Microsoft Technology Licensing, LLC

Lucent Technologies Inc.

Alcatel Lucent

Alcatel-lucent USA Inc.

Ramachandran Ramjee

Thomas F La Porta

Vishnu Navda

Venkata N Padmanabhan

Thomas Y Woo

Kannan Varadhan

Tirunell V Lakshman

Li Erran Li

Thyagarajan Nandagopal

Kazutaka Murakami

Sandra Ramos Thuel

Furquan Ahmed Ansari

Martin Havemann

Jitendra Dattatraya Padhye

Bozidar Radunovic

Malathi Veeraraghavan

Ranjita Bhagwan

Ashish Sharma

Krishna Kant Chintalapudi

Paramvir Bahl

Harish Viswanathan

Dharma Kiritkumar Shukla

Krishan Kumar Sabnani

Kun Tan

Tian Bu

Bernard D Aboba

Richard Wayne Buskens

Krishna Chintalapudi

Luca Salgarelli

Pankaj Sharma

Muthian Sivathanu

Saikat Guha

Yow-Jian Lin

Rimma V Nehme

Miguel Oom Temudo De Castro

Nishanth Chandran

Gopala Srihari Narlanka

Nipun Kwatra

Shravan Rayanchu

Mun-Choon Chan

Bozidar Radunović

Horia Vlad Balan

Mun Choon Chan

Michael Buettner

Devarayasamudra Chandramouli Vageesh

Romil Bhardwaj

Ranveer Chandra

Thomas Moscibroda

Arun N Netravali

Kamal Jain

Richard Dennis Gitlin

Prashanth Hosabettu Mohan

Milind Madhav Buddhikot

Scott Corey Miller

Pradeep Padala

Saikrishna Badrinarayanan

Lenin Ravindranath Sivalingam

George Varghese

Ashok Anand

Ganesan Ramalingam

Apurv Bhartia

Atul Katiyar

Danny Raz

Thomas Yat Chung Woo

Gerard Terence Foster

Yuval Shavitt

Bhalakumaaran Erode Ranganathan

Mark Haner

Sahil Suneja

Sandeep P Karanth

Vaibhav Sharma

Thomas F LaPorta

Shreshth Singhal

Michael Patrick Deitz

Srinidhi Viswanatha

Alfred Vaino Aho

Li Li

Sneha Kasera

Tathagata Das

Prasun Sinha

Bhavish Aggarwal

Gopi Krishna Tummala

Antonios Papadimitriou

Aaron Schulman

Manveer Singh Chawla

Bozidar Raduvic

Deeparnab Chakrabarty

Pushkar Chitnis

Chitra Muthukrishnan

Athula Balachandran

Ananth Balashankar

Eugenio Magistretti

Harmeet Singh

Venkat N Padmanabhan

The disclosure herein describes scheduling execution of artificial intelligence (AI) workloads in a cloud infrastructure platform. A global scheduler receives AI workloads associated with resource ticket values. The scheduler distributes the AI workloads to nodes based on balancing resource ticket values. Local schedulers of the nodes schedule AI workloads on resources based on the resource ticket values of the AI workloads. Based on scheduling the AI workloads, coordinator services of the local schedulers execute the distributed AI workloads on the infrastructure resources of the nodes. The disclosure further describes scheduling AI workloads based on priority tiers. A scheduler receives AI workloads, and each AI workload is associated with a priority tier indicative of a preemption priority while being executed. The AI workloads are scheduled for execution on a distributed set of nodes based on the priority tiers and then execute based on the scheduling.

Scheduler for planet-scale computing system

The disclosure herein describes platform-level checkpointing for deep learning (DL) jobs. The checkpointing is performed through capturing two kinds of state data: (i) GPU state (device state), and (ii) CPU state (host state). The GPU state includes GPU data (e.g., model parameters, optimizer state, etc.) that is located in the GPU and GPU context (e.g., the default stream in GPU, various handles created by the libraries such as DNN, Blas, etc.). Only a fraction of the GPU memory is copied because the checkpointing is done in a domain-aware manner. The 'active' memory contains useful data like model parameters. To be able to capture the useful data, memory management is controlled to identify which parts of the memory are active. Also, to restore the destination GPU to the same context/state, a mechanism is used to capture such state-changing events on an original GPU and replayed on a destination GPU.

Transparent pre-emption and migration for planet-scale computer

A 'Data Transfer Tool' extracts, labels and stores user data or information that may be confined within application silos during user interaction with arbitrary apps. The Data Transfer Tool enables sharing of this otherwise siloed data shared across and between authorized apps. The Data Transfer Tool provides a task-centric approach to define and extract structured and semantically meaningful information from source applications by providing multiple semantic models that are individually tailored to particular source apps. The Data Transfer Tool applies an accessibility API or the like of the OS in combination with the semantic model for the source app to scrape user entered or selected data or information from the source app. The Data Transfer Tool enables and creates new user experiences and increases user efficiency when interacting with various apps by making the scraped data or information available to subscribing destination apps approved to receive that data or information.

Automated data transfer from mobile application silos to authorized third-party applications

This document relates to camera calibration. One example uses real-world distances and image coordinates of object features in images to determine multiple candidate camera calibrations for a camera. This example filters out at least some of the multiple candidate camera calibrations to obtain remaining calibrations, and obtains a final calibration for the camera from the remaining calibrations

Automatic camera calibration

In some embodiments, an encryption system secures data using a homomorphic encryption. The encryption system encrypts a number by encrypting a number identifier of the number and combining the number and the encrypted number identifier using a mathematical operation to generate an encrypted number. The encrypted numbers may be stored at a server system along with their number identifiers. The server system can then generate an aggregation (e.g., sum) of the encrypted numbers and provide the aggregation, the encrypted numbers, and the number identifiers. The encryption system can then separate the aggregation of the numbers from the aggregation of the encrypted numbers using an inverse of the mathematical operation used in the encryption to effect removal of an aggregation of the encrypted number identifiers of the numbers from the aggregation of the encrypted numbers. The separated aggregation of the numbers is an aggregation of the plurality of the numbers.

Aggregation of encrypted data

A 'Skip-Correlator' ensures carrier sensing symmetry between wireless devices of arbitrary transmission power levels, thereby enabling fair sharing of available spectrum of a wireless channel between the wireless devices. In various implementations, receivers of a wireless device receive wireless preambles from neighboring wireless transmitters. This preamble includes an indication of a transmission power level of the neighboring wireless transmitter. The wireless device then selects and correlates a subset of samples of the received preamble as a function of a transmission power level of that wireless device. The remaining samples of the preamble are skipped by the receiving wireless device, hence the use of the term 'Skip-Correlator.' Further, the wireless device computes a total preamble energy of the correlated samples of the received preamble. Finally, the wireless device implements carrier sensing symmetry between itself and the neighboring wireless transmitter as a function of the total preamble energy.

Skip-correlation based symmetric carrier sensing with multiple power levels

An encryption system stores encrypted values for aggregation is provided. The encryption system accesses an input set with input values. For each distinct value in the input set of input values, the encryption system generates an output set with an encrypted output value corresponding to each input value. The encryption system sets the encrypted output value for a corresponding input value to an encryption of an indicator of a match when the corresponding input value is the same as that distinct value. Otherwise, the encryption sets the encrypted output value for the corresponding input value to an encryption of an indicator of no match. The encrypted output values can then be aggregated to generate an encrypted aggregation based on input values that match, and the encrypted aggregation can be decrypted to generated a decrypted aggregation based on the input values that match.

Aggregation based on splayed data

A dynamic channel selection approach for wireless communication networks is provided by measuring an ineffective communication metric on a currently-used channel. The network can switch channels if the ineffective communication metric from a device on the network satisfies a channel selection condition. To change the wireless communication network to the new channel, all network devices in the wireless communication network are instructed to switch to the new channel. As network communications start on the new channel, the ineffective communication metric measurements begin again on the new channel. Communications on the new channel continue until the channel selection condition is satisfied on the new channel, at which point another random selection of a channel is executed. The process continues in such iterations during the operation of the network and its network devices.

Dynamic channel selection in a wireless communication network

A method for managing data communication of a mobile device in a mobile network is provided. The method may include receiving a request, at a cloud proxy, to retrieve Internet data from the Internet. The request may originate from an application, and the Internet data may have a plurality unique objects. The method may also include aggregating the Internet data and two or more of the unique objects from the Internet. Furthermore, the method may include forwarding the Internet data and the two or more unique objects to the mobile device in one transmission.

Cellular data communication for mobile devices

A multiple independent narrow-channel wireless network and method for transmitting and received data over a wireless network using a fragmented frequency spectrum. The system and method uses a plurality of narrow wireless channels obtained from splitting a wide wireless channel. Each narrow channel performs sending, receiving, and carrier sensing. Moreover, each narrow channel is independent such that data can be sent or received without any interference from other narrow channels and without synchronization. Embodiments of the system and method include a compound radio having a compound receiver and a compound transmitter. The compound transmitter includes an inter-radiolet symbol synchronization module, to permit use of a single inverse fast Fourier transform block, and a dynamically configurable filter array, to mitigate leakage between channels. The compound receiver uses fraction data rate processing to optimize efficiency. A throughput maximal metric technique is used to determine its frequency of operation in white spaces.

Multiple independent narrow channels in wireless networks

One or more mobile devices and a cloud server computing device are utilized to provide Internet access to one or more client computing devices. One of the clients is designated as a Wi-Fi access point. The Wi-Fi access point implements a reverse-infrastructure Wi-Fi mode which tethers available mobile devices and other clients to the access point. The cloud server periodically computes an optimal number of the mobile devices to be used for data striping, and transmits a webpage to a requesting client by striping data associated with the webpage across the optimal number of the mobile devices.

Energy-efficient on-the-fly Wi-Fi hotspot using mobile devices

The SoftRouter architecture separates the implementation of control plane functions from packet forwarding functions. In this architecture, all control plane functions are implemented on general purpose servers called the control elements (CEs) that may be multiple hops away from the forwarding elements (FEs). A network element (NE) or a router is formed using dynamic binding between the CEs and the FEs. There is a protocol failover mechanism for handling failovers initiated by FEs to transfer control from one CE to another CE.

SoftRouter protocol failovers

An end host redundancy elimination system and method to provide redundancy elimination as an end system service. Embodiments of the system and method use optimization techniques that reduce server central processing unit (CPU) load and memory footprint as compared to existing approaches. For server storage, embodiments of the system and method use a suite of highly-optimized data structures for managing metadata and cached payloads. An optimized asymmetric max-match technique exploits the inherent structure in data maintained at the server and client and ensures that client processing load is negligible. A load-adaptive fingerprinting technique is used that is much faster than current fingerprinting techniques while still delivering similar compression. Load-adaptive means that embodiments of the fingerprinting technique can adapt CPU usage depending on server load. Embodiments of the system and method operate above the transmission control protocol (TCP) layer, thereby reducing the number of roundtrips needed for data transfer.

Asymmetric end host redundancy elimination for networks

Novel, Internet-related architectures, methods and devices are proposed that are based on a fundamentally different philosophy: hosts (e.g., source and destination nodes) are given the ability to specify their access control policies to the network they are a part of, and the network enforces these policies. The architecture proposed is mobility friendly to the ever increasing number of mobile hosts and is scalable as well.

Methods and systems for providing controlled access to the internet

Growing community of inventors

Bengaluru, India

Ramachandran Ramjee