Preventing service interruptions by predicting outages in a satellite network
This disclosure describes techniques for predicting and accommodating for outages in a satellite network using crowdsourced data. An example method includes receiving outage data indicating first outages experienced by first endpoints in a first geographical region. The first outages, for instance, include interruptions in communication between first satellites and the first endpoints. The example method further includes predicting, based on the outage data, a second outage comprising an interruption in communication between at least one second satellite and a second endpoint in a second geographical region. Further, the example method includes causing the second endpoint to transmit user data over a secondary network in advance of the second outage.
1 . A method performed by a system, the method comprising:
receiving outage data indicating first outages experienced by first endpoints in a first geographical region, the first outages comprising interruptions in communication between first satellites and the first endpoints;
receiving contextual data indicating:
radio frequency (RF) conditions impacting the first geographical region and a second geographical region;
weather conditions in the first geographical region and the second geographical region; and
network conditions associated with a satellite network comprising the first satellites and at least one second satellite;
predicting, based on the outage data and the contextual data, a future time associated with a second outage comprising an interruption in communication between the at least one second satellite and a second endpoint in a second geographical region; and
causing the second endpoint to transmit user data over a secondary network in advance of the future time associated with the second outage.
2 . The method of claim 1 , wherein the outage data is received from at least one proxy connected to the first endpoints.
3 . The method of claim 1 , wherein the outage data comprises locations of the first endpoints during the first outages, times of the first outages, and durations of the first outages.
4 . The method of claim 1 , wherein the first satellites comprise at least one Low Earth Orbit (LEO) satellite.
5 . The method of claim 1 , wherein predicting the second outage comprises:
training a machine learning model using a first portion of the outage data; and
in response to training the machine learning model, predicting the second outage based on a second portion of the outage data and the machine learning model.
6 . The method of claim 1 , wherein the secondary network comprises a terrestrial cellular network.
7 . The method of claim 1 , wherein causing the second endpoint to transmit user data over the secondary network in advance of the second outage comprises:
transmitting, to a proxy connected to the second endpoint, an instruction to transmit the user data to the secondary network.
8 . The method of claim 1 , wherein the contextual data further indicates
meteorological events impacting the first geographical region and the second geographical region; and
at least one inoperative satellite among the first satellites and/or at least one second satellite.
9 . The method of claim 1 , wherein the contextual data further indicates meteorological events impacting the first geographical region and the second geographical region the meteorological events comprising solar flares.
10 . A system, comprising
at least one processor; and
one or more non-transitory media storing instructions that, when executed by the system, cause the system to perform operations comprising:
receiving outage data indicating first outages experienced by first endpoints in a first geographical region, the first outages comprising interruptions in communication between first satellites and the first endpoints;
receiving contextual data indicating radio frequency (RF) conditions impacting the first geographical region and a second geographical region, weather conditions in the first geographical region and the second geographical region, or network conditions associated with a satellite network comprising the first satellites and at least one second satellite;
predicting, based on the outage data and the contextual data, a second outage comprising an interruption in communication between the at least one second satellite and a second endpoint in a second geographical region; and
in response to predicting the second outage and prior to the second outage, transmitting, to a proxy communicatively coupled to the second endpoint, a message indicating the second outage.
11 . The system of claim 10 , wherein the outage data is received from at least one proxy connected to the first endpoints.
12 . The system of claim 10 , wherein the outage data comprises locations of the first endpoints during the first outages, times of the first outages, and durations of the first outages.
13 . The system of claim 10 , wherein the first satellites comprise at least one Low Earth Orbit (LEO) satellite.
14 . The system of claim 10 , wherein predicting the second outage comprises:
training a machine learning model using a first portion of the outage data; and
in response to training the machine learning model, predicting the second outage based on a second portion of the outage data and the machine learning model.
15 . The system of claim 10 , wherein the message indicating the second outage comprises an instruction to transmit user data associated with the second endpoint over a terrestrial network, rather than the satellite network.
16 . The system of claim 15 , wherein the terrestrial network comprises a cellular network.
17 . An outage prediction system, comprising:
at least one processor; and
memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
receiving outage data indicating first outages experienced by endpoints disposed in a grid of geographical regions, the first outages comprising interruptions in communication between a satellite network and the endpoints, the satellite network comprising low earth orbit (LEO) satellites;
receiving contextual data indicating radio frequency (RF) conditions in the geographical regions, weather conditions in the geographical regions, and network conditions associated with the satellite network;
predicting, based on the outage data and the contextual data, a second outage comprising an interruption in communication between the satellite network and a particular endpoint among the endpoints, the particular endpoint being in one of the geographical regions; and
in response to predicting the second outage, transmitting, to a proxy communicatively coupled to the particular endpoint, an instruction to route data associated with the particular endpoint over a terrestrial cellular network in advance of the second outage.
18 . The system of claim 17 , wherein the first outages are experienced by at least one of the endpoints in a first geographical region among the geographical regions,
wherein the second outage is predicted to impact the particular endpoint in a second geographical region among the geographical regions, and
wherein the first geographical region borders the second geographical region.
19 . The system of claim 17 , wherein predicting, based on the outage data and the contextual data, the second outage comprising the interruption in communication between the satellite network and the particular endpoint comprises:
training a machine learning (ML) model based on a first portion of the outage data and a first portion of the contextual data;
in response to training the ML model, predicting the second outage based on the ML model, a second portion of the outage data, and a second portion of the contextual data.
20 . The system of claim 17 , wherein the first outages and the second outage are caused by at least one of:
an inoperative LEO satellite among the LEO satellites; or
a weather pattern moving across the geographical regions.