Proof of Life

Summary

Cattle are some of the most valuable collateral in agriculture, and some of the hardest to verify. They move, get sold, get sick, and die. Lenders usually confirm a herd exists by sending someone out to count it, sometimes only once a year. Between counts, the loan is backed by paperwork.

That gap has been exploited, and at record cattle prices, it is more expensive than ever. At the same time, banks, private lenders, and blockchain platforms are looking for ways to finance or tokenize livestock as a real-world asset.

A ledger can make a record permanent, but it cannot make the record true. Something has to connect the token to the animal.

This paper describes how data from Seismi sensors, worn by each animal, can serve as continuous proof of life. The data can show that a specific animal:

• exists and is alive,
• is the animal it claims to be,
• is counted only once,
• is where it is supposed to be.

We describe the signals, the fraud patterns they are designed to catch, and how daily attestations can be written to any distributed ledger. We also describe planned safeguards that make the data itself hard to forge: signing on the device, tamper evidence in hardware, and an audit trail at tagging.

The problem: collateral that walks

Livestock lending has always run on trust and periodic inspection. Recent cases show what happens when the inspection comes too late.

• The "ghost herd." A Washington cattleman billed a major meatpacker and another company for the cost of buying and feeding about 265,000 cattle that did not exist. He collected about $244 million between roughly 2016 and 2020, and was sentenced to 11 years in federal prison¹. The story was later told in the public radio series Ghost Herd².
• An empire that wasn't there. A Kentucky operator presented lenders and investors with what looked like an 80,000-head operation and took in about $170 million. When the lender finally did a full physical count, the first in more than four years of lending, it found 8,916 animals³.
• Cattle sold that were never bought. In February 2026, federal prosecutors indicted five people tied to a Texas company. They allegedly collected more than $220 million from over 2,200 people, each told their money was buying specific individual cattle⁴. (The charges are allegations; the defendants are presumed innocent.)
• A forged bill of sale. In September 2026, a Missouri bank vice president pleaded guilty in a scheme that took at least $9.4 million. He borrowed money to buy cattle, never bought them, and gave a lender a bill of sale for 850 cow-calf pairs that the named seller says it never sold⁵.

These are not isolated cases. Court records describe the same handful of patterns again and again.

Why these schemes last so long

Bank examiners expect lenders to inspect livestock collateral periodically. Someone other than the loan officer should do the inspection, and the results should be compared against records of animals bought and sold¹¹.

In practice, that leaves long gaps. Many schemes surface only when the borrower misses payments⁹, or when a full count finally happens years in³. In one of the cases above, the fraudster was himself a bank loan officer⁵.

What lenders lack is a source of truth that does not depend on the borrower's paperwork or the loan officer's word.

Why it matters more now

Cattle prices hit records in 2026. Feeder steers at Oklahoma City reached an all-time high of $388.06 per hundredweight in the first week of May¹². At that price, a single 775-pound feeder steer was worth about $3,000.

Every phantom animal is now a bigger loss. And every honest animal is bigger collateral that lenders often can't fully count. In Brazil, the firm that structured a livestock-backed loan said the market traditionally discounted cattle heavily. For example, a cow valued at R$20,000 might count as only R$8,000 of collateral¹³.

Why a ledger alone doesn't solve it

Tokenization and distributed ledgers solve real problems:

• A shared record that cannot be quietly edited.
• Clear ownership and transfer.
• Visibility for multiple lenders, which helps prevent the same collateral from being pledged twice.

Livestock tokenization is already happening:

• Brazil. In July 2026, farmers tokenized dairy cows on the country's national exchange to secure credit. Collars monitored the animals' health, behavior, and location, and each animal got a unique registered code^{13, 14}.
• Argentina. A platform represents individual cattle as NFTs on a blockchain¹⁵, with the goal of helping producers access credit and insurance¹⁶.
• United States. A ranching technology company is promoting blockchain-based digital identities for individual animals¹⁷.

But a ledger only records what it is told. If the input is a spreadsheet of animals that don't exist, the ledger will faithfully preserve a lie.

Researchers call this the "oracle problem." Anything that feeds outside information into a blockchain becomes the point where trust can fail¹⁸. Researchers have studied the problem specifically for tracing physical, one-of-a-kind goods, where the record has to stay tied to a specific real object¹⁹.

Tokenized finance has a partial answer for other assets: proof of reserve. An independent feed continuously verifies that the assets backing a token actually exist. That feed can also be used to stop new tokens from being issued without backing²⁰.

Livestock need the same thing. The reserve is alive, and the only trustworthy witness is the animal itself.

The approach: the animal as its own witness

A Seismi sensor is attached to each animal. It continuously measures signals that only a living animal of that species produces:

• Movement and activity
• Heart rate
• Respiratory rate
• Temperature

Fixed readers at water points, and other locations in the pen or pasture, receive that data as the animals go about their day.

No single reading proves much. Together, over days and weeks, the readings form a record that is very hard to fake: a specific, living animal, behaving like a bovine, showing up where it should.

The signals Seismi data can provide

1. Identity binding. At processing, each Seismi sensor is paired to the animal's existing identity.

• Official ID. Since November 2024, certain classes of U.S. cattle moving interstate, mostly breeding stock, must carry official ear tags that are both visually and electronically readable²¹. Pairing to that official EID ties the sensor record to the identity regulators and buyers already use.
• One record each. From that point, each sensor speaks for one animal, and each animal has one record.
• Optional biometric. A photo of the animal's muzzle can be enrolled as a second, tag-independent identity check. Muzzle patterns are individually distinctive. One 2026 study re-identified the same 48 cattle 131 days later with 96% accuracy²².

2. A living-animal signature.

• Gait and activity. Bovine movement has a recognizable pattern, including gait, activity cycles, and rest.
• Rumination. Cattle spend hours a day chewing cud, in a distinctive rhythm that an ear-tag accelerometer can detect^{23, 24}. A cart, a person, or a dog does not ruminate.
• Vital signs. Heart rate, respiration, and temperature fall within bovine ranges and change in physiologically plausible ways.
• Individual baseline. Animals differ in measurable ways. In one study, a rumination model trained on each individual cow reached 98.4% accuracy, versus 86.2% for a single model shared across the herd²³. A sudden, unexplained change in an animal's baseline can indicate the sensor has been moved to a different animal.

A sensor taped to a cart, carried by a person, strapped to a dog, or left in a box produces patterns that are different. The system is designed to detect those differences.

3. Regular presence at water.

• Every animal drinks. Beef cattle in cool climates drink on average 4–7 times a day²⁵. On large grazing properties, most visits to water come within 48 hours of the animal's previous visit²⁶.
• Each visit is a record. Readers at water points record every visit with a timestamp, as a routine part of the animal's life.
• The expected interval depends on the operation. It is set for each one: daily or more often in a feedlot, a day or two on open range. An animal that misses its expected visits is flagged for someone to go check.
• Presence is the point. An animal at the water is not always drinking; one study put the odds at 42–54%²⁷. Proof of life needs presence, not intake.

4. Location and co-location.

• Precise in-pen location is proven. A commercial ear-tag system has located feedlot cattle to within about half a meter, every few seconds. It has also measured how many animals stand within 3 meters of each other²⁸.
• Several technologies can do it. Our current implementation uses Bluetooth direction finding, introduced in Bluetooth 5.1²⁹. The same approach can be built on:
  • Bluetooth Channel Sounding, introduced in Bluetooth 6.0, which estimates distance to roughly 10–30 cm³⁰
  • ultra-wideband (UWB), already paired with accelerometers to monitor cattle behavior³¹
  • satellite positioning, or future location technologies.

  The method does not depend on any one of them.

• Proximity shows duplicates. Location shows how close animals are to each other over time.

5. Weight.

• Weighing at the water point. Walk-over scales at water points record each animal's tag and live weight automatically as it walks in to drink³².
• Verified weight. Paired with a Seismi sensor, weight becomes a verified measurement of a known, living animal. Weight is a large part of what collateral is worth.

6. Tamper and loss signals. When a sensor is removed or falls off, its data changes in recognizable ways:

• Heartbeat and respiration disappear.
• Temperature drifts toward ambient.
• Motion goes still, or stops matching a living animal.

The system is designed to flag these events for someone to investigate.

Making the data itself trustworthy (planned)

Signals only help if the data carrying them is genuine. A fraudster who cannot fake an animal may try to fake the data instead: editing stored records, replaying old readings, or running modified firmware. Where there is opportunity for fraud, fraudsters will be creative. Seismi is designing three safeguards against this. They are planned, not yet shipping.

Signed at the source. Each Seismi device is planned to carry its own cryptographic key pair, created when the device is manufactured.

• A private key that never leaves the device. The device signs the data it produces. Each reading can be traced to the exact sensor that made it, and any later edit breaks the signature.
• A public key anyone can check. The matching public key is recorded at manufacture and can be published on a ledger. Investors and auditors can then verify the provenance of the data themselves.
• A proven pattern. Smartphones already work this way. An Android phone can prove that a key lives in secure hardware, using a root key provisioned to the device at the factory³³.

Tamper evidence in hardware. Signal 6 describes how removal shows up in the data. Planned hardware safeguards would also record tampering in the device itself.

• One-way markers. One-time-programmable electronic fuses (eFuses) can permanently record an event, such as the case being opened, in a way that cannot be reset. Phones use the same kind of tamper-evident storage to refuse older, vulnerable software, a protection called rollback protection³⁴. Apple devices carry a liquid contact indicator that turns red on contact with water³⁵.
• Tampering is reported, not hidden. A device that detects tampering keeps working. It signs and transmits its data as before, with a flag saying it has been tampered with. A false alarm costs only a check, and a real tamper cannot be quietly erased.

Accountability at tagging. Seismi plans to add audit features at enrollment, the moment a sensor is paired to an animal:

• Who tagged it. The person who tags each animal is cryptographically recorded, so they can be held accountable later.
• What was tagged. A photo of the animal is recorded at tagging. Content-provenance standards such as C2PA can bind the image to its capture, so an edited or swapped photo can be detected³⁶.
• Signed by the phone. Phone platforms can prove a request came from a genuine app on genuine hardware: App Attest on iOS³⁷ and key attestation on Android³³.
• On the same ledger. The enrollment record can be stored alongside the animal's health data, on the same ledger.

Together, these safeguards significantly raise the cost of fraud. Forging a record would mean defeating a device's secure hardware, not just editing a spreadsheet.

Fraud scenarios and how sensor data can detect them

A closer look: detecting duplicate tags on one animal

Official ear tags can be duplicated. An animal can carry two or three tags, and nothing about a tag itself says how many others are on the same head.

Sensor data changes that. Two Seismi sensors on the same animal share one body. That shows up in several independent ways:

• Motion: their traces move together, step for step.
• Vital signs: their heart and breathing rhythms line up.
• Water visits: they arrive and leave at the same moments.
• Location: they stay at a distance from each other that two separate animals would not keep, hour after hour.

Two different animals may graze together or drink side by side. How many neighbors an animal keeps nearby varies through the day and is measurable²⁸. But two animals do not stay unnaturally close for an entire day, and they do not share a heartbeat.

Location may prove the most reliable of these signals, more so than motion or temperature.

This is an anomaly-detection problem, similar to how card networks spot fraudulent transactions. Statistics and machine learning can weigh the signals together. No single signal has to be perfect. The combination makes the duplicate stand out.

To build and test this, Seismi plans to deliberately double-tag a subset of animals in upcoming clinical trials. That will produce labeled examples of exactly this case.

From sensor to ledger: the attestation pipeline

Seismi's role is to turn raw sensor data into a trustworthy statement about each animal that a lender or tokenization platform can rely on. The same feed is independent of both the borrower's paperwork and the loan officer.

1. 1

   Enroll

   Each animal is tagged and paired to its official ID, optionally with a muzzle photo. In a planned audit feature, the tagger's identity and a provenance-protected photo are recorded at the same time. When a loan finances a purchase, funds can be released in stages as the purchased animals are enrolled and verified at the receiving site.

2. 2

   Collect

   Sensors on each animal transmit data to readers at water points and other locations. The data is uploaded to the Seismi platform. In the planned design, each reading is signed by the device that made it.

3. 3

   Evaluate

   For each animal, each day, the platform can evaluate the proof-of-life signals: identity, living-animal signature, presence, location, weight, and tamper or loss events. It assigns a status, such as verified alive and on site, flagged, or not seen, along with a confidence level.

4. 4

   Attest

   The platform can produce a daily attestation for each animal and for the herd as a whole.

   • Each animal. Each attestation includes the animal ID, date, status, confidence, and a summary of the supporting evidence. Each can be digitally signed.
   • The herd. The herd attestation answers: "Of the animals pledged, how many were verified alive and on site today?"
   • Value. Combined with verified weights and a public market price, such as USDA-reported feeder cattle prices¹², it can also estimate the herd's current collateral value.
5. 5

   Anchor

   The attestation, or a cryptographic hash of it, can be written to the lender's or platform's ledger of choice. That could be a public blockchain, a permissioned ledger, or a lender's own system of record.

   • In the simplest design, the full sensor data stays off-chain.
   • It remains available, with the producer's authorization, for audit or dispute resolution. Anyone holding the data can recompute the hash and confirm the attestation was not altered.
   • Alternatively, the sensor data itself can be compressed and written to the ledger, so it cannot be altered later and stays directly accessible to investors.
   • Each device's public key can be published on the same ledger, so anyone can confirm which sensor produced the data.
6. 6

   Act

   The lender or platform applies its own rules to the attested data. Examples:

   • Gate token issuance. Issue tokens only for animals with active proof of life, the way proof-of-reserve systems can stop new tokens being issued without backing²⁰.
   • Recalculate loan-to-value daily as head count, weight, and price change.
   • Trigger a check. Start a physical inspection or margin call when verified count or value drops below a threshold.
   • Hold a buffer. Keep extra verified animals in reserve. The Brazilian deal included about 20% more animals than needed so collateral could be maintained if animals died¹³.
   • Swap animals properly. Retire or swap a token when an animal dies or is sold. The replacement must carry its own proof-of-life history before it counts.
   • Close out at sale or harvest. When a pledged animal is read at a sale barn or packing plant, the platform can close its token and direct the proceeds. Several U.S. states already run central filing systems that tell registered buyers which farm products carry a lender's security interest³⁸. An animal-level ledger can deliver that notice at the moment of sale.
   • Verify insurance claims. A sensor-recorded death event can support an insurance claim.

Variations

The method is not tied to a particular ledger, device, or deployment:

• Ledger and token. Public, permissioned, or private ledgers. Fungible herd-level tokens, individual tokens per animal, or traditional loan records.
• Frequency. Attestations can run hourly, daily, or weekly.
• Sensors. Ear tags today, with the same method applying to collars, boluses, nose rings, or other animal-worn devices.
• Readers. Fixed readers at water points, feed bunks, gates, alleys, and scales, or mobile and handheld readers for spot audits.
• Identity. Official EID, visual tags, muzzle or other biometrics, or any combination.
• Who attests. Seismi alone, or Seismi alongside other independent data providers in a multi-party or decentralized oracle network.
• Cross-checks. Other independent records, such as water meter volumes, feed deliveries, or scale weights, can confirm the totals add up.

Integration is through Seismi's API, the same one available to third-party farm management systems today.

What this means

For lenders.

• Collateral is checked every day instead of once a year.
• Shortfalls show up in days, not after the money is gone.
• Verification no longer rests on the borrower's paperwork or a single loan officer.
• Physical inspections become targeted rather than routine.

For producers.

• Honest operators can prove their herd is real without hosting inspectors.
• Collateral that is verified every day may not need to be discounted as heavily. That can mean access to more capital, at better terms, from more sources, including tokenized credit markets that were not open to them before.
• The same sensors also monitor herd health, so the data pays for itself twice.

For tokenization platforms. The token is tied to a living animal with continuous evidence behind it, not to a spreadsheet. That is the difference between a real-world asset and a claim about one.

Limitations and validation

Proof of life is a strong signal, not a guarantee. We want to be clear about where this stands.

• Validation is in progress. The detection methods described here are being developed and validated in field trials, including deliberate double-tagging to test duplicate detection. Where this paper says the system "can" or "is designed to" detect something, that is the goal being validated.
• Sensor accuracy varies. Published accuracy for detecting behaviors like rumination with ear-tag sensors ranges widely by device, algorithm, and conditions^{23, 24}. It can drop under heat stress³⁹. That is why proof of life combines many signals rather than relying on one.
• Flags are prompts, not verdicts. Sensors fail, readers miss reads, and animals sometimes skip a drink. A flag should trigger a check, not an automatic default.
• Planned safeguards are not yet built. Signing on the device, hardware tamper evidence, and the audit trail at tagging are planned features. Signing proves where data came from, not that the sensor is on the right animal. That is what the other signals are for.
• Biometrics are a complement. Muzzle-based identification is promising, but current studies are small and short-term²².
• It complements audits. Especially early on, sensor attestation is best used alongside periodic physical verification. Over time, it can reduce how often those visits are needed.
• Producer data rights matter. Sensor data belongs to the operation that generates it. Sharing attestations with lenders or platforms should happen only with the producer's authorization.

Public disclosure

Seismi is publishing this paper as an open technical description of using animal-worn sensor data to verify livestock collateral and tokenized livestock assets. We believe tying real-world animal data to financial records is a natural next step for the industry. We want the approach to be widely understood and available.

Work with us

If you are a lender, a tokenization or distributed-ledger platform, or a producer interested in using sensor data to verify livestock assets, we'd like to talk.

Get in Touch Cattle Sensor Integrations

References

1. U.S. Attorney's Office, Eastern District of Washington. "Tri-Cities Rancher Sentenced to Eleven Years in Federal Prison and Ordered to Pay $244 Million in Restitution for 'Ghost Cattle' Fraud." October 2022. https://www.justice.gov/usao-edwa/pr/tri-cities-rancher-sentenced-eleven-years-federal-prison-and-ordered-pay-244-million
2. KUOW and Northwest Public Broadcasting. Ghost Herd (podcast series). 2023. https://www.npr.org/podcasts/1141136265/ghost-herd
3. Patrick, M., and Thomas, P. "The Cattle Empire That Turned Out to Be a Giant Ponzi Scheme." The Wall Street Journal, June 2026. https://www.wsj.com/finance/cattle-empire-ponzi-scheme-3d245791
4. U.S. Attorney's Office, Northern District of Texas. "Five Federally Indicted for Alleged $220 Million Nationwide Cattle Fraud Scheme; One is Fugitive on FBI Wanted List." February 12, 2026. https://www.justice.gov/usao-ndtx/pr/five-federally-indicted-alleged-220-million-nationwide-cattle-fraud-scheme-one
5. U.S. Attorney's Office, Western District of Missouri. "Butler, Missouri Bank Vice President and Former Pastor Pleads Guilty to Financial Fraud Involving an Alleged Cattle-Buying Scheme." September 9, 2026. https://www.justice.gov/usao-wdmo/pr/butler-missouri-bank-vice-president-and-former-pastor-pleads-guilty-financial-fraud
6. Texas and Southwestern Cattle Raisers Association. "The beef with cattle fraud." https://tscra.org/the-beef-with-cattle-fraud/
7. U.S. Attorney's Office, Western District of Missouri. Guilty plea in $8 million cattle fraud schemes (Asbury). 2011. https://www.justice.gov/archive/usao/mow/news2011/asbury.ple.html
8. CBS Texas. "North Texas Man Accused Of Fraud, Used Cattle As $5.8M In Collateral." 2018. https://www.cbsnews.com/texas/news/north-texas-man-fraud-cattle-5m-collateral/
9. ACAMS. "All Hat, No Cattle: What the AML Professional Needs to Know About Cattle Fraud." 2020. https://www.acams.org/sites/default/files/2020-08/ALL%20HAT,%20NO%20CATTLE-%20What%20the%20AML%20Professional%20Needs%20to%20Know%20About%20Cattle%20Fraud.pdf
10. U.S. Attorney's Office, District of Montana. "Lewistown man sentenced for bank fraud in obtaining loans for cattle ranching business." 2023. https://www.justice.gov/usao-mt/pr/lewistown-man-sentenced-bank-fraud-obtaining-loans-cattle-ranching-business
11. Office of the Comptroller of the Currency. Comptroller's Handbook: Agricultural Lending. March 2020. https://www.occ.gov/topics/supervision-and-examination/credit/commercial-credit/agricultural-lending.html
12. USDA Economic Research Service. Livestock, Dairy, and Poultry Outlook: May 2026 (LDP-M-383). https://www.ers.usda.gov/media/20881/ldp-m-383.pdf
13. Simão, A. "Vacas tokenizadas movimentam R$ 100 mil na primeira operação na B3." CNN Brasil, July 21, 2026. https://www.cnnbrasil.com.br/agro/vacas-tokenizadas-movimentam-r-100-mil-na-primeira-operacao-na-b3/
14. Acuna, O. "Brazilian farmers tokenized dairy cows to get loans, bypassing bank lending limits." CoinDesk, July 24, 2026. https://www.coindesk.com/markets/2026/07/24/brazilian-farmers-tokenized-dairy-cows-to-get-loans-bypassing-bank-lending-limits
15. Contexto Ganadero. "Argentina y Bolivia ya usan tokens ganaderos, pero Colombia sigue a la espera." 2025. https://www.contextoganadero.com/tendencias/argentina-y-bolivia-ya-usan-tokens-ganaderos-pero-colombia-sigue-a-la-espera
16. Periódico El Emprendedor. "Blockchain permite convertir ganados en una garantía para conseguir capital." https://www.periodicoelemprendedor.com/ve/breves/item/3942-blockchain-permite-convertir-ganados-en-una-garant%C3%ADa-para-conseguir-capital.html
17. Hinkle, J. A. "Blockchain is quietly moving into the cattle industry." TheStreet, February 25, 2026. https://www.thestreet.com/crypto/innovation/blockchain-is-quietly-moving-into-the-cattle-industry
18. Caldarelli, G. "Understanding the Blockchain Oracle Problem: A Call for Action." Information 11, no. 11 (2020): 509. https://doi.org/10.3390/info11110509
19. Caldarelli, G., Rossignoli, C., and Zardini, A. "Overcoming the Blockchain Oracle Problem in the Traceability of Non-Fungible Products." Sustainability 12, no. 6 (2020): 2391. https://doi.org/10.3390/su12062391
20. Chainlink. "What Are Proof of Reserves?" Chainlink Education Hub. https://chain.link/education-hub/proof-of-reserves
21. Congressional Research Service. "Animal Disease Traceability: Electronic Identification Requirements" (R48169). 2024. https://www.congress.gov/crs-product/R48169
22. Yardımcı, S. O., Orhan, İ., and Bahar, F. İ. "Short-Term Longitudinal Biometric Identification of Cattle Using Planum Nasolabiale Images and a Two-Stage Deep Learning Framework." Veterinary Sciences 13, no. 9 (2026): 924. https://doi.org/10.3390/vetsci13090924
23. Chang, A. Z., Fogarty, E. S., Moraes, L. E., García-Guerra, A., Swain, D. L., and Trotter, M. G. "Detection of rumination in cattle using an accelerometer ear-tag: A comparison of analytical methods and individual animal and generic models." Computers and Electronics in Agriculture 192 (2022): 106595. https://doi.org/10.1016/j.compag.2021.106595
24. Wolfger, B., et al. "Technical note: Accuracy of an ear tag-attached accelerometer to monitor rumination and feeding behavior in feedlot cattle." Journal of Animal Science (2015). https://pubmed.ncbi.nlm.nih.gov/26115302/
25. Williams, L. R., et al. "Drinking frequency effects on the performance of cattle: a systematic review." Journal of Animal Physiology and Animal Nutrition (2017). https://doi.org/10.1111/jpn.12640
26. Williams, L. R., Fox, D. R., Bishop-Hurley, G. J., and Swain, D. L. "Use of radio frequency identification (RFID) technology to record grazing beef cattle water point use." Computers and Electronics in Agriculture 156 (2019): 193–202. https://doi.org/10.1016/j.compag.2018.11.025
27. Shane, D. D., White, B. J., Larson, R. L., Amrine, D. E., and Kramer, J. L. "Probabilities of cattle participating in eating and drinking behavior when located at feeding and watering locations by a real time location system." Computers and Electronics in Agriculture 127 (2016): 460–466. https://doi.org/10.1016/j.compag.2016.07.005
28. White, B. J., Goehl, D. R., McMeniman, J. P., Batterham, T., Booker, C. W., and McMullen, C. "Determination of Behavioral Changes Associated with Bovine Respiratory Disease in Australian Feedlots." Animals 13, no. 23 (2023): 3692. https://doi.org/10.3390/ani13233692
29. Pau, G., Arena, F., Gebremariam, Y. E., and You, I. "Bluetooth 5.1: An Analysis of Direction Finding Capability for High-Precision Location Services." Sensors 21, no. 11 (2021): 3589. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8196737/
30. audioXpress. "Bluetooth SIG Introduces Updated Specification for True Distance Awareness." September 2024. https://audioxpress.com/news/bluetooth-sig-introduces-updated-specification-for-true-distance-awareness
31. Benaissa, S., et al. "Improved cattle behaviour monitoring by combining Ultra-Wideband location and accelerometer data." Animal 17 (2023): 100730. https://doi.org/10.1016/j.animal.2023.100730
32. FutureBeef. "Automated cattle weighing: what are your options?" 2026. https://futurebeef.com.au/resources/automated-cattle-weighing/
33. Android Developers. "Verify hardware-backed key pairs with key attestation." https://developer.android.com/privacy-and-security/security-key-attestation
34. Android Open Source Project. "Use Verified Boot" (rollback protection). https://source.android.com/docs/security/features/verifiedboot/verified-boot
35. Apple Support. "Water and other liquid damage to iPhone or iPod isn't covered by warranty." https://support.apple.com/en-us/109350
36. Coalition for Content Provenance and Authenticity (C2PA). Content Credentials: C2PA Technical Specification, version 2.4. https://spec.c2pa.org/specifications/specifications/2.4/specs/C2PA_Specification.html
37. Apple Developer Documentation. "Establishing your app's integrity" (App Attest). https://developer.apple.com/documentation/devicecheck/establishing-your-app-s-integrity
38. National Agricultural Law Center. "Figuring the Federal Farm Products Rule: Complying with the Notice Requirements" (fact sheet). https://nationalaglawcenter.org/wp-content/uploads/assets/articles/Figuring-the-Federal-Farm-Products-Complying-with-the-Notice-Requirements-Fact-Sheet-1-1.pdf
39. USDA Agricultural Research Service. "Effects of heat stress on the accuracy of an ear-tag accelerometer for monitoring rumination and eating behavior in dairy-beef cross cattle using an automated gold standard." https://www.ars.usda.gov/research/publications/publication/?seqNo115=413272

© 2026 Seismi, Inc. Version 1.0, published October 1, 2026.