Digital Commons Network^™

Grounding Structure Designs That Mitigate Undesired Resonances In High-Speed Signals, Yansheng Wang, Hanfeng Wang, Ken Wu

Defensive Publications Series

This publication discusses a specific ground-signal-signal-ground (GSSG) configuration of a high-speed differential signal pair. Simulations show that frequency resonances occur when a quarter-wavelength resonant structure is unintentionally created, which causes the observed frequency resonances. To remove the resonances, this publication recommends two design methods. Each proposed method has its own benefits and limitations. The proposed solutions may be treated as design rules for similar high-speed differential signal pairs.

Optimizing For Interpupillary Distance In Augmented Reality Eyewear, Shiblee Hasan

Defensive Publications Series

To create an accurate field of view (FoV), augmented reality (AR) eyewear is optimally matched to the user’s interpupillary distance (IPD). However, the IPD varies from person to person. This disclosure describes AR eyewear that (with user permission) automatically measures the IPD of a user and adjusts its depth and width to match the user’s IPD. The user facing interface is personalized based on the physical FoV of the user’s eyes.

Improving Call Quality By Generation Of Replacement Packets, Mukesh Agrawal, Marie Janssen, Zaccariah Troy Bowling

Defensive Publications Series

Voice communications often suffer from lost packets due to, e.g., radio channel fades, network errors, etc., resulting in audio glitches and poor call quality. This disclosure describes machine learning techniques to replace lost packets. By masking audio glitches and providing continuity, the replacement packets improve call quality as perceived by call participants.

Recommending Content Based On Anticipated Knowledge Gain, Victor Cărbune, Alexandru Damian

Defensive Publications Series

This disclosure describes machine learning techniques to automatically remove, reorder, or prioritize segments of content such that users can quickly access information that is new to them. With user permission, data such as user profile, historical content consumption patterns and its relevance, etc. is obtained. This data is used to chunk and rank historical content segments based on relevance to the user. A machine learning model is trained to identify new content (or segments thereof) based on the knowledge gain that the content is likely to provide to the user.

Multi-Camera Arrays To Detect Posture, Jonathan D. Hurwitz, Christina L. Gilbert

Defensive Publications Series

This disclosure describes techniques that, with user permission, automatically detect the posture of a user of a personal computer. The user’s posture is detected using hardware typically found on a desktop or laptop PC, e.g., cameras, without requiring extra sensors or other additional hardware. If the detected posture is determined to be poor, the user is alerted. By detecting and alerting the user of poor posture, the techniques can help forestall health problems. The techniques are implemented with user permission. Users are provided with options to turn off posture detection.

Anti-Spoof Detection For In-Screen Fingerprint Sensors, Vishwath Mohan, James Brooks Miller

Defensive Publications Series

User authentication prior to accessing contents of a secured device or container can be achieved in many ways, including textual passwords, picture passwords, locks, keys, or other codes. Many devices also allow for biometric authentication via fingerprint scanners, iris scanners, facial recognition, or the like for increased security and simplified user experience. In concept, a biometric scanner would eliminate all but a single individual from obtaining access to the device or container. However, many biometric scanners can be fooled, tricked, or spoofed. For example, an unauthorized user can bypass some fingerprint scanners obtaining a latent fingerprint, manipulating a copy of …

Non-Annual Queries And Events, Bryan Horling

Defensive Publications Series

This disclosure describes techniques that provide a user with information relating to non-annual events. For example, such events include a 10,000-day birthday, billionth second since wedding, etc. With user permission, a platform such as a social media service, a virtual assistant, etc. accesses information such as personal profile, contact information, online calendars, etc. to determine non-annual events likely of interest to a user, and to formulate answers to time-seeking queries requested by the user in non-annual units. An option is provided to the user to share milestones expressed in non-annual time units.

User Equipment (Ue) Bimodal User Interface (Ui), Jakob Foerster, Oli Gaymond

Defensive Publications Series

This publication describes a user equipment (UE), such as a smartphone, that offers a user a bimodal user interface (UI) having an input mode and an output mode. The user may use various techniques to instruct the UE whether the UI needs to be in input mode or in output mode. As described herein, input mode refers to a user experience in which the user instructs the UE what to do, such as perform a computation, write an email, make a call, draw a figure, take notes, and other user-initiated activities. The UI in input mode silences and hides some …

Context Based Speech Enhancement Techniques For Voice And Media Applications, Snehitha Singaraju

Defensive Publications Series

Smart devices continue to proliferate, and they provide a variety of functions to assist end users. Voice-interactive (voice-activated) and audio-interactive devices continue to increase in popularity, availability, and functionality. Voice-interactive and audio-interactive devices may improve by monitoring and determining context when distinguishing speech from background noise and distinguishing speech intended for the voice-interactive and audio-interactive device and speech nearby but irrelevant to the device. A machine-learning model can process detected keywords, contextual cues, historical cues, and other information in determining relevant speech from background noise.

Method For Classifying Wake-Up Alarms Using Machine Learning (Ml), Thomas G. Price

Defensive Publications Series

This publication describes an operating system (OS) of a user equipment (UE), such as a smartphone, that can classify a wake-up alarm from other types of alarms. The importance of such classification stems from the fact that when a user sets a wake-up alarm on the smartphone, they may or may not turn on a do-not-disturb (DND) feature. In addition, the UE’s DND feature may differ, such that it may silence calls and alerts, but keep the luminosity of the UE’s screen the same, it may silence calls and alerts and lower the luminosity, or a combination thereof. To aid …

A Smart Task Assistant To Instruct Users In How To Perform A Task Based On Machine Learning Techniques By Voice, Hp Inc

Defensive Publications Series

A voice assistant that guide users with different levels of knowledge through an operation or

repair. The proposed solution is a camera that can be attached to the user’s security glasses or in a headset

that will transmit the Point of View (POV) and receive instructions about what is the next action that the

user needs to perform to complete a given task. A machine learning algorithm is responsible to process all

the data input and give instructions to the user about how to perform an action. The instructions are

transmitted to the user’s ear-piece.

This technology streamlines the maintenance …

A Proactive Network Analyzer Using X86 Platform With Linux-Based Operating System, Hp Inc

Defensive Publications Series

Often times we need to deal with complex network issues during debugging. E.g., MAC/IP address

conflicts, magic package loss, network cable disconnected…etc. Issues like these require an engineer’s

expertise and time to check the log in the network environment and sometimes the symptom doesn’t

happen anymore.

We run into countless network issues during the product development stage. We hope to create a

proactive network analyzer that filters/spots the error once we install it in the network. This will

eliminate the needs for post capturing the log and the time to analyze it.

On-Device Dialog Management, Vinod D. Krishnan, Yanhong Chen

Defensive Publications Series

This disclosure describes techniques for on-device dialog management that enable a voice-based multi-turn dialog without the need for network connectivity. Although the techniques support a fully hands-free mode, parts of the dialog can also be entered by the user using a touch or other interface.

Communication Between Moving Vehicles Using A Virtual Assistant, Surath Sundram, Rupika Chatterjee

Defensive Publications Series

This disclosure describes virtual assistant based techniques for communication between vehicles in proximity to each other. Users register pertinent information such as license plate number, contact information, etc. at a registry. While driving, a user requests their virtual assistant to send a message to a nearby vehicle, e.g., by providing an observed license plate number or a portion thereof. With user permission, location data is accessed to identify nearby vehicles that match the user’s specification. When a nearby vehicle is identified that matches the user specification, the virtual assistant checks if the user of the identified vehicle permits being contacted. …

Customized Query Responses Based On Available Appliances, Ashish Duggal

Defensive Publications Series

This disclosure describes techniques to provide customized responses to user queries for household tasks, e.g., use of an appliance such as a washer/dryer. The responses are customized based on information regarding appliances available within the household, obtained with user permission. The query may be received by a virtual assistant, e.g., available via a household appliance, or via device such as a smartphone, smart speaker, personal computer, etc. Information regarding the appliances is obtained, e.g., by indexing product manuals and other information. Results are provided that guide the user to complete the household task using an available appliance.

Wireless-Connection Drop Prediction, Warning, And Protection, Jim Sun, Aaron Lee

Defensive Publications Series

This publication describes systems and techniques directed to wireless-connection drop prediction, warning, and protection. A wireless-communication device, such as a smartphone, includes a wireless-connection manager application. The wireless-communication device, having a wireless connection to an access point of a wireless network, performs operations under the direction of the wireless-connection manager application to predict, warn, and protect a user of instances where the wireless connection may drop. The operations include determining that a quality metric of a signal supporting the wireless connection to the access point does not meet a threshold, determining that another access point to the wireless network is …

Long Conversation Mode For Virtual Assistant, Surath Sundram, Rupika Chatterjee

Defensive Publications Series

This disclosure describes a virtual assistant that, with specific user permission, can perform queries and offer suggestions based on user conversation and context. When configured by the user with suitable permissions, the virtual assistant can perform queries without the user explicitly issuing a hotword to activate the virtual assistant. For example, the user can enable a long conversation mode for a virtual assistant on a device such as a smartphone, tablet, smartwatch, etc. The virtual assistant can automatically provide suggestions, responses, etc.

Hybrid On-Device Cloud Scheme For Re-Identification Of Persons Based On Shared Embedding Gallery, Nhat Vu, Daniele Midi, Tianchun Yang, Bill N. Schilit

Defensive Publications Series

Generally, the present disclosure is directed to a system of facial and/or person recognition via machine learning and Internet of Things (IoT). In particular, in some implementations, the systems and methods of the present disclosure can include or otherwise leverage a machine learning and IoT system or device to track and/or identify a person based on video images taken by one or more device(s). For example, a hybrid on-device and cloud scheme can enable locally-derived embeddings from multiple camera devices to be sent to a shared cloud space which can cluster the embeddings to generate a person model for a …

Text Data Embedded Into A Voice Call, Jim Sun, Aaron Lee, Gary Jian

Defensive Publications Series

This publication describes systems and techniques directed to embedding text data into a voice call. A first wireless-communication device, such as a smartphone, includes a text-voice manager application. The first wireless-communication device performs the voice call with a second wireless-communication device that includes a peer text-voice manager application. The first wireless-communication device, by executing instructions of the text-voice manager application, performs operations that include receiving text data for a text message, converting the text data to a voice message, detecting silent periods in a stream of a voice transmission, and embedding the voice message into the detected silent periods in …

Fingerprint-Matching Algorithms, Firas Sammoura, Kuntal Sengupta, Jean-Marie Bussat

Defensive Publications Series

This publication describes a fingerprint-matching algorithm that uses a fusion of minutiae and pattern-correlation matching. Instead of extracting minutiae points, the algorithm extracts small patches from the fingerprint image and transforms them into rotationally-invariant vectors. The algorithm divides each image to be evaluated, herein called an “enrolled” image, into “M” number of patches with a sliding distance of one (1) pixel. The algorithm also extracts “N” number of random patches from a stored image, herein called a “verify” image, and calculates the similarity between the “verify” rotationally-invariant vectors and the “enrolled” rotationally-invariant vectors. At this stage, the algorithm merges vectors …

Audio-Recording Techniques Using Machine Learning (Ml), Snehitha Singaraju, Moonseok Kim

Defensive Publications Series

This publication describes audio-recording techniques, which allow a user equipment (UE) to determine a user’s intent and apply audio beamforming and wind-noise filtering based on the determined intent. This provides the user with an enhanced user experience (UX) by enabling the user to focus on the moment and use the UE to record or capture the acoustic signals that best-represent the auditory scene. Among other aspects, the invention uses machine learning (ML) to analyze the auditory scene and the environment of the user.

Transferring Application Software From One User Equipment (Ue) To Another, Victor Carbune, Daniel Keysers, Thomas Deselaers

Defensive Publications Series

An operating system (OS) of a user equipment (UE), such as a smartphone, supports various first-party and third-party application software. Many application software offer personalized user experience (UX) depending on user settings. A user may spend a considerable amount of time and effort downloading the application software and configuring their settings. The user, however, periodically replaces the smartphone, and often they download and configure the settings of various application software again on the new UE, same as they had done on the previous UE. This leads to a lengthy process of migrating from one UE to another. This publication describes …

Serverless Cybersecurity Training

Defensive Publications Series

Training students and professionals in cybersecurity practices is expensive due to the complex laboratory environments required to properly isolate the trainees. Traditional methods include downloading and executing a virtual machine locally or accessing cloud-based virtual machines. In this work, this cost is reduced by obtaining static files from server that cause a user’s browser to execute an HTML5 Service Worker and a Web Worker. The Service Worker intercepts browser requests to one or more domains and redirects them to the Web Worker. The Web Worker, which provides the vulnerable services the trainee interacts with, responds to the user.

Efficient Queue Management In A Packet Buffer Between Processes, Ian Wells, Kyle Andrew Donald Mestery

Defensive Publications Series

Techniques are described herein for queueing between multiple threads and processes. These techniques provide for a low-lock queue to allow multiple transmitters and receivers to successfully use a single queue efficiently. In particular, the techniques presented herein provide tactics for passing packets in a shared memory area from one process with threads to another process with a potentially different number of threads, requiring different transmitter (Tx) and receiver (Rx) queue counts on the two sides of the connection and avoiding stalls in the multiple workers as they operate on shared data structures.

Customizing User Interfaces Based On Usage Preferences And Practices, Matthew Urquhart

Defensive Publications Series

User interfaces (UI) of applications are designed to provide elements connected to all of the various features regardless of their frequency of use. The inclusion of features rarely used by a user within the application UI can lead to a cluttered interface that wastes screen real estate and makes it difficult for the user to select the elements for frequently used operations, thus slowing the user down. Applications sometimes provide users the option to control the visibility and grouping of the menu options and/or toolbar buttons; however, these options require manual input from the user. This disclosure describes adaptive UI …

Load Balancing Of Phone Traffic Generated From Web Interactions, Jared M. Zimmerman, Travis Neilson

Defensive Publications Series

Click-to-call actions, e.g., provided in online advertisements, enable users to call businesses by clicking on a link. When a number of users simultaneously perform click-to-call actions, such simultaneous requests can outstrip the capacity of a business to respond to queries. The techniques of this disclosure detect click-to-call attempts in excess of the capacity of a business and distribute caller load amongst multiple businesses that offer similar goods or services.

Open Loop State Machine Synchronization, Hp Inc

Defensive Publications Series

This is a method for using an open loop state machine synchronization on a shared bus to emulate

electromechanical hardware devices for software development.

Trusted Certificate Installation Service For Printers With Embedded Web Server, Hp Inc

Defensive Publications Series

Herein a web‐based service is proposed which provides printer users with a straightforward and simple way to secure

their connection to printers supporting Embedded Web Server [EWS] functionality.

Media Suggestions Based On Face Recognition, Will Hennessy

Defensive Publications Series

To initiate or continue media playback on a device such as a smart speaker, the user needs to provide input, e.g., a voice command, touchscreen-based input, etc. This disclosure describes techniques to automatically recommend media content to users upon detecting the user’s presence, e.g., using facial recognition techniques. With user permission, media consumption, e.g., music, radio shows, podcasts, television shows, online videos, etc. of a user is analyzed to determine usage patterns. User-specific media suggestions are determined based on the usage patterns. When it is detected that the user is near a media playback device, e.g., smart speaker, smart television, …

Enabling Trust Between People And Mr/Vr Users, Anonymous Anonymous

Defensive Publications Series

The present disclosure describes a system that enables trust between a Mixed Reality (MR)/Virtual Reality (VR) user and people in the vicinity, so that the people can be assured if the user is aware of their presence or not. A light strip with LEDs is embedded in an MR/VR headset, along with a camera. The camera records a video of the user’s surroundings. The video is wirelessly transferred to an Application Program Interface (API) of an MR/VR application. A pre-trained person detection model of the API detects people in vicinity of the user. The API determines a type of session …