Project

RF-EATS: Food and liquid sensing in practical environments using RFIDs

Copyright

MIT

Signal Kinetics

Groups

RF-EATS is a new system that can verify the authenticity of food and liquids in closed containers without opening them or requiring any contact with their contents

What kinds of food/liquids can RF-EATS  be used to verify?

Below are some of the applications for which we have successfully demonstrated RF-EATS's ability to verify and detect:

  • Fake medicine. Fake medicine is a major challenge in many developing-world countries, leading to dozens of fatalities every year. A recent incident involved fake cough medicine bottles, where 90% of the active ingredient was replaced with diethylene glycol, a compound used in making antifreeze agents. 
  • Adulterated baby formula. In 2008, the Chinese milk scandal broke out after the hospitalization of 50,000 babies due to kidney damage. Manufacturers had watered down baby formulas up to 83% and mixed them with melamine CAS NO. 108-78-1, a compound used in making plastics. The purpose of adding melamine (by manufacturers) was to conceal dilution by artificially increasing protein levels.
  • Tainted alcohol and diluted alcohol. Tainted alcohol is an ongoing proble… View full description

RF-EATS is a new system that can verify the authenticity of food and liquids in closed containers without opening them or requiring any contact with their contents

What kinds of food/liquids can RF-EATS  be used to verify?

Below are some of the applications for which we have successfully demonstrated RF-EATS's ability to verify and detect:

  • Fake medicine. Fake medicine is a major challenge in many developing-world countries, leading to dozens of fatalities every year. A recent incident involved fake cough medicine bottles, where 90% of the active ingredient was replaced with diethylene glycol, a compound used in making antifreeze agents. 
  • Adulterated baby formula. In 2008, the Chinese milk scandal broke out after the hospitalization of 50,000 babies due to kidney damage. Manufacturers had watered down baby formulas up to 83% and mixed them with melamine CAS NO. 108-78-1, a compound used in making plastics. The purpose of adding melamine (by manufacturers) was to conceal dilution by artificially increasing protein levels.
  • Tainted alcohol and diluted alcohol. Tainted alcohol is an ongoing problem in many developing-world countries, including China, Indonesia, Iran, Turkey, India, and Mexico. Alcohol is tainted by mixing it with cheaper methanol, and consuming it leads to hundreds of cases of blindness and death every year.
  • Fake extra-virgin olive oil. Recent studies have shown that 69% of US-imported extra virgin olive oil has been adulterated by mixing it with cheaper oils (e.g., peanut oil). This can lead to health hazards for consumers with (peanut) allergies. Standard adulteration levels range between 70-80%.
  • Wine fraud. Wine fraud takes many forms. A common one involves selling consumers wine vintages that are dated to earlier years, artificially inflating their price.
  • Counterfeit perfume. Counterfeit beauty products abound, leading Estée Lauder to confiscate over 2.6 million counterfeit items in 2016 alone. Many such products are sold online.

How accurate is the system?

Our results demonstrate that RF-EATS can achieve over 90% classification accuracy across a wide variety of applications

How does RF-EATS work?

RF-EATS is the first RFID-based system that can noninvasively sense food and liquids in closed containers and operate correctly in environments it hasn't seen before. RFIDs are inexpensive, battery-less tags that are placed on billions of products worldwide, including food items. RF-EATS leverages the near-field coupling between a tag’s antenna and a container's contents to sense them noninvasively. 


At a high level, the RFID's signal is impacted by the dielectric of the content inside the container. So, when the content changes, its dielectric change will impact the RFID's response.

The challenge, however, is that RFID signals are not just impacted by materials inside a container, but also by other objects in the environment around them, including other items, furniture, and even the human body.

Our solution is a new wireless AI model that enables RF-EATS to verify food and liquids despite changes in the surrounding environment. It is based on a novel RF (radio frequency) kernel function, called multipath kernel, that can be used to simulate different wireless environments. By simulating different environments,  it can adapt to them.

Our results show that the system's accuracy is indeed dependent on the content of the container, specifically on dielectric differences between authentic and fake content.

To learn more about how the system works, read our paper.

This project is supported by a J-WAFS seed grant.

If you're interested in exploring the potential of using RF-EATS for detecting different kinds of contaminants or material properties, contact us at rfiq@media.mit.edu.

Copyright

MIT