This article was first published on <a href="https://news.mit.edu/2023/tackling-counterfeit-seeds-unclonable-labels-0322" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>. &nbsp;<hr>Average crop yields in Africa are consistently far below those expected, and one significant reason is the prevalence of counterfeit seeds whose germination rates are far lower than those of the genuine ones. The World Bank estimates that as much as half of all seeds sold in some African countries are fake, which could help to account for crop production that is far below potential.There have been many attempts to prevent this counterfeiting through tracking labels, but none have proved effective; among other issues, such labels have been vulnerable to hacking because of the deterministic nature of their encoding systems. But now, a team of MIT researchers has come up with a kind of tiny, biodegradable tag that can be applied directly to the seeds themselves, and that provides a unique randomly created code that cannot be duplicated.The new system, which uses minuscule dots of silk-based material, each containing a unique combination of different chemical signatures, is described today in the journal Science Advances in a <a href="https://www.science.org/doi/10.1126/sciadv.adf1978" target="_blank">paper</a> by MIT&rsquo;s dean of engineering Anantha Chandrakasan, professor of civil and environmental engineering Benedetto Marelli, postdoc Hui Sun, and graduate student Saurav Maji.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1679919238187-MIT-SeedTracking-02-press.jpg" style="width: 766px;" class="fr-fic fr-dib">Researchers developed a simple drop-casting approach that produces tags that are less than a tenth of an inch in diameter. Credit: Courtesy of the researchersThe problem of counterfeiting is an enormous one globally, the researchers point out, affecting everything from drugs to luxury goods, and many different systems have been developed to try to combat this. But there has been less attention to the problem in the area of agriculture, even though the consequences can be severe. In sub-Saharan Africa, for example, the World Bank estimates that counterfeit seeds are a significant factor in crop yields that average less than one-fifth of the potential for maize, and less than one-third for rice.Marelli explains that a key to the new system is creating a randomly-produced physical object whose exact composition is virtually impossible to duplicate. The labels they create &ldquo;leverage randomness and uncertainty in the process of application, to generate unique signature features that can be read, and that cannot be replicated,&rdquo; he says.What they&rsquo;re dealing with, Sun adds, &ldquo;is the very old job of trying, basically, not to get your stuff stolen. And you can try as much as you can, but eventually somebody is always smart enough to figure out how to do it, so nothing is really unbreakable. But the idea is, it&rsquo;s almost impossible, if not impossible, to replicate it, or it takes so much effort that it&rsquo;s not worth it anymore.&rdquo;The idea of an &ldquo;unclonable&rdquo; code was originally developed as a way of protecting the authenticity of computer chips, explains Chandrakasan, who is the Vannevar Bush Professor of Electrical Engineering and Computer Science. &ldquo;In integrated circuits, individual transistors have slightly different properties coined device variations,&rdquo; he explains, &ldquo;and you could then use that variability and combine that variability with higher-level circuits to create a unique ID for the device. And once you have that, then you can use that unique ID as a part of a security protocol. Something like transistor variability is hard to replicate from device to device, so that&rsquo;s what gives it its uniqueness, versus storing a particular fixed ID.&rdquo; The concept is based on what are known as physically unclonable functions, or PUFs.The team decided to try to apply that PUF principle to the problem of fake seeds, and the use of silk proteins was a natural choice because the material is not only harmless to the environment but also classified by the Food and Drug Administration in the &ldquo;generally recognized as safe&rdquo; category, so it requires no special approval for use on food products.&ldquo;You could coat it on top of seeds,&rdquo; Maji says, &ldquo;and if you synthesize silk in a certain way, it will also have natural random variations. So that&rsquo;s the idea, that every seed or every bag could have a unique signature.&rdquo;Developing effective secure system solutions has long been one of Chandrakasan&rsquo;s specialties, while Marelli has spent many years developing systems for applying silk coatings to a variety of fruits, vegetables, and seeds, so their collaboration was a natural for developing such a silk-based coding system toward enhanced security.&ldquo;The challenge was what type of form factor to give to silk,&rdquo; Sun says, &ldquo;so that it can be fabricated very easily.&rdquo; They developed a simple drop-casting approach that produces tags that are less than one-tenth of an inch in diameter. The second challenge was to develop &ldquo;a way where we can read the uniqueness, in also a very high throughput and easy way.&rdquo;For the unique silk-based codes, Marelli says, &ldquo;eventually we found a way to add a color to these microparticles so that they assemble in random structures.&rdquo; The resulting unique patterns can be read out not only by a spectrograph or a portable microscope, but even by an ordinary cellphone camera with a macro lens. This image can be processed locally to generate the PUF code and then sent to the cloud and compared with a secure database to ensure the authenticity of the product. &ldquo;It&rsquo;s random so that people cannot easily replicate it,&rdquo; says Sun. &ldquo;People cannot predict it without measuring it.&rdquo;And the number of possible permutations that could result from the way they mix four basic types of colored silk nanoparticles is astronomical. &ldquo;We were able to show that with a minimal amount of silk, we were able to generate 128 random bits of security,&rdquo; Maji says. &ldquo;So this gives rise to 2 to the power 128 possible combinations, which is extremely difficult to crack given the computational capabilities of the state-of-the-art computing systems.&rdquo;Marelli says that &ldquo;for us, it&rsquo;s a good test bed in order to think out-of-the-box, and how we can have a path that somehow is more democratic.&rdquo; In this case, that means &ldquo;something that you can literally read with your phone, and you can fabricate by simply drop casting a solution, without using any advanced manufacturing technique, without going in a clean room.&rdquo;Some additional work will be needed to make this a practical commercial product, Chandrakasan says. &ldquo;There will have to be a development for at-scale reading&rdquo; via smartphones. &ldquo;So, that&rsquo;s clearly a future opportunity.&rdquo; But the principle now shows a clear path to the day when &ldquo;a farmer could at least, maybe not every seed, but could maybe take some random seeds in a particular batch and verify them,&rdquo; he says.The research was partially supported by the U.S. Office of Naval research and the National Science Foundation, Analog Devices Inc., an EECS Mathworks fellowship, and a Paul M. Cook Career Development Professorship.<hr>&quot;Reprinted with permission of <a href="https://news.mit.edu/2023/tackling-counterfeit-seeds-unclonable-labels-0322" id="isPasted" rel="noopener noreferrer" target="_blank">MIT News</a>&quot; &nbsp;

As a way to reduce seed counterfeiting, MIT researchers developed a silk-based tag that, when applied to seeds, provides a unique code that cannot be duplicated. Credit: Photograph courtesy of the researchers. Edited by Jose-Luis Olivares, MIT

Fake seeds can cost farmers more than two-thirds of expected crop yields and threaten food security. Trackable silk labels could help.

Tackling counterfeit seeds with "unclonable" labels

Detailed virtual copies of physical objects, called digital twins, are opening doors for better products across automotive, health care, aerospace and other industries. According to a new study, cybersecurity may also fit neatly into the digital twin portfolio.&nbsp;As more robots and other manufacturing equipment become remotely accessible, new entry points for malicious cyberattacks are created. To keep pace with the growing cyber threat, a team of researchers at the National Institute of Standards and Technology (NIST) and the University of Michigan devised a cybersecurity framework that brings digital twin technology together with machine learning and human expertise to flag indicators of cyberattacks.In a paper published in&nbsp;<a data-extlink="" href="https://ieeexplore.ieee.org/document/10049398" target="_blank">IEEE Transactions on Automation Science and Engineering</a>, the NIST and University of Michigan researchers demonstrated the feasibility of their strategy by detecting cyberattacks aimed at a 3D printer in their lab. They also note that the framework could be applied to a broad range of manufacturing technologies.&nbsp;Cyberattacks can be incredibly subtle and thus difficult to detect or differentiate from other, sometimes more routine, system anomalies. Operational data describing what is occurring within machines &mdash; sensor data, error signals, digital commands being issued or executed, for instance &mdash; could support cyberattack detection. However, directly accessing this kind of data in near real time from operational technology (OT) devices, such as a 3D printer, could put the performance and safety of the process on the factory floor at risk.&ldquo;Typically, I have observed that manufacturing cybersecurity strategies rely on copies of network traffic that do not always help us see what is occurring inside a piece of machinery or process,&rdquo; said NIST mechanical engineer Michael Pease, a co-author of the study. &ldquo;As a result, some OT cybersecurity strategies seem analogous to observing the operations from the outside through a window; however, adversaries might have found a way onto the floor.&rdquo;Without looking under the hood of the hardware, cybersecurity professionals may be leaving room for malicious actors to operate undetected.&nbsp;<h2>Taking a Look in the Digital Mirror</h2>Digital twins aren&rsquo;t your run-of-the-mill computer models. They are closely tied to their physical counterparts, from which they extract data and run alongside in near real time. So, when it&rsquo;s not possible to inspect a physical machine while it&rsquo;s in operation, its digital twin is the next best thing.&nbsp;In recent years, digital twins of manufacturing machinery have armed engineers with an abundance of operational data, helping them accomplish a variety of feats (without impacting performance or safety), including predicting when parts will start to break down and require maintenance.&nbsp;In addition to spotting routine indicators of wear and tear, digital twins could help find something more within manufacturing data, the authors of the study say.&ldquo;Because manufacturing processes produce such rich data sets &mdash; temperature, voltage, current &mdash; and they are so repetitive, there are opportunities to detect anomalies that stick out, including cyberattacks,&rdquo; said Dawn Tilbury, a professor of mechanical engineering at the University of Michigan and study co-author.&nbsp;To seize the opportunity presented by digital twins for tighter cybersecurity, the researchers developed a framework entailing a new strategy, which they tested out on an off-the-shelf 3D printer.The team built a digital twin to emulate the 3D printing process and provided it with information from the real printer. As the printer built a part (a plastic hourglass in this case), computer programs monitored and analyzed continuous data streams including both measured temperatures from the physical printing head and the simulated temperatures being computed in real time by the digital twin.&nbsp;The researchers launched waves of disturbances at the printer. Some were innocent anomalies, such as an external fan causing the printer to cool, but others, some of which caused the printer to incorrectly report its temperature readings, represented something more nefarious.&nbsp;So, even with the wealth of information at hand, how did the team&rsquo;s computer programs distinguish a cyberattack from something more routine? The framework&rsquo;s answer is to use a process of elimination.&nbsp;The programs analyzing both the real and digital printers were pattern-recognizing machine learning models trained on normal operating data, which is included in the paper, in bulk. In other words, the models were adept at recognizing what the printer looked like under normal conditions, also meaning they could tell when things were out of the ordinary.&nbsp;If these models detected an irregularity, they passed the baton off to other computer models that checked whether the strange signals were consistent with anything in a library of known issues, such as the printer&rsquo;s fan cooling its printing head more than expected. Then the system categorized the irregularity as an expected anomaly or a potential cyber threat.&nbsp;In the last step, a human expert is meant to interpret the system&rsquo;s finding and then make a decision.&nbsp;&ldquo;The framework provides tools to systematically formalize the subject matter expert&rsquo;s knowledge on anomaly detection. If the framework hasn&rsquo;t seen a certain anomaly before, a subject matter expert can analyze the collected data to provide further insights to be integrated into and improve the system,&rdquo; said lead-author Efe Balta, a former mechanical engineering graduate student at the University of Michigan and now a postdoctoral researcher at ETH Zurich.Generally speaking, the expert would either confirm the cybersecurity system&rsquo;s suspicions or teach it a new anomaly to store in the database. And then as time goes on, the models in the system would theoretically learn more and more, and the human expert would need to teach them less and less.&nbsp;In the case of the 3D printer, the team checked its cybersecurity system&rsquo;s work and found it was able to correctly sort the cyberattacks from normal anomalies by analyzing physical and emulated data.&nbsp;But despite the promising showing, the researchers plan to study how the framework responds to more varied and aggressive attacks in the future, ensuring the strategy is reliable and scalable. Their next steps will likely also include applying the strategy to a fleet of printers at once, to see if the expanded coverage either hurts or helps their detection capabilities.&ldquo;With further research, this framework could potentially be a huge win-win for both maintenance as well as monitoring for indications of compromised OT systems,&rdquo; Pease said.<hr>Paper: E. C. Balta, M. Pease, J. Moyne, K. Barton and D. M. Tilbury. Cyber-Attack Detection Digital Twins for Cyber-Physical Manufacturing Systems. IEEE Transactions on Automation Science and Engineering. Published online Feb. 22, 2023. DOI: <a data-extlink="" href="https://doi.org/10.1109/TASE.2023.3243147%20" target="_blank">10.1109/TASE.2023.3243147</a> &nbsp;This article was first published here:&nbsp;<a href="https://www.nist.gov/news-events/news/2023/02/how-digital-twins-could-protect-manufacturers-cyberattacks" id="isPasted" rel="noopener noreferrer" target="_blank">https://www.nist.gov/news-events/news/2023/02/how-digital-twins-could-protect-manufacturers-cyberattacks</a>

A new and improved strategy for detecting cyberattacks on manufacturing systems, such as 3D printers, involves using AI to monitor a digital twin that mimics and is fed real-time data from the physical system. Credit: N. Hanacek/NIST

Detailed virtual copies of physical objects, called digital twins, are opening doors for better products across automotive, health care, aerospace and other industries. According to a new study, cybersecurity may also fit neatly into the digital twin portfolio. 

How Digital Twins Could Protect Manufacturers From Cyberattacks

This is the third&nbsp;article in a&nbsp;<a href="https://www.wevolver.com/article/iot-needs-trust-the-current-state-of-iot-security" rel="noopener noreferrer" target="_blank">3-part series</a>&nbsp;that explains the current state of IoT security and introduces cutting-edge, highly secure MCUs from Alif SemiconductorTM.&nbsp;There is a surge of interest in Machine Learning (ML) and Artificial Intelligence (AI) applications. This interest is partly driven by the explosion in data generation, which enables modern enterprises to access very large volumes of digital data from a variety of different sources, such as business information systems and internet-connected devices. Based on these data, enterprises can develop ML systems that exhibit better performance than ever before.Introducing edge machine learning (EdgeML)Until recently, most ML systems and applications were deployed within cloud computing infrastructures. This provided the means for managing large volumes of data while easing access to the computing capacity needed for training and executing ML models. In recent years there has been a shift of data from remote cloud data centers to local edge clusters and devices, i.e., a shift to the edge computing paradigm. This shift is motivated by the need to support real-time and power-efficient applications that operate close to the field instead of transferring large volumes of data to the cloud. The advent of edge computing has given rise to EdgeML, a novel paradigm for developing and deploying ML applications. Rather than moving data over wide area networks and running ML models inside the cloud, EdgeML applications execute models within various edge computers and devices. This includes edge clusters, gateways, embedded devices, Internet of Things (IoT) devices, and microcontrollers.ML designers leverage edge computing and edge ML to achieve the following benefits for their applications:<ul><li>Low-Latency Operations for Mission Critical Tasks: Edge processing can be much faster than cloud processing, as it does not incur the latency of transferring data from the source to the cloud. This makes EdgeML suitable for low-latency applications such as real-time identification of abnormal human behaviors in surveillance applications and real-time defect detection in manufacturing processes.&nbsp;</li></ul>EdgeML applications can also benefit from the capacity of cloud computing when required. Specifically, it is possible to collect, aggregate, and process data points from multiple edge devices in the cloud. This enables cloud/edge designs and deployments to leverage many data points toward building, training, and executing accurate data models.<h3>EdgeML security challenges</h3>ML designers are commonly resorting to EdgeML configurations to enhance the privacy and data protection advantages of their applications when compared to the respective cloud applications. This is because edge configurations reduce data transfers, which limits the attack surface of the ML applications. EdgeML applications keep sensitive data at their source, which reduces the number of potential data breaches.Despite these benefits, EdgeML deployments create new security challenges for EdgeML designers, notably challenges associated with the peculiarities of their data sources and edge devices (e.g., IoT devices). This is because edge devices cannot benefit from the security protection functionalities of the cloud data center. Rather they reside at the edge of the network where fewer protective measures are deployed. This renders sensitive information and the devices themselves vulnerable to a variety of security threats. &nbsp;Therefore, EdgeML designers must consider and remedy prominent vulnerabilities of edge ML devices, including:<ul><li>Malware attacks: In EdgeML deployments, there are always opportunities for adversaries to use malicious code to hijack devices.&nbsp;For example, an attacker may maliciously install older firmware with known security vulnerabilities.</li><li>Data theft: &nbsp;Edge devices are vulnerable to breaches that lead to data theft. For instance, cybercriminals can bypass encryption to steal sensitive data.</li><li>IP theft: There are also cases where malicious parties attempt to steal Intellectual Property (IP) such as software code, ML models, and algorithms. One of the best ways to do this is to breach the memory protection of the device.</li><li>Rollbacks to Vulnerable Firmware:&nbsp;Adversaries can also replace a device&rsquo;s firmware with an older version that has known security weaknesses. In this case, they can compromise the device without getting easily noticed.</li><li>Device Cloning: It is also possible to copy the identity code of the device to manufacture illegal copies of it. This can occur in several ways, such as leveraging software vulnerabilities in different systems that share source code or libraries . Illegal copies with the counterfeit identity code then look like legitimate devices. This is another way to compromise an enterprise&rsquo;s IP and reduce its product revenue.</li><li>Network Breaches: Internet-connected devices are also vulnerable to networking attacks. For example, hackers are likely to spoof the device&rsquo;s authentication to gain access and commit hostile actions.&nbsp;</li><li>Attacks against ML models: EdgeML deployments are also vulnerable to attacks that compromise the operation of ML models. For example, hackers can poison the training data to compromise the proper operation of an ML model. Malicious parties can also compromise the operation of deep neural networks by means of adversarial inputs that cannot be correctly classified. This is commonly known as an evasion attack.&nbsp;</li></ul>In a world with billions of IoT devices, these challenges can lead to large-scale problems and security attacks. Back in 2016, <a href="https://en.wikipedia.org/wiki/Mirai_(malware)" target="_blank">the Mirai malware</a> was used to compromise thousands of IoT devices, which formed a botnet and launched Distributed Denial of Service (DDoS) attacks. This had severe consequences for major internet service providers and made popular sites (e.g., Twitter, Amazon, Spotify, Netflix, Reddit) unreachable to some users. As these types of attacks become more frequent, it is important that edge ML designers devise effective solutions to the unique security challenges of EdgeML applications.<h3>Popular security solutions for edge applications</h3>To alleviate the security vulnerabilities of non-trivial EdgeML systems, EdgeML designers can implement various security solutions and controls. These solutions address security at different levels (e.g., operating system, application, network) towards remedying vulnerabilities and threats of different natures. A representative set of such solutions and controls follows:<ul><li>Data Encryption: Data encryption is key to safeguarding data at rest. Effective encryption solutions partition data according to their type (e.g., IoT data, device logs, ML models) and provide support for multiple integrity checks. Moreover, they decrypt data only for authenticated users from a small set of origins and trust domains.&nbsp;</li><li>Distributed Key Management: To strengthen the effectiveness of the encryption mechanisms, there are security solutions that split unique random keys into parts that reside in different locations. The different pieces of the keys are integrated in memory for short amounts of time only, which makes it very hard for adversarial parties to tamper with them and gain access to the encryption and decryption operations.</li><li>Operating System (OS) Security Solutions: OS level solutions ensure that the device runs the most secure version of its operating system. In this direction, IoT and security vendors design and implement secure &ldquo;hardened&rdquo; configurations of the OS. This entails tasks like removing packages that are not needed, creating and configuring encrypted partitions, configuring stringent data access policies for OS functionalities, and embedding advanced monitoring and logging features within the OS.</li><li>Application-Level Security Solutions: These solutions protect ML applications against different forms of tampering. For instance, they implement multi-factor authentication that makes it very difficult for hackers to gain access to ML models and applications. Multi-factor authentication requires users to supply multiple authentication credentials (e.g., passwords, PIN codes), beyond physical access to the device. As another example, application-level security ensures the use of secure components (e.g., libraries) for the application at hand, as well as support for the secure downloading of components (e.g., ML models) to prevent data and IP theft.</li><li>Network-Level Security Solutions: At this level, solutions aim at ensuring secure communications between edge devices and other applications. In this direction, measures like mutual authentication of applications or devices, Transport Level Security (TLS) over Virtual Private networks (VPN), and encryption of sensitive data exchanged over the network are usually implemented. Moreover, special provisions for secure communications that perform firmware updates are made to alleviate malicious changes to the device firmware.</li><li>ML-Level Security Solutions: Solutions at this level safeguard the model training process. In this direction, they can use unique encryption keys to safeguard the security and confidentiality of the device&rsquo;s communication with the backend computer where the training takes place. This is important for alleviating poisoning attacks. Moreover, the parameters of the ML models are encrypted and obfuscated, which renders any interception attempt useless. &nbsp;</li></ul><h3>Putting the solutions to work in practical use cases</h3>A balanced combination of the above-listed multi-level security solutions is required for every practical ML design and deployment. Consider, for example, the design and deployment of ML-based security and surveillance solutions in smart city environments. These solutions leverage edge computing and edge devices (e.g., cameras) to provide real-time detection of events. For instance, camera feeds are typically processed at the edge to identify license plates and abnormal behaviors of vehicles and citizens. To this end, ML-based computer vision algorithms are deployed within the edge devices.Such deployments are susceptible to a variety of attacks. For instance, an adversary could gain access and deploy malware on the cameras to change their functionality. Strong authentication measures and network-level security are therefore required.&nbsp;As another example, hackers could attempt to steal sensitive data, such as license plates of vehicles in a specific crossing or even the ML models used for detecting abnormalities. The risk of such attacks can be minimized based on secure configurations of the device and proper encryption measures. Similarly, network security and OS-level measures can significantly reduce the risk of malicious firmware updates. Furthermore, ML-level measures can avoid changes to the operation of the ML models and eventually to the proper operation of the security and surveillance applications.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1675882894331-Figure-shows-some-results-of-our-end-to-end-license-plate-detection-and-recognition+%281%29.png" style="width: 725px;" class="fr-fic fr-dib">License plate detection requires rigorous security. Image credit:&nbsp;<a href="https://www.researchgate.net/profile/Zain-Masood-4" rel="noopener noreferrer" target="_blank">Zain Masood. </a><h3>Alif semiconductor security solutions&nbsp;</h3>Alif provides world-class, scalable, and integrated security solutions for state-of-the-art EdgeML deployments. The company&rsquo;s solutions alleviate the earlier presented challenges based on a rich set of security features and functionalities:<ul><li>Secure boot against malware attacks: In Alif&rsquo;s EdgeML deployments, all code images are signed prior to their deployment and use in edge devices. Moreover, granular firewalls are configured and used to protect assets like the device&rsquo;s memory and Central Processing Units (CPUs). &nbsp;Also, the entire security process employs granular controls for each asset in isolation.</li><li>Data encryption to prevent data theft: Alif&rsquo;s provides strong encryption for both data within the device and data transmitted over the device&rsquo;s network. Hence, it is pointless for a hacker to access the data without the required private key. Alif provides TLS security and uses Public Key Infrastructure (PKI) certificates to protect the payloads of sensitive data.</li><li>Code encryption to prevent IP theft: Similar to the data theft case, in Alif&rsquo;s solutions IP theft is avoided thanks to code encryption. The company&rsquo;s security solutions provide proper in-memory interfaces in support of code decryption in real time. Hence, the solution does not impose any performance penalty on the EdgeML application performance.&nbsp;</li><li>Secure network communications:&nbsp;Alif minimizes the risk of network breaches based on support for bi-directional cloud-network authentication. The authentication process leverage certificates verified with a proper Certificate Authority. Moreover, signatures of trusted servers are stored to ensure that all network communications originate from and terminate at trusted computers and devices.</li><li>Hardware Unique Key (HUK) prevents device Cloning: Alif programs all devices based on a HUK. The HUK is generated at the factory and its authenticity is verified by a special security subsystem of the device. This makes it impossible to clone the device.</li><li>Authenticated and signed software versions: Alif&rsquo;s security solutions do not permit the downgrading of a device&rsquo;s firmware. They only allow the loading of equal or later versions of the firmware. In this direction, Alif provides mechanisms for authenticating (&ldquo;signed&rdquo;) software versions by means of a proper signature.</li></ul>The above-listed measures enable EdgeML designers to implement a holistic, end-to-end approach to safeguarding the operation of large-scale EdgeML applications, even in the light of stringent security requirements.For more information about Alif&rsquo;s products and services for state-of-the-art cloud/edge computing deployments, visit <a href="https://alifsemi.com/ensemble/" rel="noopener noreferrer" target="_blank">Alif&#39;s website.</a>This is the third article in a&nbsp;<a href="https://www.wevolver.com/article/iot-needs-trust-the-current-state-of-iot-security" rel="noopener noreferrer" target="_blank">three-part series</a>&nbsp;examining IoT security problems and solutions as of today.&nbsp;<a href="https://www.wevolver.com/article/iot-needs-trust-the-current-state-of-iot-security" rel="noopener noreferrer" target="_blank">The first article</a> provided an overview of the current state of IoT security.&nbsp;<a href="https://www.wevolver.com/article/how-cutting-edge-microcontrollers-build-security-from-the-ground-up" rel="noopener noreferrer" target="_blank">The second article</a> introduced Alif&#39;s latest product.&nbsp;<h2>About the sponsor: Alif Semiconductor</h2>Alif Semiconductor was founded in 2019 as a result of our co-founders&#39; longtime observations of a rapidly growing market need, and an inadequate response by suppliers, for broad, scalable, connected, AI-enabled embedded computing solutions that are genuinely power efficient. That&rsquo;s why we created a new class of products in the industry, that we call fusion processors, to enable our customers to do much more through innovative low-power techniques, extreme integration, accelerated AI/ML, high security, ubiquitous wireless connectivity, and operating system diversity.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjc1ODgxOTU1MjEyLTE2NzU4ODE5NTUyMTIucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" class="fr-fic fr-dii"><div id="_com_5" language="JavaScript"> </div>

With more IoT devices performing mission-critical tasks - security is essential. To achieve the potential of smart devices that implement ML and AI in their applications, we must protect the critical intellectual property that represents the majority of the solution’s value. 

Security Guide for Edge ML Designers

Alif Semiconductor’s™ microcontrollers meet the market need for low-power, AI-enabled embedded solutions with unmatched levels of multi-layered security, and a fully integrated secure element.

How cutting-edge microcontrollers build security from the ground up

Our love affair with smart devices has led to an explosive growth in the number of devices we surround ourselves with. But whereas they allow us to keep an eye on our homes when we are away or help us track the expiry date of items stored in our fridges, they also leave us vulnerable to intruders, who can use them to hack their way into our homes.For instance, if we choose weak passwords for our internet-connected smart devices or fail to update their security systems, we risk creating access points for hackers.Once a malicious mind has entered your network via devices such as security cameras or fridges, they can try to exploit you by taking charge of your computer and encrypting files until you pay a ransom to have them decrypted. Or they could use your network as a gateway to harm others by orchestrating attacks from hundreds of thousands of hacked devices simultaneously.<h2>Identifying digital ghost ships</h2>In a new project, DTU Associate Professor Emmanouil Vasilomanolakis aims to develop a method that can detect such &lsquo;digital ghost ships&rsquo;&mdash;devices in the sea of smart electronics that have been neglected and pose a security threat. As he explains: Knowing where they are is crucial to alerting the owner so they can either make them safe to use or disable them.&ldquo;We believe that digital ghost ships are a real security threat,&rdquo; he says.Emmanouil Vasilomanolakis coined the term digital ghost ship, which references the ghost ships of the seas&mdash;vessels that have no crew on board to safely steer them.It&rsquo;s not just in people&rsquo;s homes that gadgets with poorly maintained security features pose a problem. In fact, the stakes are often much higher for businesses and organizations that use and rely on smart devices if an intruder gets through and creates havoc.Emmanouil Vasilomanolakis points out that the healthcare industry is a good example: &ldquo;Hospitals use more and more devices that need internet connectivity. If these devices are hacked and stop functioning, we may have a life and death situation.&rdquo;He explains that even cheap devices such as surveillance cameras for your home that can&rsquo;t do much can be powerful tools for hackers &ndash; especially if they gain access to a large number of devices at the same time and use them to stage an attack on another target:&ldquo;If you can access only one device, it&rsquo;s not a very powerful attack. But of course, if you can use one million devices, that creates a serious security threat.&rdquo;Such attacks can be used to e.g., force authorities&rsquo; websites offline, as was seen when Chinese hackers &nbsp;managed to temporarily force Taiwanese government websites offline during the visit of US speaker of the House of Representatives, Nancy Pelosi, to Taiwan in August. Hackers can also use it to cause significant disruption to commercial sites, effectively blocking actual customers from purchasing goods for periods of time.<h2>A more finely meshed safety net</h2>Commercial services are already available that allow users to scan the internet and identify internet-connected devices. Emmanouil Vasilomanolakis aims to create a much more finely meshed safety net that scans and detects only actual digital ghost ships while omitting properly maintained gadgets.The system will also be trained to avoid so-called honeypots and other false positives. A honeypot is a detection system that developers create to attract attackers to a secure system to study their behaviour.The researchers will investigate novel ways of creating network signatures of digital ghost ships. A network signature is a footprint that has been left following unauthorized access. The aim is to enrich these signatures with device fingerprinting capabilities. Collecting such fingerprints provides information about the software and hardware of the device in question, making it easier to identify its type.DTU will collaborate with the University of Cambridge for this part of the project.

Researchers at DTU have set out to create novel tools to identify internet-of-things gadgets that leave the door open to hacker attacks. Knowing where these devices are and making them secure is a powerful weapon in the fight against attackers.

Setting out to sink the internet's digital ghost ships

Most of us are aware that large tech companies such as Google, Facebook, Apple, Microsoft, Twitter, Netflix and Spotify have massive amounts of data about our online behaviour, so they almost know us better than we do. But who else has access to your personal information? It&rsquo;s almost impossible to know, and a new EU-supported project in which DTU is involved wants to change that.&ldquo;The problem is that our data is everywhere, and it is difficult to find out which companies have access to it and have sold it on. We want to make it all more transparent,&quot; says Weizhi Meng, associate professor in cyber security at DTU Compute.DataVaults can be seen as a digital safe where you can store your personal data and control who has access and when. You are notified when companies, authorities or others access your information, and the platform can assess the risks associated with sharing your information in various contexts.Personal data such as your age, gender and address may seem entirely innocent, but if all those details fall into the wrong hands, it can have serious consequences.&quot;In a digitalized world, personal data is the most important asset. If your data has been leaked, hackers can use it to create a false identity of you. They can potentially create fake passports or credit cards in your name or send you targeted phishing emails based on what they already know. That is why it is so important to protect our personal data,&rdquo; says Weizhi Meng.<h2>Businesses&nbsp;are still challenged by GDPR</h2>It&rsquo;s impossible to discuss data privacy without mentioning GDPR. The EU&#39;s legislation on the protection of personal data shook the IT world when it was introduced in 2018 and is still a cause of confusion to many companies. DataVaults can help not only individuals but also companies that can use the platform to collect information in a secure and legal manner about everything from employees to customers and business partners.&quot;It&rsquo;s a big challenge for many companies to comply with GDPR. They don&#39;t know how to implement the GDPR guidelines because it&#39;s very complicated,&rdquo; says Weizhi Meng.A survey by the Council for Digital Security has shown that almost half of all small and medium-sized companies are challenged by the legal requirements for personal data protection, and 67% believe it is difficult to assess how data can be used ethically.&quot;There is no doubt that it has been an uphill battle for the vast majority of companies,&quot; says Henning Mortensen, chairman of the Council for Digital Security.Most recently, the so-called Schrems II ruling in the EU has created uncertainty. In practice, it makes it problematic for businesses to use American cloud services and platforms, such as Google Analytics, because they may transfer personal data to their parent company in the US which is against GDPR regulations.&quot;It&rsquo;s a huge problem for all companies and also authorities, which can make GDPR regulations a major barrier for them. At best, it is unclear what to do when using international services, especially cloud services, and at worst, they cannot use them at all,&quot; says Henning Mortensen.That&rsquo;s why he welcomes an initiative like DataVaults.&quot;There is clearly a need for solutions that can collect data in a way that protects our personal data. It could help create a greater willingness to make one&#39;s data available for public benefit,&quot; says Henning Mortensen.&nbsp;<h2 id="isPasted">Privacy-friendly big data</h2>In essence, DataVaults wants to build up our trust so we can confidently share data without fear of it being misused. There is great potential in big data, where large data sets are used to see new connections, but it must be done responsibly. One of the project&#39;s partners is the French healthcare platform Andaman7, which uses DataVaults to collect data on patients that they use for clinical research.&quot;If you want to look at how many people suffer from a certain disease, we can collect the results but anonymize the individuals,&quot; says Weizhi Meng.Similarly, a Spanish solar cell manufacturer participates in the project and uses DataVaults to obtain data on users&#39; electricity consumption to predict the demand for electricity at different times of the day.But DataVaults also wants to give back the financial incentives to us. Whereas Google and Facebook bring in billions of ad dollars from harvesting our data, users of DataVaults can be rewarded for sharing their data. It could be monetary, but also as the Italian municipality of Prato is doing by rewarding residents who share their cultural preferences and habits with tickets to a performance for instance.However, DataVaults cannot do anything about the information we have already freely given away.&quot;We can protect the data in the DataVaults, but if Google and Facebook already have it, then there is not much we can do,&quot; says Weizhi Meng.<h2>Future solution</h2>After three years, the DataVaults project is coming to an end, and the platform&rsquo;s beta version is now ready. Long term the hope is that the EU or another authority will be interested in taking ownership of the platform, as it requires consumer trust and resources to handle the personal data of potentially millions of people.&ldquo;The more people using DataVaults, the better it will be. When it comes to big data, it obviously works best with as many users as possible,&quot; says Weizhi Meng.In the future, the pressure on our personal data will only increase, and the DTU associate professor believes that, although GDPR is a good start, there is a need to continue to improve the rules and solutions within the protection of personal data.&ldquo;It will only become more important in the future. Suppose we enter the era of the metaverse (virtual worlds in 3D), then personal data becomes the top priority. When everything becomes digitalized, those who have data can control everything,&rdquo; says Weizhi Meng.

We need better control over our personal data if we are to keep hackers and commercial interests from knowing our every move. DTU plays an important role in a large EU-funded research project developing a digital vault that lets you manage who has access to your personal data.

Taking back privacy control

Trustworthy and reliable applications are fueling the worldwide IoT success story. This article explains why protecting IoT edge devices is crucial in an environment with threats on the rise.  


IoT needs trust: The current state of IoT security

You&rsquo;ve probably heard or read about the amazing capabilities of some recent AI models, specifically large-scale language models like <a href="https://openai.com/api/" rel="noopener" target="_blank">GPT</a> (GPT-3 at the time of this writing), <a href="https://www.ai21.com/" rel="noopener" target="_blank">AI21</a>, or <a href="https://huggingface.co/docs/transformers/model_doc/bloom" rel="noopener" target="_blank">BLOOM</a>. Perhaps you&rsquo;ve tried one of these models yourself, either directly or as part of another product. For example, a writing assistant like <a href="https://www.wordtune.com/" rel="noopener" target="_blank">Wordtune</a>, <a href="https://blog.you.com/introducing-youwrite-an-ai-writing-assistant-on-you-com-a9e05080c8ea" rel="noopener" target="_blank">YouWrite</a>, or <a href="https://www.jasper.ai/" rel="noopener" target="_blank">Jasper</a>. Or, more recently, <a href="https://chat.openai.com/chat" rel="noopener" target="_blank">ChatGPT</a>.As their name suggests, large-scale language models (LLMs) were developed and trained with language data. But it turns out that they can be applied to other use cases, outside language. <a href="https://github.com/features/copilot" rel="noopener" target="_blank">GitHub Copilot</a> is probably one of the best-known examples of this. It&rsquo;s a tool that helps you write software. You describe to the model what you want to do, and it suggests code.It&rsquo;s also no big secret anymore that these models are started to get used for <a href="https://mergeflow.net/report?key=16b87a34-a80f-9dc3-0626-407bae0f2938" rel="noopener" target="_blank">drug discovery</a>:<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjcxMTg4NTY5NjA5LWltYWdlLTEucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" style="width: 300px;" class="fr-fic fr-dib"><figure id="isPasted"><figcaption>Mergeflow data on large language models (LLMs) used for drug discovery. <a href="https://mergeflow.net/report?key=16b87a34-a80f-9dc3-0626-407bae0f2938" rel="noopener" target="_blank">Click here</a> to see the report.</figcaption></figure>Probably because of these applications outside language, more and more people refer to these models as <a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=gpt-3*%7cgpt3*%7cgpt-j%7cgpt-neo%7cai21%7celeutherai%7cdall-e%7cinstructgpt*%7cwebgpt*%7c%22language+model%22%7c%22language+models%22%7copt-175b%7c%22open+pretrained+transformer%22%7calexatm%7c%22alexa+teacher+model%22%7c%22stable+diffusion%22%7c%22foundation+models%22%7ctext-davinci*%7calphacode%7c%22github+copilot%22%7ccodewhisperer&q=-%22more+explainable+AI%22&fid=10291" rel="noopener" target="_blank">foundation models</a>, rather than language models: They are the foundation for many use cases, not just language-related ones.So, what are some other non-language examples where LLMs, or foundation models, are currently being explored? I used Mergeflow to find out.You&rsquo;ll notice that all my examples below are research papers. This might be an indicator of how nascent the field of &ldquo;doing things with foundation models&rdquo; is. But even if you&rsquo;re a &ldquo;product person&rdquo;, or a &ldquo;company person&rdquo;, I encourage you to take a look: the time it takes to get from research paper to product can be really short when it comes to LLMs.<h2 id="isPasted">Materials discovery</h2>Materials discovery is the process of identifying new materials and understanding their properties. It can help inform decisions about which materials are best suited for particular applications. And it helps accelerate the development process by providing insights into how materials behave under different conditions.I used Mergeflow to research <a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=%22Computational+Materials+Discovery%22&q=%22Foundation+Models%22" rel="noopener" target="_blank">foundation models for materials discovery, or computational chemistry</a>, and then zoomed in on research papers that didn&rsquo;t deal with language. For example, I excluded papers on materials-related information extraction from text.This paper, for instance&hellip;<a href="https://arxiv.org/abs/2206.13578" rel="noopener" target="_blank">Materials Transformers Language Models for Generative Materials Design: a benchmark study</a>&hellip;used data from material databases (<a href="https://icsd.products.fiz-karlsruhe.de/" rel="noopener" target="_blank">ICSD</a>, <a href="https://oqmd.org/" rel="noopener" target="_blank">OQMD</a>, and <a href="https://materialsproject.org/" rel="noopener" target="_blank">Materials Project</a>) to predict new materials. The authors verified the materials using <a href="https://mergeflow.net/profiles/fw/search/summary?q=%22DFT+calculations%22%7C%22DFT+calculation%22&date=default3" rel="noopener" target="_blank">DFT calculations</a>. DFT (Density Functional Theory) calculations are used to determine the energy and electronic properties of a molecule (I&rsquo;m not a materials scientist but I assume that this is basically like doing a plausibility check for a hypothesized material).And this paper&hellip;<a href="https://arxiv.org/abs/2211.15895" rel="noopener" target="_blank">Composition based oxidation state prediction of materials using deep learning</a>&hellip;uses LLMs to predict <a href="https://mergeflow.net/profiles/fw/search/summary?q=%22oxidation+state%22%7C%22oxidation+states%22&date=default3" rel="noopener" target="_blank">oxidation states</a> of inorganic materials. The oxidation state of an atom is a measure of its electron-accepting or electron-donating ability. Atoms with a positive oxidation state are electron-donating, while atoms with a negative oxidation state are electron-accepting. According to Mergeflow&rsquo;s data and analytics, oxidation states play a role in battery and fuel cell technology, and in catalyst design:<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjcxMTg4NjU4ODMxLWltYWdlLTIucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" style="width: 300px;" class="fr-fic fr-dib"><figure id="isPasted"><figcaption><a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=%22oxidation+state%22%7c%22oxidation+states%22" rel="noopener" target="_blank">Oxidation states</a> play a role in battery or fuel cell technology, as well as for&nbsp;<a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=%22oxidation+state%22%7C%22oxidation+states%22&q=catalys%2A" rel="noopener" target="_blank">catalyst design</a>. Screenshot from Mergeflow.</figcaption></figure> <h2 id="isPasted">Cybersecurity</h2>In cybersecurity, an <a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=%22attack+packet%22%7C%22attack+packets%22" rel="noopener" target="_blank">attack packet</a> is a piece of data that is used to attack a computer or network. Attack packets can be used to perform a variety of attacks, including <a href="https://mergeflow.net/profiles/fw/search/summary?q=%22denial+of+service%22%7Cdenial-of-service%2A&date=default3" rel="noopener" target="_blank">denial of service</a>, <a href="https://mergeflow.net/profiles/fw/search/summary?q=man-in-the-middle%2A&date=default3" rel="noopener" target="_blank">man-in-the-middle</a>, <a href="https://mergeflow.net/profiles/fw/search/summary?q=phishing%2A&date=default3" rel="noopener" target="_blank">phishing</a>, <a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=cyber%2A&q=%22buffer+overflow%22%7Cbuffer-overflow%2A" rel="noopener" target="_blank">buffer overflow</a>, <a href="https://mergeflow.net/profiles/fw/search/summary?q=%22SQL+injection%22&date=default3" rel="noopener" target="_blank">SQL injection</a>, and <a href="https://mergeflow.net/profiles/fw/search/summary?q=spoofing%2A&date=default3" rel="noopener" target="_blank">spoofing</a>.Imagine you&rsquo;re in charge of the security of a system, and you see attack packets hitting your system. And based on what kinds of attack packets you see, you have to infer what kind of attack you might be dealing with.When you do this, chances are you&rsquo;ll see lots of such packets every day. For example, the authors of this paper&hellip;<a href="https://arxiv.org/abs/2209.00263" rel="noopener" target="_blank">Attack Tactic Identification by Transfer Learning of Language Model</a>&hellip;used data from a <a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=honeypot&q=-A23%2A%5Bipc%5D%7Cinsect%2A%7Cdisease%7Cfood" rel="noopener" target="_blank">honeypot</a> (= a system that aims to attract attackers and learn more about their tactics and tools) with 227,000 packets per day. Obviously, you need some kind of analytics to help you interpret such volumes of data.And this is where foundation models come in: They used LLMs to learn attack patterns from (text) descriptions of known attacks, and then used the model to classify previously-unseen attack packets. (so this is a &ldquo;borderline case&rdquo; &mdash; yes, the model uses language to train but then is used to classify data that&rsquo;s not language, i.e. the attack packets).<h2 id="isPasted">Building management</h2><a href="https://mergeflow.net/profiles/fw/search/summary?date=default3&q=%22building+management%22&q=-%22building%2C+management+and+communication%22" rel="noopener" target="_blank">Building management</a> refers to the process of overseeing the day-to-day operations of a building. It involves managing the maintenance of the building systems, such as heating, ventilation and air conditioning (HVAC), electrical systems, plumbing, and fire suppression systems. It also includes other aspects, such as tenant lease management, but the focus here is on the infrastructure aspects.Building management is quite an active area of technology, as you can see in the <a href="https://mergeflow.net/report?key=605a0240-4733-54b2-6afe-e479731e41fd" rel="noopener" target="_blank">report from Mergeflow</a> below (click on the screenshot below to get to the report):<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjcxMTg4NzE2NTQxLWltYWdlLTMucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" style="width: 300px;" class="fr-fic fr-dib"><figure id="isPasted"><figcaption>A&nbsp;<a href="https://mergeflow.net/report?key=605a0240-4733-54b2-6afe-e479731e41fd" rel="noopener" target="_blank">&ldquo;building management&rdquo;</a> report from Mergeflow.</figcaption></figure>Perhaps a bit similar to the cybersecurity example above, building management systems produce large amounts, or streams, of data. The challenge then is to interpret the data. For example, given data streams from all kinds of building systems, how do you know where you could optimize energy consumption?The problem is, you need labeled data. In other words, you need to know what certain kinds of data mean (e.g. what do certain combinations of weather and other sensor data mean with respect to energy consumption?). And in order to substantially reduce the amount of hand-labeled data needed for a good model, this paper&hellip;<a href="https://www.mdpi.com/1996-1073/15/23/8894" rel="noopener" target="_blank">Semantic-Similarity-Based Schema Matching for Management of Building Energy Data</a>&hellip;uses a combination of foundation models and&nbsp;<a href="https://mergeflow.net/profiles/fw/search/summary?q=%22active+learning%22&q=%22machine+learning%22&date=default3" rel="noopener" target="_blank">active learning</a>. Active learning is a machine learning technique that focuses on selecting the best samples to query from a dataset, in order to get the most accurate answer with the fewest queries. It involves selecting a small subset of data points and querying them for their values, then using the results to choose the next set of samples to query. This allows the machine learning algorithm to quickly home in on the correct answer without having to query all of the data points initially.

Language models outside language. Image generated by Stable Diffusion.

You've probably heard about the amazing capabilities of some recent AI models, such as GPT, AI21, or BLOOM. Perhaps you use one of these models yourself. Either directly, or through another product like Wordtune, YouWrite, Jasper – or ChatGPT.

It turns out that while these models are trained on language data, they can be used for other applications as well.

In my latest article, you can read more about applications in materials discovery, cybersecurity, and even building management.

Applying large-scale language models outside language: Examples from materials discovery, cybersecurity, and building management

A major vulnerability in a networking technology widely used in critical infrastructures such as spacecraft, aircraft, energy generation systems and industrial control systems was exposed by researchers at the University of Michigan and NASA.<iframe src="https://www.youtube.com/embed/tw1QLVQw8Go?&wmode=opaque&rel=0&enablejsapi=1&origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" style="width: 767px; height: 432px;"></iframe>It goes after a network protocol and hardware system called time-triggered ethernet, or TTE, which greatly reduces costs in high-risk settings by allowing mission-critical devices (like flight controls and life support systems) and less important devices (like passenger WiFi or data collection) to coexist on the same network hardware. This blend of devices on a single network arose as part of a push by many industries to reduce network costs and boost efficiency.That coexistence has been considered safe for more than a decade, predicated on a design that prevented the two types of network traffic from interfering with one another. The team&rsquo;s attack, called PCspooF, was the first of its kind to break this isolation.In one compelling demonstration, the team used real NASA hardware to recreate a planned Asteroid Redirection Test. The experimental setup controlled a simulated crewed capsule, specifically at the point in the mission when the capsule prepared to dock with a robotic spacecraft.&ldquo;We wanted to determine what the impact would be in a real system,&rdquo; said <a href="https://web.eecs.umich.edu/~barisk/" target="_blank">Baris Kasikci</a>, the Morris Wellman Faculty Development Assistant Professor of Computer Science and Engineering. &ldquo;If someone executed this attack in a real spaceflight mission, what would the damage be?&rdquo;With one small malicious device, the team was able to seamlessly introduce disruptive messages to the system, creating a cascading effect that ended in the capsule veering off course and missing its dock entirely.Here&rsquo;s how it works: The attack emulates the network switches, which are high-stakes traffic controllers in TTE networks, by sending out fake synchronization messages. These messages are normally intended to keep network devices running on a shared schedule, allowing the most important devices to communicate quickly.&ldquo;Normally, no device besides a network switch is allowed to send this message, so in order to get the switch to forward our malicious message, we conducted electromagnetic interference into it over an Ethernet cable,&rdquo; said Andrew Loveless, U-M doctoral student in computer science and engineering and subject matter expert at the NASA Johnson Space Center.That interference serves as an envelope for the fake synchronization message. The noise causes just enough of a gap in the switch&rsquo;s normal operation to allow the message to pass through. An easily concealed bit of circuitry on a malicious device, connected to the network via Ethernet, can inject these messages as many times as necessary to throw everything out of whack.&ldquo;Once the attack is underway, the TTE devices will start sporadically losing synchronization and reconnecting repeatedly,&rdquo; Loveless said.This disruption will gradually lead to time-sensitive messages being dropped or delayed, causing systems to operate unpredictably and, at times, catastrophically. But the researchers explain how to prevent this attack, too.Replacing copper Ethernet with fiber optic cables or installing optical isolators between switches and untrusted devices would eliminate the risk of electromagnetic interference, though this would come with cost and performance tradeoffs. Other options involve changes to the network layout, so that malicious synchronization messages can never access the same path taken by the legitimate ones.&ldquo;Some of these mitigations could be implemented very quickly and cheaply,&rdquo; Kasikci said.The team disclosed their findings and proposed mitigations to major companies and organizations using TTE and to device manufacturers in 2021.&ldquo;Everyone has been highly receptive about adopting mitigations,&rdquo; Loveless said. &ldquo;To our knowledge, there is not a current threat to anyone&rsquo;s safety because of this attack. We have been very encouraged by the response we have seen from industry and government.&rdquo;

A new attack discovered by the University of Michigan and NASA exploits a trusted network technology to create unexpected and potentially catastrophic behavior