Intersatellite imaging data transfer

Patent metadata

Field	Value
Publication number	US-12451960-B2
Application number	US-202318306856-A
Country	US
Kind code	B2
Filing date	Apr 25, 2023
Priority date	Apr 25, 2023
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

A computing device including a processor configured to receive satellite status data from satellites included in a satellite constellation. The processor is further configured to determine a link topology of the satellites. Based at least in part on the satellite status data and the link topology, the processor is further configured to identify a first satellite constellation subset including one or more selected satellite pairs. Identifying the one or more selected satellite pairs includes computing respective link utility values associated with a plurality of candidate pairs of satellites included in the satellite constellation based at least in part on the satellite status data and the link topology. The one or more selected satellite pairs are selected based at least in part on the link utility values. The processor is further configured to transmit, to the satellites included in the first satellite constellation subset, instructions to perform intersatellite imaging data transfer.

First claim

The invention claimed is:

1. A computing device comprising:
a processor configured to:
receive satellite status data from each of a plurality of satellites included in a satellite constellation, wherein:
the satellite status data is associated with a time window of a plurality of time windows; and
for each of the plurality of satellites, the satellite status data associated with that satellite includes a label distribution of a plurality of labels computed during the time window;

determine a link topology of the plurality of satellites at the time window;
based at least in part on the satellite status data and the link topology, identify a first satellite constellation subset including one or more selected satellite pairs, wherein identifying the one or more selected satellite pairs includes:
computing respective link utility values associated with a plurality of candidate pairs of satellites included in the satellite constellation based at least in part on the satellite status data and the link topology; and
selecting the one or more selected satellite pairs based at least in part on the link utility values; and

transmit, to the satellites included in the first satellite constellation subset, instructions to perform intersatellite imaging data transfer.

2. The computing device of claim 1 , wherein, for each of the plurality of satellites, the satellite status data associated with that satellite includes:
a model staleness value of a local copy of a machine learning model during the time window; and
a training loss value of the local copy of the machine learning model.

3. The computing device of claim 2 , wherein the processor is further configured to:
receive local model update data from the plurality of satellites; and
update a central copy of the machine learning model based at least in part on the local model update data.

4. The computing device of claim 1 , wherein, for each of the plurality of satellites, the plurality of labels are computed at the local copy of the machine learning model during the time window.

5. The computing device of claim 4 , wherein:
the processor is configured to compute a plurality of utility values including the link utility values associated with the candidate pairs, respective local data utility values associated with the plurality of satellites, and respective transfer utility values associated with the plurality of candidate pairs;
when identifying the one or more selected satellite pairs, the processor is further configured to:
compute the local data utility values and the transfer utility values based at least in part on the satellite status data and the link topology; and
compute the respective link utility value associated with each of the candidate pairs based at least in part on the transfer utility value of the candidate pair and the local data utility value of a candidate recipient satellite included in the candidate pair.

6. The computing device of claim 5 , wherein the processor is configured to compute the local data utility values at least in part by, for each of the plurality of satellites:
computing a first distance between the label distribution and an independent identically distributed (i.i.d.) label distribution;
estimating a first number of images stored at the satellite;
estimating a second number of the images with which training is configured to be performed at the local copy of the machine learning model during one or more subsequent time windows; and
computing the local data utility value based at least in part on the first distance, the first number, the second number, and a batch size used during training of the local copy of the machine learning model.

7. The computing device of claim 5 , wherein the processor is configured to compute the transfer utility values at least in part by, for each of the plurality of candidate pairs:
for the candidate recipient satellite included in the candidate pair, computing a second distance between the label distribution and an independent identically distributed (i.i.d.) label distribution;
computing a third distance between an updated label distribution of the candidate recipient satellite and the i.i.d. label distribution, wherein the updated label distribution is computed based at least in part on:
a label distribution of a candidate sender satellite included in the candidate pair; and
a third number of images transferred from the candidate sender satellite to the candidate recipient satellite; and

computing the transfer utility value based at least in part on the second distance and the third distance.

8. The computing device of claim 1 , wherein:
the one or more selected satellite pairs each include a sender satellite and a recipient satellite;
the instructions to perform the intersatellite imaging data transfer include instructions to transfer a plurality of images from the sender satellite to the recipient satellite.

9. The computing device of claim 8 , wherein the processor is further configured to transmit, to a second satellite constellation subset including satellites not included in the selected pairs, instructions to perform training at the local copies of the machine learning model respectively stored at the satellites included in the second satellite constellation subset.

10. The computing device of claim 1 , wherein the processor is further configured to:
compute respective local computing utility values associated with the plurality of satellites based at least in part on the satellite status data;
identify a model aggregation subset of the satellite constellation based at least in part on the local computing utility values and a time-varying model aggregation threshold; and
transmit, to satellites included in the model aggregation subset, instructions to aggregate their respective local copies of the machine learning model.

11. The computing device of claim 10 , wherein the processor is further configured to:
for each of a plurality of model aggregation candidate pairs:
compute a model aggregation time of the respective local copies of the machine learning model stored at the satellites included in the model aggregation candidate pair; and
determine whether the model aggregation time is shorter in duration than the time window; and

identify the model aggregation subset based at least in part on the determinations of whether the model aggregation time is shorter in duration than the time window.

12. The computing device of claim 11 , wherein the processor is further configured to compute the model aggregation time least in part by performing a shortest-path tree search over the link topology.

13. The computing device of claim 1 , wherein the satellite status data associated with the satellite further includes:
an amount of memory available at the satellite during the time window; and/or
an amount of energy available at the satellite during the time window.

14. The computing device of claim 1 , wherein the satellite status data associated with the satellite further includes:
a satellite-ground-station link bandwidth; and
an intersatellite link bandwidth.

15. A method for use with a computing device, the method comprising:
receiving satellite status data from each of a plurality of satellites included in a satellite constellation, wherein:
the satellite status data is associated with a time window of a plurality of time windows; and
for each of the plurality of satellites, the satellite status data associated with that satellite includes a label distribution of a plurality of labels computed during the time window;

determining a link

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Hsieh Tsu-Wang
• Arzani Behnaz
• Olsen Peder Andreas
• Chandra Ranveer
• Li Chenning

Classifications

• H04B7/18586

  Arrangements for data transporting, e.g. for an end to end data transport or check · CPC title

• H04B7/18584

  Arrangements for data networking, i.e. for data packet routing, for congestion control (data switching networks H04L12/00) · CPC title

• H04B7/18521 Primary

  Systems of inter linked satellites, i.e. inter satellite service (for optical links between satellites H04B10/118) · CPC title