Scientists at the MIT Media Lab have designed a wireless system that leverages the inexpensive RFID tags already on hundreds of billions of products to detect potential food contamination—with no hardware alterations needed. With the simple, scalable system, the scientists aim to offer food-safety detection to the general public.
Food safety instances have made headlines around the world for causing illness and death almost every year for the past 20 years. Back in 2008, for example, 50,000 babies in China were hospitalized after eating infant formula contaminated with melamine, an organic compound used to produce plastics, which is poisonous in high concentrations. Then, this April, over 100 people died in Indonesia from drinking alcohol adulterated, partly, with methanol, a deadly alcohol frequently used to dilute liquor for sale in black markets globally.
The researchers’ system, referred to as RFIQ, comprises of a reader that senses tiny changes in wireless signals discharged from RFID tags when the signals interact with food. For this research, they concentrated on alcohol and baby formula, but in the future, consumers might have their own reader and software to conduct food-safety sensing before purchasing almost any product. Systems could also be executed in smart fridges or in supermarket back rooms to constantly ping an RFID tag to automatically sense food spoilage, the scientists say.
The technology is based on the fact that certain variations in the signals released from an RFID tag match the levels of certain contaminants within that product. A machine-learning model “learns” those correlations and, given a new material, can predict if the material is pure or contaminated, and the level of concentration. In experiments, the system sensed alcohol diluted with methanol with 97% accuracy and baby formula spiked with melamine with 96% accuracy.
“In recent years, there have been so many hazards related to food and drinks we could have avoided if we all had tools to sense food quality and safety ourselves,” says Fadel Adib, an assistant professor at the Media Lab who is co-author on a paper illustrating the system, which is being presented at the ACM Workshop on Hot Topics in Networks. “We want to democratize food quality and safety, and bring it to the hands of everyone.”
The co-authors of the paper include: postdoc and first author Unsoo Ha, postdoc Yunfei Ma, visiting researcher Zexuan Zhong, and electrical engineering and computer science graduate student Tzu-Ming Hsu.
The power of “weak coupling”
Other sensors have also been engineered for sensing spoilage or chemicals in food. But those are very specialized systems, where the sensor is coated with chemicals and trained to sense particular contaminations. The goal of the Media Lab scientists instead is for broader sensing.
We’ve moved this detection purely to the computation side, where you’re going to use the same very cheap sensor for products as varied as alcohol and baby formula.
Fadel Adib, Assistant Professor, Media Lab.
RFID tags are stickers having minute, ultra-high-frequency antennas. They can be seen on food products and other items, and each costs about three to five cents. Traditionally, a wireless device known as a reader pings the tag, which powers up and releases a distinctive signal containing information about the product it’s fixed to.
The scientists’ system exploits the fact that, when RFID tags power up, the small electromagnetic waves they release travel into and are distorted by the ions and molecules of the contents in the container. This process is called “weak coupling.” Basically, if the material’s property changes, so do the properties of the signal.
A basic instance of feature distortion is with a container of air versus water. If a container is empty, the RFID will respond at about 950 MHz at all times. If it’s full of water, the water absorbs some of the frequency, and its highest response is about only 720 MHz. Feature distortions tend to be far more fine-grained with varied materials and varied contaminants. “That kind of information can be used to classify materials … [and] show different characteristics between impure and pure materials,” Ha says.
In the scientists’ system, a reader releases a wireless signal that powers the RFID tag on a food container. Electromagnetic waves enter the material within the container and return to the reader with phase (angle) and distorted amplitude (strength of signal).
When the reader extracts the signal characteristics, it transmits those data to a machine-learning model on a separate computer. In training, the scientists tell the model which feature changes relate to pure or impure materials. For this research, they used pure alcohol and alcohol contaminated with 25, 50, 75, and 100% methanol; baby formula was contaminated with a mixed percentage of melamine, from 0 to 30%.
“Then, the model will automatically learn which frequencies are most impacted by this type of impurity at this level of percentage,” Adib says. “Once we get a new sample, say, 20 percent methanol, the model extracts [the features] and weights them, and tells you, ‘I think with high accuracy that this is alcohol with 20 percent methanol.’”
Broadening the frequencies
The system’s concept develops from a method known as radio frequency spectroscopy, which excites a material with electromagnetic waves over a broad frequency and measures the different interactions to establish the material’s makeup.
But there was one huge difficulty in adapting this method for the system: RFID tags only power up at a very tight bandwidth wavering about 950 MHz. Extracting signals in that narrow bandwidth wouldn’t capture any useful information.
The scientists further developed on a sensing method they formulated earlier, called two-frequency excitation, which transmits two frequencies—one for activation, and one for sensing—to measure numerous more frequencies. The reader transmits a signal at about 950 MHz to power the RFID tag. When it stimulates, the reader transmits another frequency that sweeps a range of frequencies from around 400 to 800 MHz. It detects the feature variations spanning all these frequencies and feeds them to the reader.
“Given this response, it’s almost as if we have transformed cheap RFIDs into tiny radio frequency spectroscopes,” Adib says.
Since the shape of the container and other environmental aspects can impact the signal, the team is presently working on guaranteeing the system can account for those changes. They are also aiming to increase the system’s capabilities to sense many types of contaminants in numerous different materials.
We want to generalize to any environment. That requires us to be very robust, because you want to learn to extract the right signals and to eliminate the impact of the environment from what’s inside the material.
Fadel Adib, Assistant Professor, Media Lab.