Instruments flood a site with raw data. A data centre in a 20 ft container runs the models beside them and sends only results upstream. Seven decisions, each worked out live.
Container
20 ft · 6 racks
Cryptography
SHA-256 · ECDSA P-256
Chapters
7
01 · Raw is heavy
Bitrate
Raw video outruns any backhaul.
An uncompressed frame is width × height × bits per pixel; 8-bit 4:2:0 colour costs 12 bits a pixel. Multiply by the frame rate and the cameras on site and the stream dwarfs a satellite or microwave link long before it reaches a region.
Resolution
Frame rate
Per camera, raw
2.99 Gb/s
Site total, raw
23.89 Gb/s
Per day
257.99 TB
Frames to analyse each second
240
02 · Inference at the edge
Results
Ship the answer, not the pixels.
The GPU servers in the container’s racks turn frames into detections. A binary record of class (2 bytes), confidence (4), box (4 × 4), timestamp (8) and camera (2) is 32 bytes; that, not the frame, is what leaves the site.
Record encoding
Record size
32 bytes
Results upstream
41.0 kb/s
Reduction against raw
583,200 : 1
Results per day
442.4 MB
03 · Latency budget
Round trip
Light in fibre sets the floor.
Glass slows light to about c/1.47, close to 4.9 µs a kilometre. A control loop that must answer within its deadline pays that twice, out and back, before a single switch, queue or model runs. Distance alone can rule a region out.
Round trip against the deadline
Edge, 0.5 km
8.0 ms
Fibre round trip
11.8 ms
Cloud, total
19.8 ms
Deadline
20 ms
Farthest region that still meets it
1,224 km
04 · Store and forward
Outage
When the link drops, the site keeps working.
Inference never stops for the backhaul. Results queue on local disk in order, and when the link returns the backlog drains at whatever capacity the live stream leaves free: drain time = backlog ÷ (link − live rate).
Backhaul
Disk for the queue
Live rate upstream
41.0 kb/s
Backlog at the end of the outage
663.6 MB
Drain time after restore
4.4 min
Outage the disk can hold
12.4 years
05 · Attestation
SHA-256
Prove what booted before trusting what it says.
Each boot stage hashes the next before handing over control and extends a platform configuration register: PCR = SHA-256(PCR ‖ digest). A register can only be extended, never set, so its final value proves the whole chain. A remote verifier compares a signed quote of it with the golden value. Tamper with a stage and watch every later value change.
Model matches digest 22a2043d13a5e837… · PCR c8c2e6e596843a93…
Verifier
matches the golden value
06 · Signed plan
ECDSA P-256
A model update is a signed plan, rolled out in waves.
The plan names the model by its digest and the waves of the fleet that take it. It is signed with ECDSA over P-256 and SHA-256; every node verifies the signature against a key it already trusts before it fetches a byte. Change one character after signing and no node moves.
The same box plans and checks a survey flight. Ground sample distance is sensor width × altitude ÷ (focal length × image width); overlap sets the spacing of the lines. In Canada, Part IX of the Canadian Aviation Regulations caps routine flight at 122 m above ground.
Camera
Ground sample distance
2.74 cm/px
Footprint per image
150 × 100 m
Lines over 600 × 400 m
10 · 44 m apart
Images at 75 % forward overlap
257
Flight at 10 m/s
6.40 km · 10.7 min
Altitude against 122 m
22 m under
Building the site…
Edge · ExplainerInteractive
Compute where the data is born.
Instruments flood a site with raw data. A data centre in a 20 ft container runs the models beside them and sends only results upstream. Seven decisions, each worked out live.
01 · Raw is heavyBitrate
Raw video outruns any backhaul.
An uncompressed frame is width × height × bits per pixel; 8-bit 4:2:0 colour costs 12 bits a pixel. Multiply by the frame rate and the cameras on site and the stream dwarfs a satellite or microwave link long before it reaches a region.
Resolution
Frame rate
Per camera, raw
–
Site total, raw
–
Per day
–
Frames to analyse each second
–
02 · Inference at the edgeResults
Ship the answer, not the pixels.
The GPU servers in the container’s racks turn frames into detections. A binary record of class (2 bytes), confidence (4), box (4 × 4), timestamp (8) and camera (2) is 32 bytes; that, not the frame, is what leaves the site.
Record encoding
Record size
–
Results upstream
–
Reduction against raw
–
Results per day
–
03 · Latency budgetRound trip
Light in fibre sets the floor.
Glass slows light to about c/1.47, close to 4.9 µs a kilometre. A control loop that must answer within its deadline pays that twice, out and back, before a single switch, queue or model runs. Distance alone can rule a region out.
Round trip against the deadline
Farthest region that still meets it
–
04 · Store and forwardOutage
When the link drops, the site keeps working.
Inference never stops for the backhaul. Results queue on local disk in order, and when the link returns the backlog drains at whatever capacity the live stream leaves free: drain time = backlog ÷ (link − live rate).
Backhaul
Disk for the queue
Live rate upstream
–
Backlog at the end of the outage
–
Drain time after restore
–
Outage the disk can hold
–
05 · AttestationSHA-256
Prove what booted before trusting what it says.
Each boot stage hashes the next before handing over control and extends a platform configuration register: PCR = SHA-256(PCR ‖ digest). A register can only be extended, never set, so its final value proves the whole chain. A remote verifier compares a signed quote of it with the golden value. Tamper with a stage and watch every later value change.
Tamper with a stage
Verifier
–
06 · Signed planECDSA P-256
A model update is a signed plan, rolled out in waves.
The plan names the model by its digest and the waves of the fleet that take it. It is signed with ECDSA over P-256 and SHA-256; every node verifies the signature against a key it already trusts before it fetches a byte. Change one character after signing and no node moves.
Plan in transit
Signature
–
Verification on each node
–
Nodes updating now
–
07 · Physical AISurvey
A drone that flies its own survey.
The same box plans and checks a survey flight. Ground sample distance is sensor width × altitude ÷ (focal length × image width); overlap sets the spacing of the lines. In Canada, Part IX of the Canadian Aviation Regulations caps routine flight at 122 m above ground.
Camera
Ground sample distance
–
Footprint per image
–
Lines over 600 × 400 m
–
Images at 75 % forward overlap
–
Flight at 10 m/s
–
Altitude against 122 m
–
PrivacyConfidentialité
Your privacy choicesVos choix de confidentialité
We use only the cookies this site needs to work. With your permission, we also use optional cookies to remember preferences, measure visits and show content from other providers. Nothing optional runs until you choose, and you can change your choice at any time in Privacy choices. Details are in our cookie and storage notice.
Notice at collection: this site uses the cookies it needs to work. Optional cookies, which remember preferences, measure visits and show content from other providers, stay off until you allow them. We do not sell your personal information or share it for cross-context behavioral advertising. You can opt out at any time with Do Not Sell or Share My Personal Information or in Privacy choices. Details, including the categories of personal information we collect and how long we keep them, are in our privacy policy and our cookie and storage notice.
Nous n'utilisons que les témoins nécessaires au fonctionnement de ce site. Avec votre permission, nous utilisons aussi des témoins facultatifs pour retenir vos préférences, mesurer la fréquentation et afficher des contenus d'autres fournisseurs. Rien de facultatif ne fonctionne tant que vous n'avez pas choisi, et vous pouvez modifier votre choix en tout temps dans Choix de confidentialité. Les détails figurent dans notre avis sur les témoins et le stockage.
Avis au moment de la collecte : ce site utilise les témoins nécessaires à son fonctionnement. Les témoins facultatifs, qui retiennent vos préférences, mesurent la fréquentation et affichent des contenus d'autres fournisseurs, restent désactivés tant que vous ne les autorisez pas. Nous ne vendons pas vos renseignements personnels et ne les communiquons pas à des fins de publicité comportementale intercontextuelle. Vous pouvez vous y opposer en tout temps au moyen du lien Ne pas vendre ni communiquer mes renseignements personnels ou dans Choix de confidentialité. Les détails, y compris les catégories de renseignements personnels recueillis et leur durée de conservation, figurent dans notre politique de confidentialité et notre avis sur les témoins et le stockage.
Global Privacy Control detected. We have honored it as your request to opt out of the sale and sharing of your personal information and of targeted advertising.Signal Global Privacy Control détecté. Nous l'avons respecté comme une demande d'opposition à la vente et à la communication de vos renseignements personnels ainsi qu'à la publicité ciblée.
We use only the cookies this site needs to work. With your permission, we also use optional cookies to remember preferences, measure visits and show content from other providers. Nothing optional runs until you choose, and you can change your choice at any time in Privacy choices. Details are in our cookie and storage notice.
Notice at collection: this site uses the cookies it needs to work. Optional cookies, which remember preferences, measure visits and show content from other providers, stay off until you allow them. We do not sell your personal information or share it for cross-context behavioral advertising. You can opt out at any time with Do Not Sell or Share My Personal Information or in Privacy choices. Details, including the categories of personal information we collect and how long we keep them, are in our privacy policy and our cookie and storage notice.
Nous n'utilisons que les témoins nécessaires au fonctionnement de ce site. Avec votre permission, nous utilisons aussi des témoins facultatifs pour retenir vos préférences, mesurer la fréquentation et afficher des contenus d'autres fournisseurs. Rien de facultatif ne fonctionne tant que vous n'avez pas choisi, et vous pouvez modifier votre choix en tout temps dans Choix de confidentialité. Les détails figurent dans notre avis sur les témoins et le stockage.
Avis au moment de la collecte : ce site utilise les témoins nécessaires à son fonctionnement. Les témoins facultatifs, qui retiennent vos préférences, mesurent la fréquentation et affichent des contenus d'autres fournisseurs, restent désactivés tant que vous ne les autorisez pas. Nous ne vendons pas vos renseignements personnels et ne les communiquons pas à des fins de publicité comportementale intercontextuelle. Vous pouvez vous y opposer en tout temps au moyen du lien Ne pas vendre ni communiquer mes renseignements personnels ou dans Choix de confidentialité. Les détails, y compris les catégories de renseignements personnels recueillis et leur durée de conservation, figurent dans notre politique de confidentialité et notre avis sur les témoins et le stockage.
Global Privacy Control detected. We have honored it as your request to opt out of the sale and sharing of your personal information and of targeted advertising.Signal Global Privacy Control détecté. Nous l'avons respecté comme une demande d'opposition à la vente et à la communication de vos renseignements personnels ainsi qu'à la publicité ciblée.
Your browser sent a Global Privacy Control signal. We treat it as a refusal for the purposes marked below, and those switches stay off.Votre navigateur a envoyé un signal Global Privacy Control. Nous le traitons comme un refus pour les finalités indiquées ci-dessous, et ces interrupteurs restent désactivés.
Strictly necessaryStrictement nécessaires
Always activeToujours actif
Needed for the site to work, to keep you signed in, to protect it from attacks and to remember your privacy choices and the settings you pick. Always on.Nécessaires au fonctionnement du site, pour vous garder connecté, le protéger contre les attaques et retenir vos choix de confidentialité et les réglages que vous choisissez. Toujours actifs.
1 service
Consent preferences
ProviderFournisseur
Nexus Observatory Inc.
RetentionConservation
6 months, or 12 months where an opt-out law applies (cookie); browser session (region); until cleared (receipts)
Remember optional settings beyond the ones you pick yourself. Off until you allow it.Retenir des réglages facultatifs autres que ceux que vous choisissez vous-même. Désactivés tant que vous ne les autorisez pas.
Off
Count visits and see how pages are used, so we can improve the site. Off until you allow it.Compter les visites et voir comment les pages sont utilisées, afin d'améliorer le site. Désactivés tant que vous ne les autorisez pas.
Off
Show videos, maps and live data layers from other providers. They receive your IP address and may set their own cookies. Off until you allow it, or load a single item once.Afficher des vidéos, des cartes et des couches de données en direct d'autres fournisseurs. Ceux-ci reçoivent votre adresse IP et peuvent déposer leurs propres témoins. Désactivés tant que vous ne les autorisez pas, ou chargez un seul élément une fois.
4 services
Site Planner "Plan anywhere" vector basemap and label glyphs (OpenMapTiles schema)
Site Planner "Plan anywhere" elevation (Terrarium) for hillshade, line of sight, profiles and sky view
ProviderFournisseur
Terrain: Mapzen Terrain Tiles (AWS Open Data), incl. SRTM, GMTED, ETOPO1, USGS 3DEP and Canadian DEM data
Hosts contactedHôtes contactés
elevation-tiles-prod.s3.amazonaws.com
Observatory Builder weather layers (GDPS 2 m air temperature, global; radar precipitation rate, Canada) and the 2 m air temperature at the site (WMS GetFeatureInfo) that seeds the synthetic run
ProviderFournisseur
Data Source: Environment and Climate Change Canada
Hosts contactedHôtes contactés
geo.weather.gc.ca
Observatory Builder hazard layer and report (earthquakes of magnitude 2.5 and above in the past 30 days, GeoJSON summary feed)
ProviderFournisseur
Earthquakes: U.S. Geological Survey
Hosts contactedHôtes contactés
earthquake.usgs.gov
Off
Measure campaigns and personalise advertising. We use none today. Off until you allow it, and refused when your browser sends Global Privacy Control.Mesurer les campagnes et personnaliser la publicité. Nous n'en utilisons aucun actuellement. Désactivés tant que vous ne les autorisez pas, et refusés lorsque votre navigateur transmet le signal Global Privacy Control.
Nexus Observatory Inc. keeps a record of each choice as proof: a random consent ID, the time, your region and the rules applied, the choices, the notice shown, the page, a shortened IP address, a one-way code of your browser type and, if you are signed in, your account. Records are kept while your choice applies and for 3 years after, then deleted.Nexus Observatory Inc. conserve un registre de chaque choix à titre de preuve : un identifiant de consentement aléatoire, l'heure, votre région et les règles appliquées, les choix, l'avis affiché, la page, une adresse IP abrégée, un code à sens unique du type de navigateur et, si vous êtes connecté, votre compte. Les registres sont conservés tant que votre choix s'applique et pendant 3 ans par la suite, puis supprimés.
Your consent ID (quote it in any privacy request):Votre identifiant de consentement (à indiquer dans toute demande relative à la confidentialité) :