the same size as the candidate image, we compute the origin index of the matching region in the reference image (i.e., the topleft corner). For each matched keypoint group k(w, h) and k ′ (w′ , h′ ), we calculate the origin index of the matching region in the reference image based on the relative location of its corresponding keypoint in the candidate image via k ′ − k. We then employ a voting mechanism on these potential origin indices to determine the final matching region.
(2)
where Lww , Lwh , Lhw , and Lhh are Gaussian second-order derivatives at point w and scale σ. SURF approximates the Hessian matrix using box filters D, allowing for efficient computation through integral images: Det (Happrox ) = Dww Dhh − (αDwh )2 ,
(3)
where Dww , Dhh , and Dwh are approximations using box filter, and α is a weighting factor to balance the filter responses. Subsequently, SURF creates a scale space by upscaling the filter size rather than downscaling the image to initially generate keypoints. SURF assigns a dominant orientation to each keypoint based on the Haar wavelet [27] responses within a circular region surrounding the keypoint. After keypoints detected and oriented, SURF constructs a feature descriptor for each keypoint by analyzing the intensity changes in its local neighborhood, specifically using Haar wavelet responses. The local neighborhood is divided into several square sub-regions. For each sub-region, SURF calculates the Haar wavelet responses in both horizontal (dw ) and vertical (dh ) directions. The descriptor for each sub-region consists of the sums of these responses: d=
X
dw ,
X
dh ,
X
|dw |,
X
|dh | .
(c) Optical/SAR data.(d) ROIs (downsampled).
Fig. 7: Illustration of the change detector on various data modalities, including multi-spectral, optical, and Synthetic Aperture Radar (SAR).
,
(b) ROIs.
D. Change Detector 1) Ensemble Differencing: For the candidate image I and the selected reference Iref , we compute the difference between these two image matrices using ∆I = I − Iref . The resulting differencing matrix ∆I highlights the pixel discrepancies between these two images, i.e., the RoIs in the satellite image during the revisiting interval. Ideally, the output should be a sparse differencing matrix, where most elements are zeros. However, even when comparing two highly similar images, the subtraction can yield a differencing matrix with numerous pixel value locations (i.e., dense matrix), potentially leading to a large output size. This discrepancy in size can be attributed to prevalent image noise, arising from optical hardware imperfections and environmental variations. Such subtle pixel perturbations, i.e., RoNs, can significantly reduce the effectiveness of the ensemble differencing process. 2) Local Compensation: RoN identification. Regarding such a dense differencing matrix, we propose a local compensation method. Exploiting the similarity between layers in an image, we capitalize on the “one-to-many” notion that pixel-point variations across layers are akin. This allows us to pinpoint noise points in one single channel to identify RoNs, thereby enhancing computational efficiency. Specifically, focusing on the first channel (typically the Red channel) of the
(4)
Keypoints in the candidate image and the reference image are considered matches if their feature descriptors are sufficiently similar, measured by the Euclidean distance. Global consensus voting. After keypoints matching, we perform fine alignment steps. The SURF matching method relies on the homography of the matched keypoints, which is a perspective transformation between two planes. However, this transformation can introduce distortions or deformations in the matching region, significantly affecting the accuracy of the
5
Jetson TX2 Electronic Box
Flat Test Board
Drive
Electronic Box
Solar Panel
Power Jack
(a) Flat-Sat platform.
(b) Anonymous flight model.
Fig. 9: Flat-Sat deployment. TABLE II: Statistics of application datasets. Dataset
Fig. 8: Illustration of satellite tracks and ground station locations on the Earth map under Mercator projection.
SpaceNet 7 [2] LSCIDMR [29] DynamicEarthNet [3] MTGL40-5 [30] SatSOT [31]
differencing matrix ∆I(w,h,0) , we traverse all pixel values for RoNs detection. Noise introduces subtle perturbations in pixel values, which tend to be near zero but not exactly zero. These perturbations, i.e., RoNs, can be identified by comparing the pixel values against a predefined local compensation threshold τ . We then retain the original values for RoIs, while setting the values in RoNs to zero via ( ∆I(h,w,n) =
∆I(h,w,n) , if ∆I(h,w,0) > max(I(h,w,0) ) × τ 0, otherwise.
( V [k], if c = col index[k] && ri ≤ k < ri+1 0, otherwise,
#AOIs 101 11 75 40 105
Mean GSD (m) 4 3 0.5 -
Temporal Resolution Monthly Daily Daily Yearly Sub-secondly
Time Span 2 years 1 year 2 years 5 years -
Source PlanetScope Himawari-8 PlanetFusion Google Earth Skybox, etc.
(AOI: area of interest, GSD: ground sampling distance).
Satellite constellation simulator. We evaluate Stride using the simulator that models orbital dynamics, data acquisition, and satellite communication behaviors. The simulator leverages real-world TLE orbit descriptors from CelesTrak to describe the orbital characteristics of Earth-orbiting satellites. As illustrated in Figure 8, we consider three representative Earth observation satellites: Landsat-8, Sentinel-2A, and SKYSATA, operating during one revisiting cycle (i.e., 24 hours). Additionally, ten ground stations are evenly distributed across the globe. Flat-Sat testbed. To quantify the system overhead, we evaluate Stride on a Flat-Sat testbed (see Figure 9a), a bench-top satellite prototype integrating key flight subsystems, including the OBC, communication interface module (CIM), attitude determination and control system (ADCS), and electrical power system (EPS). These components are interconnected using flight-equivalent harnessing. The Flat-Sat is derived from an already launched satellite platform; a representative replica is shown in Figure 9b. For anonymity, we omit identifying details of the original flight model. Applications. As illustrated in Table II, we evaluate Stride across five popular applications. Default parameters. For the cloud indicator, Stride utilizes 3 bands, with thresholds τmono = 0.7 and τmulti = 0.7. For the reference selector, the algorithm parameters are τλ = τϕ = 0.01 and ϵ = 1 × 107 . For the change detector, the local compensation parameter τ is set to 0.15. Benchmarks. We compare Stride with four baselines: (1) Raw, a plain Earth observation satellite system without data compression; (2) JPEG-2000 [10], a conventional imagery compression method; (3) CCSDS 121.0-B-3 [11], an imagery compression standard made by the Consultative Committee for Space Data Systems (CCSDS). (4) Earth+ [14], the recent reference uplinking-based compression method. Metrics. We consider the following performance aspects: (1) Transmission load. We utilize the size ratio (of the original images) to measure the transmission load of the satellite downlink transmission. (2) Reconstruction quality. We employ the Structural Similarity Index Measure (SSIM)
(5)
Sparse representation. Following ensemble differencing and local compensation, a sparse matrix of RoIs is obtained, with most values being zero, as illustrated in Figure 7. To optimize its size for transmission, we encode it in Compressed Sparse Row (CSR) [28] format. CSR comprises three components: non-zero element value array V , the column index array Col index, and the row pointer array Row ptr. For a sparse matrix ∆I(H,W,N ) with a total of N Z non-zero elements, we first reshape it into ∆I′(H×W, N ) then apply CSR encoding. With these three arrays, the CSR representation of the sparse matrix ∆I′(H×W, N ) can be represented by ∆I′(i,j) =
#Image/ Frame 1,423 104,390 54,750 27,664
(6)
where i and j represent the indices of the row and column of ∆I′ , respectively. V. M ETHODOLOGY Software. We implement Stride as a Python-based processing pipeline that operates in two stages: (1) satellites capture raw observations and transform them into intermediate representations, and (2) ground stations receive and reconstruct the original data. Satellite and ground compute systems. We deploy our applications on NVIDIA Jetson TX2 as the OBC platform, as the commodity computing devices (e.g., NVIDIA Jetson TX2/AGX Orin mobile GPUs) are compatible with being deployed in LEO satellites [4]. Specifically, Jetson TX2 is built with an 256-core NVIDIA PascalTM -family GPU architecture, featuring 8 GB memory, and operating in 7.5/15 W power modes. The ground station is equipped with a workstation (with Intel Core i7-10700 2.90 GHz CPU, 64 GB RAM, and NVIDIA GeForce RTX 3080 GPU).
6