Hello from FedInvent,
On Tuesday, while we were digging through the latest round of taxpayer-funded patents, we watched the horrific shooting on a subway car in Brooklyn unfold. Being stuck in a subway car with an armed crazy is any New Yorker's worst nightmare.
The shooter has been apprehended and charged with one federal terrorism charge. Federal and city prosecutors now need to assemble their evidence and prepare to go to court.
The New York Police Department (NYPD) asked the public to upload any video that might be helpful to their investigation. Unfortunately, the cameras in the station weren't working. Video from the public was essential to the investigation. The news media scoured social media looking for videos of the incident. The media also aired conversations with eyewitnesses.
The videos taken from inside the train and on the platform provide important timeline evidence of what happened. In addition, these videos have pictures of potential witnesses. Pictures of the faces of the people in the videos may be processed through facial recognition software to help identify the shooter and find people who can both provide testimony and help fill in the blanks on what took place on the train.
The NYPD and the FBI sent their agents out to collect videos from the city's cameras, the stores in the area, and surveillance cameras in other stations so that the police could figure out when and where the shooter entered the subways system and how he got away. Once the police found a key to a U-Haul van at the crime scene, investigators probably started looking at license plate readers, E.Z. Pass data, and the cameras on the roads into and around Brooklyn to locate the van. This case will generate a tsunami of digital evidence. The challenge will be to weave it together so it that can be used in court to tell the story of what happened on Tuesday.
One of the compelling issues is how to validate the authenticity of digital evidence. How do you know the images haven't been altered? How do you authenticate YouTube videos? How do you create a digital timeline? Who are the people affected by the crime?
There is a significant portfolio of federal research and development, and new inventions focused on how to authenticate digital assets and, by extension, digital evidence. Here is a sample of some technologies useful in examining digital evidence paid for by taxpayer dollars.
Forensic Examination of Video and Image Files
Researchers at IBM invented technology for image-stitching and summarizing aerial views. The stitching technology enables the creation of a panoramic view from videos of multiple cameras. This stitching capability is critical for many analytic applications, including defense, surveillance, and asset management. The invention presents image processing techniques for real-time forensic applications. DARPA funded U.S. patent 11282249, "System and Method for Perspective Preserving Stitching and Summarizing Views."
DARPA also funded inventors at Honeywell International Inc., who developed a way to detect the manipulation of an image from a smartphone. U.S. patent 11288537, "Image Forensics Using Non-Standard Pixels," enables the identification of forged images and the identification of a device model that generated a digital image. The invention leverage cues from modern smartphone camera hardware, specifically non-standard, focusing, or dual pixels. The invention uses the physical differences in the sensor hardware of smartphone cameras and other image sensing devices to assess local image statistics around the expected locations of the non-standard pixels to verify that they are intact.
Dr. Shweta Jain is an Associate Professor in the Department of Mathematics and Computer Science at John Jay College of Criminal Justice, Graduate Faculty in the Digital Forensics and Cyber Security, the inventor of 11297500, "Authenticating Digital Evidence." Dr. Jain's invention enables verifying the authenticity of legal evidence in a digital format. Her invention incorporates blockchain technology to provide integrity and spatiotemporal properties of digital evidence captured by a camera. Spatiotemporal data are data that relate to both space and time. The patent presents a computer security architecture that provides a unique approach for authenticating digital media content such as images, video, and audio recordings produced by a mobile electronic device such as a smartphone's camera. This technology is useful for journalists, eyewitnesses, crime scene or other investigators, human rights activists, insurance companies, and social media. The invention also preserves privacy or anonymity while verifying the authenticity of digital evidence. The National Science Foundation Computer & Information Science & Engineering Directorate funded this research.
Forensic Examination of Audio Files
Oak Ridge National Lab has invented techniques for the forensic authentication of digital audio recordings. The invention detects tampering with a digital recording when the frequency and phase sequences, the noise in the file, differ from the matched reference sequences. The reference sequences can identify where the file was recorded using both GPS and information on the U.S. power grid. If the location data about the recording doesn't match the expected frequency and phase sequences, the audio file has been tampered with.
The sensors, referred to as Frequency Disturbance Recorders (FDRs) by the inventors, may collect highly accurate Global Positioning System (GPS) time-stamped measurements, including frequency and phase angle measurements, at the distribution level of the power grid. Inventors from Oak Ridge National Lab received U.S. patent 11069370, "Tampering Detection and Location Identification of Digital Audio Recordings."
Inventors from the University of Florida received U.S. patent 11176960, "Method and Apparatus for Differentiating Between Human and Electronic Speak for Voice Interface Security." The patented technology is designed as a cybersecurity tool to distinguish between human and electronic voices used for Internet of Things devices and sensors. The invention leverages the unique characteristics of a human voice. The patent notes,
"The human voice is created by the complex interaction of various parts of the human anatomy. Sounds are produced by a combination of the lungs, the larynx, and the articulators (the tongue, cheeks, lips, palate, throat, and nasal cavity). The lungs force air over the rest of the vocal tract allowing it to produce sound. The larynx contains the vocal cords which are responsible for the generation of the fundamental frequency (e.g., a person's fundamental frequency is the lowest frequency present in their voice) present in the voice. Since the vocal cords are located at the bottom of what is essentially a closed tube, the fundamental frequency induces an acoustic resonance. This resonance generates harmonic frequencies of the fundamental frequency as it travels up and out of the human speaker's vocal tract. The articulators then alter the waveform generated by the vocal cords in order to produce the wide range of sound present in human speech. Specifically, articulators block or greatly diminish the amplitude of certain harmonics for different parts of speech."
This technology provides cybersecurity and may also be useful in detecting electronic files used to create deep fakes. The National Science Foundation funded this work.
DARPA-funded researchers at SRI International developed a way to automatically recognize and classify speech in live, real-time streaming audio. The technology can classify a speaker, can recognize multiple languages, and provide a score on the accuracy of the classification of the speech. The patent notes that the speech classification technology can be used to improve close captioning, forensic analysis, automatic translation, and automatic transcription services for live multi-language and multi-speaker audio streams. (Like the meetings at the United Nations.) The technologies can improve the interpretations of live speech made by speech-enabled smart devices configured with general-purpose and domain-specific chatbots or automated intelligent assistants. And it will be useful for enabling intelligence analysts and law enforcement to figure out who is speaking on wiretaps, crime scene audio, and video audio track. (See 11024291, "Real Time Class Recognition for An Audio Stream.")
Ok. Back to this week's taxpayer-funded patents.
This week there is one Bayh-Dole Scofflaw, an entity that didn’t provide the government agency that funded the research and the contract number. This week’s entry comes from SRI International Sarnoff in Princeton, NJ. SRI International is a nonprofit, independent research institute serving government and industry. SRI International received 11301703, "Robust Biometric Access Control Based on Dynamic Structural Changes in Tissue." This invention, despite its fancy name is technology to make sure the biometric being presented to access to a facility or system is being presented by a live human.
A presentation attack on a biometric security system generally involves the presentation of a manufactured artifact, recorded data representation, or altered or dead tissue to a biometric sensor to "spoof" (e.g., gain unauthorized access to) the security system or environment to which the biometric security system controls access. Their biometric access control system includes a liveness measurement unit that is configured to process the high-resolution image data to detect changes in one or more structural features that are indicative of live human tissue.
In 2017, SRI International was awarded a four-year $12.5 million contract by IARPA’s Odin Program to research and develop “dynamic biometrics” that can more effectively detect attempts to evade or dupe biometric security systems. Under this contract, SRI International will create and deliver a prototype Multi-physiological Joint Optimization and Liveness Nuances for Identity Ratification system to defeat known and unknown presentation attacks.
The press release identifies Jeffrey Lubin, Ph.D., a senior research scientist at SRI as the principal investigator for SRI’s Odin team. Dr. Lubin is the first-named inventor on this patent. We’ll chalk this one up to IARPA.
Patents By The Numbers
On Tuesday, April 12, 2022, the U.S. Patent and Trademark Office (USPTO) issued 7,240 patents. One hundred fifty (150) of these patents benefitted from taxpayer funding. Here is how they break down.
One hundred forty-two (142) patents have Government Interest Statements.
Thirty (30) have a government agency as an applicant or an assignee.
A federal department is the only assignee on 20 patents.
The 150 new patents have 212 department-level funding citations.
These patents are the work of 509 inventors.
The 493 American inventors come from 34 states and the District of Columbia.
The 16 foreign inventors come from ten (10) countries.
Six inventors of the foreign inventors are from China
There are 88 patents (59%) where at least one assignee is a college or university, the HERD.
Federally Funded Research and Development Centers (FFRDCs) received nine (9) patents.
Six (6) patents were assigned Y CPC symbols indicating that the invention may be useful in mitigating the impact of climate change.
The Big Three States and the Commonwealths
The United States has four states that call themselves commonwealths — Kentucky, Massachusetts, Pennsylvania, and Virginia. "Commonwealth" is a traditional English term for a political community founded for the common good.
This week the Commonwealth of Virginia appears in the top three states along with the Commonwealth of Massachusetts. The Commonwealth of Pennsylvania came in fourth. There are no patents this week from the Commonwealth of Kentucky,
California has 27 first-named inventors and 88 total inventors.
Massachusetts has 14 first-named inventors and 53 total inventors.
Virginia has nine first-named inventors and 26 total inventors.
Count By Department
The Tech Center Mix
The chart below shows the patent count by Technology Center.
The Health Complex
As of April 12, 2022, 725 patents granted by the U.S. Patent and Trademark Office in 2022 indicate that the inventors received funding from the Department of Health and Human Services. These patents include 2,409 different funding citations indicating that the inventor(s) received either a grant or a contract from the U.S. government that resulted in the patented invention. You can access the details here.
Before We Go
We heard from our correspondents in Ukraine after 17 days. You can read the latest update here.
If you aren't a subscriber yet, please consider subscribing. Also, please share FedInvent with other innovation enthusiasts.
We’ll be back next week with the latest FedInvent Report on April 17th’s patent application. Enjoy the holiday weekend.
Thank you for reading FedInvent.
The FedInvent Team
FedInvent tells the stories of inventors, investigators, and innovators. Wayfinder Digital's FedInvent Project follows the federal innovation ecosphere, taxpayer money, and the inventions it pays for. FedInvent is a work in progress. Please reach out if you have questions or suggestions. You can reach us at firstname.lastname@example.org.
FedInvent is a reader-supported publication focused on the federal innovation ecosphere. To receive new posts and support our work, consider becoming a free or paid subscriber.