ECE Colloquium

ECE Colloquium

Dr. Shlomo Shamai

Distinguished Professor

The Andrew and Erna Viterbi Department of Electrical Engineering

Technion-Israel Institute of Technology

 

TITLE:    A View of Information-Estimation Relations in Gaussian Networks

Defeating Cyberattacks on 3D Printers

Researchers at Rutgers and Georgia Tech develop three methods to defend against sneaky attacks

With cyberattacks on 3D printers likely to threaten health and safety, researchers at Rutgers University-New Brunswick and Georgia Institute of Technology have developed novel methods to combat them, according to a groundbreaking study.

“They will be attractive targets because 3D-printed objects and parts are used in critical infrastructures around the world, and cyberattacks may cause failures in health care, transportation, robotics, aviation and space,” said Saman Aliari Zonouz, an associate professor in the Department of Electrical and Computer Engineering at Rutgers University-New Brunswick.

He co-authored a peer-reviewed study – “See No Evil, Hear No Evil, Feel No Evil, Print No Evil? Malicious Fill Pattern Detection in Additive Manufacturing” – that was published today at the 26th USENIX Security Symposium in Vancouver, Canada. It’s the security community’s flagship event, highlighting the latest advances in protecting computer systems and networks. Among several unique techniques, the Rutgers and Georgia Tech researchers are using cancer imaging techniques to detect intrusions and hacking of 3D printer controllers. 

“Imagine outsourcing the manufacturing of an object to a 3D printing facility and you have no access to their printers and no way of verifying whether small defects, invisible to the naked eye, have been inserted into your object,” said Mehdi Javanmard, study co-author and assistant professor in the Department of Electrical and Computer Engineering at Rutgers University-New Brunswick. “The results could be devastating and you would have no way of tracing where the problem came from.”

3D printing, also called additive manufacturing, plays an increasingly important role in industrial manufacturing. But health- and safety-related products such as medical prostheses and aerospace and auto parts are being printed with no standard way to verify them for accuracy, the study says. Even houses and buildings are being manufactured by 3D printers, noted Javanmard.

Instead of spending up to $100,000 or more to buy a 3D printer, many companies and organizations send software-designed products to outside facilities for printing, Zonouz said. But the firmware in printers may be hacked.

For their study, the researchers bought several 3D printers and showed that it’s possible to hack into a computer’s firmware and print defective objects. The defects were undetectable on the outside but the objects had holes or fractures inside them.

Other researchers have shown in a YouTube video how hacking can lead to a defective propeller in a drone, causing it to crash, Zonouz noted.

While anti-hacking software is essential, it’s never 100 percent safe against cyberattacks. So the Rutgers and Georgia Tech researchers looked at the physical aspects of 3D printers.

In 3D printing, the software controls the printer, which fulfills the virtual design of an object. The physical part includes an extruder or “arm” through which filament (plastic, metal wire or other material) is pushed to form an object.

The researchers observed the motion of the extruder, using sensors, and monitored sounds made by the printer via microphones.

“Just looking at the noise and the extruder’s motion, we can figure out if the print process is following the design or a malicious defect is being introduced,” Zonouz said.

A third method they developed is examining an object to see if it was printed correctly. Tiny gold nanoparticles, acting as contrast agents, are injected into the filament and sent with the 3D print design to the printing facility. Once the object is printed and shipped back, high-tech scanning reveals whether the nanoparticles – a few microns in diameter – have shifted in the object or have holes or other defects.

“This idea is kind of similar to the way contrast agents or dyes are used for more accurate imaging of tumors as we see in MRIs or CT scans,” Javanmard said.

The next steps in their research include investigating other possible ways to attack 3D printers, proposing defenses and transferring methods to industry, Zonouz said.

“You’ll see more types of attacks as well as proposed defenses in the 3D printing industry within about five years,” he said.

Study coauthors include Christian Bayens and Raheem Beyah of the Georgia Tech, and Tuan Le and Luis Garcia of Rutgers.

Story by Todd Bates for Rutgers Today

Photo: Christian Bayens/Georgia Institute of Technology

WINLAB Team Wins prize in Ettus-Xilinx Software Defined Radio Challenge

A team from WINLAB comprising ECE graduate student Bhargav Gokalgandhi, researcher Prasanthi Maddala and Associate Director Ivan Seskar has won the 3rd prize at the RF Network on Chip (RFNoC) and Vivaldo Challenge. This challenge sponsored by Ettus Research and Xilinx Inc. rewards engineers for creating innovative open-source RF Network on Chip (RFNoC) blocks that will add to the library of available open-source blocks for programming FPGAs in Software Defined Radio development and production. The team designed a wide band channel sounder that computes the power delay profile of a multipath channel in a massive multiple antenna system.
Please find more details about the competition here

Congratulations Bhargav, Prasanthi and Ivan!
 

Rutgers Team receives NSF EAGER Grant

A team from Rutgers comprising ECE Professors Narayan Mandayam and Janne Lindqvist along with Professor Arnold Glass from Psychology has received an Early Concept Grant for Exploratory Research (EAGER) award from the NSF for a project titled "Simulated and Synthetic Data for Interdependent Communications and Energy Critical Infrastructures." This is a 2 year $200,000 project in collaboration with Florida International University (FIU), with Rutgers' share of the award being $100,000.

As part of this exploratory project, the Rutgers and FIU team will develop new mathematical foundations and computer-based learning theories for generating a wide range of simulated data sets (obtained via down sampling, aggregation of actual data) and fully-synthetic data sets (obtained via emulation and human subject studies) that model interdependence between communications and energy infrastructures. Specifically, the Rutgers team will develop synthetic data for communications critical infrastructure using the ORBIT Testbed as well as human subject studies for prosumer participation in the smart grid under emergencies. The synthetic data generated will be linked to the simulated power system data generated at FIU using transfer learning techniques. These enhanced data sets and associated data building tools will provide large-scale test data related to interdependent critical infrastructures operation.

Congratulations Narayan, Janne and Arnold!

Dario Pompili receives NSF Grant

Associate Professor Dario Pompili has been awarded a new NSF award for the project titled "Robust, scalable, distributed semantic mapping for search-and-rescue and manufacturing co-robots." This is a three-year $850,000 collaborative effort between Boston University and Rutgers University. Rutgers' share of this award is $426,161.
 
The goal of this project is to enable multiple co-robots to map and understand the environment they are in to collaborate among themselves and with human operators in education, medical assistance, agriculture, and manufacturing applications. The first characteristic of this project is that the environment is modeled semantically, that is, it contains human-interpretable labels (e.g., object category names) in addition to geometric data. This is achieved through a novel, robust integration of methods from both computer vision and robotics, and allows easier communications between robots and humans in the field. The second characteristic is that the increased computation load due to the addition of human-interpretable information is handled by judiciously approximating and spreading the computations across the entire network. 
 
Specifically, as part of this project, Dario and his team will propose a new optimization framework for semantic mapping that can handle large, dynamic, uncertain environments under significant measurement errors, studying interactions and information exchanges with humans, and allowing an intelligent sharing of the limited computational resources possessed by the network of co-robots as a whole by enabling approximations and balancing of the computations. The novel developed methods will be evaluated by emulating real-world scenarios in manufacturing and for search-and-rescue operations.
 
You can find more details on the project at the NSF page here:
 
Congratulations Dario!

Kristin Dana receives NSF Grant

Professor Kristin Dana has been awarded a new NSF award for the project titled "Seeing Surfaces: Actionable Surface Properties from Vision." This is a three year $500,000 collaborative effort between Rutgers University (Kristin Dana, PI) and Drexel University. Rutgers' share of this award is $249,931.00.

As part of this project, Kristin and her team will develop models and algorithms for estimating actionable, physical properties of surfaces from their appearance for applications in scene understanding, robotic action planning, and efficient visual sensing. The research will address the fundamental question of how computer vision can anticipate the physical properties of a surface, laying the foundation for computational vision-for-action. The research activities are centered on four specific aims: 1)  large-scale data collection of actionable physical properties and appearance measurements of everyday surfaces, 2) derivation of prediction models for deducing physical properties from local surface appearance, 3) integration of global semantic context including object and scene information, and 4) development of efficient appearance-capture optics and hardware for use in novel physics-from-appearance sensing. 

You can find more details on the project at the NSF page here.

Congratulations Kristin!

Roy Yates receives NSF Grant

Distinguished Professor Roy Yates has won a new NSF award for the project titled "Timely Updating: Principles and Applications." This is a three year $500,000 project.

As part of this groundbreaking project, Roy will study the foundations of timely updating of information. With the emergence of cyber-physical systems, real-time status updates have become an important and ubiquitous form of communication. Applications that employ vehicular status messages, security reports from computers, homes, and offices, and surveillance video from remote-controlled systems need status updates to be as timely as possible; however, this is typically constrained by limited network resources.This project will examine these real-time status updating systems using Age-of-Information metrics. 
 
You can find more details on the project at the NSF page here.
 
Congratulations Roy!
 

Emina Soljanin receives NSF Grant

Professor Emina Soljanin has received a new NSF award for the project titled "Codes for Data Storage with Queues for Data Access." This is a three year $500,000 collaborative effort between Rutgers University (Emina Soljanin, PI) and Texas A&M University. Rutgers' share of this award is $309,236.

As part of this project, Emina and her team will develop efficient algorithms for distributed storage and access of large files. Large volumes of data, which are being collected for the purpose of knowledge extraction, have to be reliably, efficiently, and securely stored. Retrieval of large data files from storage has to be fast (and often anonymous and private). Large-scale cloud data storage and distributed file systems, e.g., Amazon EBS and Google FS, have become the backbone of many applications such as web searching, e-commerce, and cluster computing. Cloud services are implemented on top of a distributed storage layer that acts as a middleware to the applications, and also provides the desired content to the users, whose interests range from performing data analytics to watching movies. This project focuses on efficient data access in distributed file systems that employ erasure codes for reliable and efficient storage.

You can find more details on the project at the NSF page at http://"https://www.nsf.gov/awardsearch/showAward…".

Pages

Subscribe to Rutgers University, Electrical & Computer Engineering RSS