MIT Research News' Journal
[Most Recent Entries]
[Calendar View]
Thursday, February 20th, 2020
| Time |
Event |
| 12:00a |
Cryptographic “tag of everything” could protect the supply chain To combat supply chain counterfeiting, which can cost companies billions of dollars annually, MIT researchers have invented a cryptographic ID tag that’s small enough to fit on virtually any product and verify its authenticity.
A 2018 report from the Organization for Economic Co-operation and Development estimates about $2 trillion worth of counterfeit goods will be sold worldwide in 2020. That’s bad news for consumers and companies that order parts from different sources worldwide to build products.
Counterfeiters tend to use complex routes that include many checkpoints, making it challenging to verifying their origins and authenticity. Consequently, companies can end up with imitation parts. Wireless ID tags are becoming increasingly popular for authenticating assets as they change hands at each checkpoint. But these tags come with various size, cost, energy, and security tradeoffs that limit their potential.
Popular radio-frequency identification (RFID) tags, for instance, are too large to fit on tiny objects such as medical and industrial components, automotive parts, or silicon chips. RFID tags also contain no tough security measures. Some tags are built with encryption schemes to protect against cloning and ward off hackers, but they’re large and power hungry. Shrinking the tags means giving up both the antenna package — which enables radio-frequency communication — and the ability to run strong encryption.
In a paper presented yesterday at the IEEE International Solid-State Circuits Conference (ISSCC), the researchers describe an ID chip that navigates all those tradeoffs. It’s millimeter-sized and runs on relatively low levels of power supplied by photovoltaic diodes. It also transmits data at far ranges, using a power-free “backscatter” technique that operates at a frequency hundreds of times higher than RFIDs. Algorithm optimization techniques also enable the chip to run a popular cryptography scheme that guarantees secure communications using extremely low energy.
“We call it the ‘tag of everything.’ And everything should mean everything,” says co-author Ruonan Han, an associate professor in the Department of Electrical Engineering and Computer Science and head of the Terahertz Integrated Electronics Group in the Microsystems Technology Laboratories (MTL). “If I want to track the logistics of, say, a single bolt or tooth implant or silicon chip, current RFID tags don’t enable that. We built a low-cost, tiny chip without packaging, batteries, or other external components, that stores and transmits sensitive data.”
Joining Han on the paper are: graduate students Mohamed I. Ibrahim, Muhammad Ibrahim Wasiq Khan, and Chiraag S. Juvekar; former postdoc associate Wanyeong Jung; former postdoc Rabia Tugce Yazicigil, who is currently an assistant professor at Boston University and a visiting scholar at MIT; and Anantha P. Chandrakasan, who is the dean of the MIT School of Engineering and the Vannevar Bush Professor of Electrical Engineering and Computer Science.
System integration
The work began as a means of creating better RFID tags. The team wanted to do away with packaging, which makes the tags bulky and increases manufacturing cost. They also wanted communication in the high terahertz frequency between microwave and infrared radiation — around 100 gigahertz and 10 terahertz — that enables chip integration of an antenna array and wireless communications at greater reader distances. Finally, they wanted cryptographic protocols because RFID tags can be scanned by essentially any reader and transmit their data indiscriminately.
But including all those functions would normally require building a fairly large chip. Instead, the researchers came up with “a pretty big system integration,” Ibrahim says, that enabled putting everything on a monolithic — meaning, not layered — silicon chip that was only about 1.6 square millimeters.
One innovation is an array of small antennas that transmit data back and forth via backscattering between the tag and reader. Backscatter, used commonly in RFID technologies, happens when a tag reflects an input signal back to a reader with slight modulations that correspond to data transmitted. In the researchers’ system, the antennas use some signal splitting and mixing techniques to backscatter signals in the terahertz range. Those signals first connect with the reader and then send data for encryption.
Implemented into the antenna array is a “beam steering” function, where the antennas focus signals toward a reader, making them more efficient, increasing signal strength and range, and reducing interference. This is the first demonstration of beam steering by a backscattering tag, according to the researchers.
Tiny holes in the antennas allow light from the reader to pass through to photodiodes underneath that convert the light into about 1 volt of electricity. That powers up the chip’s processor, which runs the chip’s “elliptic-curve-cryptography” (ECC) scheme. ECC uses a combination of private keys (known only to a user) and public keys (disseminated widely) to keep communications private. In the researchers’ system, the tag uses a private key and a reader’s public key to identify itself only to valid readers. That means any eavesdropper who doesn’t possess the reader’s private key should not be able to identify which tag is part of the protocol by monitoring just the wireless link.
Optimizing the cryptographic code and hardware lets the scheme run on an energy-efficient and small processor, Yazicigil says. “It’s always a tradeoff,” she says. “If you tolerate a higher-power budget and larger size, you can include cryptography. But the challenge is having security in such a small tag with a low-power budget.”
Pushing the limits
Currently, the signal range sits around 5 centimeters, which is considered a far-field range — and allows for convenient use of a portable tag scanner. Next, the researchers hope to “push the limits” of the range even further, Ibrahim says. Eventually, they’d like many of the tags to ping one reader positioned somewhere far away in, say, a receiving room at a supply chain checkpoint. Many assets could then be verified rapidly.
“We think we can have a reader as a central hub that doesn’t have to come close to the tag, and all these chips can beam steer their signals to talk to that one reader,” Ibrahim says.
The researchers also hope to fully power the chip through the terahertz signals themselves, eliminating any need for photodiodes.
The chips are so small, easy to make, and inexpensive that they can also be embedded into larger silicon computer chips, which are especially popular targets for counterfeiting.
“The U.S. semiconductor industry suffered $7 billion to $10 billion in losses annually because of counterfeit chips,” Wasiq Khan says. “Our chip can be seamlessly integrated into other electronic chips for security purposes, so it could have huge impact on industry. Our chips cost a few cents each, but the technology is priceless,” he quipped. | | 10:59a |
Artificial intelligence yields new antibiotic Using a machine-learning algorithm, MIT researchers have identified a powerful new antibiotic compound. In laboratory tests, the drug killed many of the world’s most problematic disease-causing bacteria, including some strains that are resistant to all known antibiotics. It also cleared infections in two different mouse models.
The computer model, which can screen more than a hundred million chemical compounds in a matter of days, is designed to pick out potential antibiotics that kill bacteria using different mechanisms than those of existing drugs.
“We wanted to develop a platform that would allow us to harness the power of artificial intelligence to usher in a new age of antibiotic drug discovery,” says James Collins, the Termeer Professor of Medical Engineering and Science in MIT’s Institute for Medical Engineering and Science (IMES) and Department of Biological Engineering. “Our approach revealed this amazing molecule which is arguably one of the more powerful antibiotics that has been discovered.”
In their new study, the researchers also identified several other promising antibiotic candidates, which they plan to test further. They believe the model could also be used to design new drugs, based on what it has learned about chemical structures that enable drugs to kill bacteria.
“The machine learning model can explore, in silico, large chemical spaces that can be prohibitively expensive for traditional experimental approaches,” says Regina Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL).
Barzilay and Collins, who are faculty co-leads for MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health, are the senior authors of the study, which appears today in Cell. The first author of the paper is Jonathan Stokes, a postdoc at MIT and the Broad Institute of MIT and Harvard.
A new pipeline
Over the past few decades, very few new antibiotics have been developed, and most of those newly approved antibiotics are slightly different variants of existing drugs. Current methods for screening new antibiotics are often prohibitively costly, require a significant time investment, and are usually limited to a narrow spectrum of chemical diversity.
“We’re facing a growing crisis around antibiotic resistance, and this situation is being generated by both an increasing number of pathogens becoming resistant to existing antibiotics, and an anemic pipeline in the biotech and pharmaceutical industries for new antibiotics,” Collins says.
To try to find completely novel compounds, he teamed up with Barzilay, Professor Tommi Jaakkola, and their students Kevin Yang, Kyle Swanson, and Wengong Jin, who have previously developed machine-learning computer models that can be trained to analyze the molecular structures of compounds and correlate them with particular traits, such as the ability to kill bacteria.
The idea of using predictive computer models for “in silico” screening is not new, but until now, these models were not sufficiently accurate to transform drug discovery. Previously, molecules were represented as vectors reflecting the presence or absence of certain chemical groups. However, the new neural networks can learn these representations automatically, mapping molecules into continuous vectors which are subsequently used to predict their properties.
In this case, the researchers designed their model to look for chemical features that make molecules effective at killing E. coli. To do so, they trained the model on about 2,500 molecules, including about 1,700 FDA-approved drugs and a set of 800 natural products with diverse structures and a wide range of bioactivities.
Once the model was trained, the researchers tested it on the Broad Institute’s Drug Repurposing Hub, a library of about 6,000 compounds. The model picked out one molecule that was predicted to have strong antibacterial activity and had a chemical structure different from any existing antibiotics. Using a different machine-learning model, the researchers also showed that this molecule would likely have low toxicity to human cells.
This molecule, which the researchers decided to call halicin, after the fictional artificial intelligence system from “2001: A Space Odyssey,” has been previously investigated as possible diabetes drug. The researchers tested it against dozens of bacterial strains isolated from patients and grown in lab dishes, and found that it was able to kill many that are resistant to treatment, including Clostridium difficile, Acinetobacter baumannii, and Mycobacterium tuberculosis. The drug worked against every species that they tested, with the exception of Pseudomonas aeruginosa, a difficult-to-treat lung pathogen.
To test halicin’s effectiveness in living animals, the researchers used it to treat mice infected with A. baumannii, a bacterium that has infected many U.S. soldiers stationed in Iraq and Afghanistan. The strain of A. baumannii that they used is resistant to all known antibiotics, but application of a halicin-containing ointment completely cleared the infections within 24 hours.
Preliminary studies suggest that halicin kills bacteria by disrupting their ability to maintain an electrochemical gradient across their cell membranes. This gradient is necessary, among other functions, to produce ATP (molecules that cells use to store energy), so if the gradient breaks down, the cells die. This type of killing mechanism could be difficult for bacteria to develop resistance to, the researchers say.
“When you’re dealing with a molecule that likely associates with membrane components, a cell can’t necessarily acquire a single mutation or a couple of mutations to change the chemistry of the outer membrane. Mutations like that tend to be far more complex to acquire evolutionarily,” Stokes says.
In this study, the researchers found that E. coli did not develop any resistance to halicin during a 30-day treatment period. In contrast, the bacteria started to develop resistance to the antibiotic ciprofloxacin within one to three days, and after 30 days, the bacteria were about 200 times more resistant to ciprofloxacin than they were at the beginning of the experiment.
The researchers plan to pursue further studies of halicin, working with a pharmaceutical company or nonprofit organization, in hopes of developing it for use in humans.
Optimized molecules
After identifying halicin, the researchers also used their model to screen more than 100 million molecules selected from the ZINC15 database, an online collection of about 1.5 billion chemical compounds. This screen, which took only three days, identified 23 candidates that were structurally dissimilar from existing antibiotics and predicted to be nontoxic to human cells.
In laboratory tests against five species of bacteria, the researchers found that eight of the molecules showed antibacterial activity, and two were particularly powerful. The researchers now plan to test these molecules further, and also to screen more of the ZINC15 database.
The researchers also plan to use their model to design new antibiotics and to optimize existing molecules. For example, they could train the model to add features that would make a particular antibiotic target only certain bacteria, preventing it from killing beneficial bacteria in a patient’s digestive tract.
“This groundbreaking work signifies a paradigm shift in antibiotic discovery and indeed in drug discovery more generally,” says Roy Kishony, a professor of biology and computer science at Technion (the Israel Institute of Technology), who was not involved in the study. “Beyond in silica screens, this approach will allow using deep learning at all stages of antibiotic development, from discovery to improved efficacy and toxicity through drug modifications and medicinal chemistry.”
The research was funded by the Abdul Latif Jameel Clinic for Machine Learning in Health, the Defense Threat Reduction Agency, the Broad Institute, the DARPA Make-It Program, the Canadian Institutes of Health Research, the Canadian Foundation for Innovation, the Canada Research Chairs Program, the Banting Fellowships Program, the Human Frontier Science Program, the Pershing Square Foundation, the Swiss National Science Foundation, a National Institutes of Health Early Investigator Award, the National Science Foundation Graduate Research Fellowship Program, and a gift from Anita and Josh Bekenstein. | | 6:04p |
The trouble with round numbers Do you have a monthly car payment, or a similar loan? Is each payment a nice round number, like $300? If so, you are hardly alone. But the appeal of that easy-to-remember payment figure may be costing you money.
That’s one implication of a new study co-authored by an MIT economist, which shows how much consumers prefer monthly payment figures that are multiples of $100 — indeed, the number of monthly consumer payments at dollar figures just above such multiples drops by 16 percent. That likely makes monthly budgeting easier for people to calculate. But as the study also shows, people select potentially unfavorable loan terms as a result.
“People budget with these round numbers and are trained to think in these monthly payment terms, going for the smallest monthly payment possible,” says MIT economist Christopher Palmer, co-author of a newly published paper detailing the results. “In particular, people really bunch around $200 or $300 or $400 a month in payments, which probably keeps them from overspending month-to-month, but it still might not be the best approach if it leads them to pay more interest over the length of the loan.”
In fact, after digging into auto loans held by more than 2 million people, Palmer and his colleagues found that this is precisely the case: Given multiple financing options, many people smooth out the monthly figures, often at less money per payment, but with notably increased long-term costs.
And while lower monthly payments are important for many, the study shows that borrowers often take such an approach when they can afford to pay more.
“One thing we did [in this study] is look at data for people with a lot of debt capacity, a low debt-to-income ratio or high credit scores, and even those people seem to make decisions based on the monthly payment amount, while ignoring the total cost of the loan,” notes Palmer, the Albert and Jeanne Clear Career Development Professor in the MIT Sloan School of Management.
The paper, “Monthly Payment Targeting and the Demand for Maturity,” appears in advance online form in the Review of Financial Studies. In addition to Palmer, the authors are Bronson Argyle and Taylor Nadauld, finance professors at Brigham Young University’s Marriott School of Business.
The natural experiment
To conduct the study, Palmer, Argyle, and Nadauld studied auto loan contracts held by 2.4 million borrowers, using 319 different lenders. The anonymized information came from a data company that works with lending firms. About 70 percent of the loans originated during the period 2012-2015, though some date to 2005. The researchers also examined another 1.3 million loan applications to get a further sense of borrowers’ fiscal circumstances.
A key feature of the study — giving the research a quasiexperimental form — involves its use of FICO scores, a basic credit rating. FICO scores range from 300 to 850, but at certain thresholds, some banks offer markedly different loans to customers. When you have a FICO score of 700, which is close to average, you may qualify for much better terms than if your score is slightly lower.
“If you have a 701 FICO score, at some banks you can get a much lower interest rate than someone with a 699 FICO score, even though if you asked the company that makes FICO scores, you’re basically the same person,” Palmer says. “But if a bank is treating similar consumers very differently, it becomes this nice laboratory for a natural experiment.”
That is, if borrowers offered a variety of loan terms have the same tendency — such as winding up with round-number monthly payments — it suggests how strongly that tendency is rooted in the behavior of consumers. The phenomenon of round-number monthly payments was quickly obvious to the researchers.
“This just jumped out of the data,” Palmer says. “You plot the data and people are bunching at hundred-dollar multiples.”
So what’s the problem, exactly?
To see why this can be a bad personal-finance habit — and clearly is, for some people — note that loans with lower monthly payments will have a greater long-term total cost, given initial purchases of the same amount.
That point applies to a second finding of the study: When consumers are offered loan terms, they respond more to changes in the maturity — the length of the loan — than changes in the interest rate.
As Palmer, Argyle, and Nadauld found, a bank offer of a 10 percent increase in loan length raises the chances that a borrower will accept the terms by 8.3 percentage points. But a bank offer of a 10 percent decrease in the interest rate raises the chances that a borrower will accept the terms by only 1 percentage point.
Why is this? As it happens, changing the maturity of the loan has a bigger impact on monthly payments, which lets more consumers bring those payments to the magic levels of $200, $300, and $400.
However, changes in loan length also bring higher long-term costs for consumers. Consider a $20,000 loan with a five-year maturity and a 5 percent interest rate. Increasing the maturity of that loan by one year lowers monthly payments by $55 but raises total interest paid by $546.
In short, by having a nose for round numbers, consumers in the new study really are paying more for their cars.
Lessons about loans
That said, Palmer acknowledges that for different people, there is not necessarily one clear answer about which approach is better: lower monthly payments or a lower long-term repayment.
“There’s not great theory on what you should do,” Palmer says. “What we would say you should do is figure out if that tradeoff worth it for you. If having lower payments today is worth paying more interest over the life of the loan, great, and there could be many reasons for that. But for many people I’d expect it could be better to try to get that loan over with more quickly with a shorter maturity.”
Palmer hopes that one practical implication of the study would be getting people to recognize that there is a tradeoff in the first place.
“Many people think monthly payments are the responsible way to talk about how much a car costs,” Palmer says. “But if you tell me you’re only going to spend $300 a month on a car, I can sell you a Mercedes if I make the car loan long enough.”
As the study shows, a significant number of people are gravitating toward a rule of thumb — round-number payments — when doing homework and comparison-shopping about loans is more useful. Still, perhaps it is the nature of auto purchasing that leads people to underinvest in shopping for loans.
“I get to test-drive the car,” Palmer says. “I don’t get to test-drive the loan.” |
|