Abstract: A system can receive respective data from respective sensors, wherein the respective data represents respective gradient values for a neural network produced by the respective first sensors according to a federated learning process. The system can transform the respective data into respective analog waveforms. The system can apply orthogonal frequency-division multiplexing to the respective analog waveforms to produce respective aligned analog waveforms. The system can create a superposition analog waveform that comprises a superposition of the respective aligned analog waveforms. The system can transmit the superposition analog waveform to an access point, wherein the access point is configured to update the neural network with the superposition analog waveform according to the federated learning process.

Abstract: The described technology is generally directed towards accelerating distributed principal components in the presence of noisy channels. A federated training based method is disclosed. The method can calculate a desired common subspace for edge devices under the coordination of a server. The server can be connected to the edge devices via noisy wireless channels. A broadband communication system can be used, wherein devices can transmit local gradients by linear analog modulation over sub-channels in communication rounds for over-the-air aggregation. Before each communication round, the server can detect information of a current region. Based on the region information, an online region-adaptive power control scheme can be applied to accelerate the process.

Abstract: Interference alignment for a multiple-input multiple-output communications network with partial connectivity is provided. A method is provided that includes determining assignments for data streams transmitted in the multiple-input multiple-output communications network. The method also includes suppressing inter-cell interference and suppressing intra-cell interference. Also provided is inter-cell/intra-cell decomposition and exploitation of partial connectivity in a multiple-input multiple-output communications network. The multiple-input multiple-output communications network can comprise three or more cells.

Abstract: Cooperative physical layer caching systems for base stations are described herein. Base stations include cache memories that store portions of content. Base stations receive requests for content files from mobile devices. Cooperative physical layer caching systems can determine contents of the cache memories and can coordinate physical transmissions of the content files based on the contents of the cache memories.

Abstract: Channel state information (CSI) is reduced for a multi input multi output (MIMO) channel, in which feeding back a part of the CSI matrix is sufficient to align interference in a MIMO network. A feedback dimension quantifies a cost of the CSI feedback of the interference in the MIMO network. A feedback profile is determined that achieves a tradeoff between degrees of freedom, antenna resources and the cost of the feedback. The feedback profile parameterizes a feedback function, which determines how the CSI matrices are fed back to the transmitter. The precoders of the network are adaptive to only the partial CSI knowledge at the transmitter.

Abstract: Signal interference associated with multi-user multiple-input multiple-output (MIMO) full-duplex wireless communication systems is reduced. In an example, first streams are received by a MIMO base station from uplink user devices based on a first availability of channel matrices and second streams are transmitted to downlink user devices based on a second availability of the channel matrices. A weighted sum of a first data rate for a first linear transceiver matrix and a second data rate of a second linear transceiver matrix is maximized based on the channel matrices, the first streams, and the second streams. The base station generates a first linear transceiver vector used in transmission of the first streams to the base station from the uplink user devices and a second linear transceiver vector used in transmission of the second streams from the base station to the downlink user devices.

Abstract: A subset of mobile device candidates in a multi-input multi-output (MIMO) channel of network devices is selected for having a feedback priority among a set of mobile device candidates. The selection is based on a set of queue state information of a plurality of mobile device queues for transmitter and receiver devices of the MIMO channel in a multi-user MIMO network. A portion of the subset of mobile device candidates can be selected by a scheduling component of the system. Communications on the MIMO channels can be scheduled to the portion of the subset of mobile device candidates based on the feedback comprising channel state information and on the queue state information of the plurality of mobile device queues.

Abstract: A wireless coverage characterization platform uses an autonomous vehicle or robot, such as an unmanned aerial vehicle or other small robot, to autonomously collect key wireless coverage parameters for an indoor environment. One or more vehicles or robots are equipped with integrated simultaneous localization and mapping sensors as well as wireless signal measurement sensors. As a vehicle traverses the indoor environment, on-board processing components process the sensor measurement data to simultaneously build an indoor map of the environment and to learn the wireless coverage characteristics of the environment incrementally. The vehicle's navigation system guides the vehicle through the environment based on the sensor measurements and the learned indoor map until a complete map of the wireless signal strength at all locations throughout the environment is obtained.

Abstract: Cooperative caching systems incorporating Plug-and-Play base stations are described herein. Plug-and-Play base stations with large caching capacities are employed in a wireless network to perform cooperative transmission with macro base stations. Each Plug-and-Play base station can either have wireless backhaul or a low-cost wired backhaul connection to the macro base stations. Cooperative caching systems can direct traffic between the Plug-and-Play base stations and the macro base stations.

Abstract: A wireless coverage characterization platform uses an autonomous vehicle or robot, such as an unmanned aerial vehicle or other small robot, to autonomously collect key wireless coverage parameters for an indoor environment. One or more vehicles or robots are equipped with integrated simultaneous localization and mapping sensors as well as wireless signal measurement sensors. As a vehicle traverses the indoor environment, on-board processing components process the sensor measurement data to simultaneously build an indoor map of the environment and to learn the wireless coverage characteristics of the environment incrementally. The vehicle's navigation system guides the vehicle through the environment based on the sensor measurements and the learned indoor map until a complete map of the wireless signal strength at all locations throughout the environment is obtained.

Abstract: Systems, methods, and apparatus are described that that increase throughput of a WiFi network. A first method is provided wherein access points monitor and keep track of their states on each resource block (frequency channel and antenna pattern) associated therewith and dynamically select the resource blocks that increase network throughput based on the instantaneous states of the access points. A second method is provided comprising employing Q-learning to determine one or more modifications of operating parameters of a network node based on observed throughput of the network and implementing the one or more modifications at the node. A third method is also provided which combines the first and second methods so as to increase network throughput at both the physical layer and the MAC layer.

Abstract: Functionality is shared among mobile devices in a peer to peer manner. A mobile device can include a plurality of functional components that can each perform respective functionality. Examples of the functionalities can include transceiver communications, processing, power, memory, input and output for the mobile device. Further, the mobile device can include a sharing component that enables sharing a particular third party functional component to replace or supplement operation of a corresponding functional component of the mobile device. The third party functional component, for instance, can be made available for sharing by at least one of a disparate mobile device or a stand alone functional component. Moreover, a host component can allow a disparate mobile device to use an available one or more of the plurality of functional components of the mobile device.

Abstract: Signal interference associated with multi-user multiple-input multiple-output (MIMO) full-duplex wireless communication systems is reduced. In an example, first streams are received by a MIMO base station from uplink user devices based on a first availability of channel matrices and second streams are transmitted to downlink user devices based on a second availability of the channel matrices. A weighted sum of a first data rate for a first linear transceiver matrix and a second data rate of a second linear transceiver matrix is maximized based on the channel matrices, the first streams, and the second streams. The base station generates a first linear transceiver vector used in transmission of the first streams to the base station from the uplink user devices and a second linear transceiver vector used in transmission of the second streams from the base station to the downlink user devices.

Abstract: Cooperative physical layer caching systems for base stations are described herein. Base stations include cache memories that store portions of content. Base stations receive requests for content files from mobile devices. Cooperative physical layer caching systems can determine contents of the cache memories and can coordinate physical transmissions of the content files based on the contents of the cache memories.

Abstract: Systems, methods, and apparatus are described that employ specially designed interference patterns to build an in-band side channel without degrading the effective throughput of the main channel. Transmitter and receiver pairs implement a two-stage dynamic interference mitigation scheme based on the partial connections in limited dimension MIMO interference channels. The first stage determines the stream assignments and the subspace constraints for beamforming vectors and zero-forcing vectors based on the partially connected topology such as the path loss, shadowing and spatial correlation. The second stage determines the beamforming vectors and the zero-forcing vectors based on the instantaneous channel state information over the subspaces obtained from the first stage. In an aspect, a controller device implements the first and second stage and assigns the transmitter and receiver pairs to perform interference alignment using the stream assignments and beamforming and zero-forcing vectors respectively.

Abstract: Channel state information (CSI) is reduced for a multi input multi output (MIMO) channel, in which feeding back a part of the CSI matrix is sufficient to align interference in a MIMO network. A feedback dimension quantifies a cost of the CSI feedback of the interference in the MIMO network. A feedback profile is determined that achieves a tradeoff between degrees of freedom, antenna resources and the cost of the feedback. The feedback profile parameterizes a feedback function, which determines how the CSI matrices are fed back to the transmitter. The precoders of the network are adaptive to only the partial CSI knowledge at the transmitter.

Abstract: A subset of mobile device candidates in a multi-input multi-output (MIMO) channel of network devices is selected for having a feedback priority among a set of mobile device candidates. The selection is based on a set of queue state information of a plurality of mobile device queues for transmitter and receiver devices of the MIMO channel in a multi-user MIMO network. A portion of the subset of mobile device candidates can be selected by a scheduling component of the system. Communications on the MIMO channels can be scheduled to the portion of the subset of mobile device candidates based on the feedback comprising channel state information and on the queue state information of the plurality of mobile device queues.

Abstract: Interference alignment for a multiple-input multiple-output communications network with partial connectivity is provided. A method is provided that includes determining assignments for data streams transmitted in the multiple-input multiple-output communications network. The method also includes suppressing inter-cell interference and suppressing intra-cell interference. Also provided is inter-cell/intra-cell decomposition and exploitation of partial connectivity in a multiple-input multiple-output communications network. The multiple-input multiple-output communications network can comprise three or more cells.

Abstract: A transmission protocol is provided for wireless communications in an OFDMA system that uses a rateless code and techniques are provided for dynamically adjusting the subcarrier resources allocated to an OFDMA relay node so that the relay is used at the right time according to the instantaneous channel state information (CSI). As a result, a higher throughput is achieved between source and destination nodes based on a low complexity algorithm for determining when relay assistance is helpful for completing a transmission between the source and destination nodes.

Abstract: Systems, methods, and apparatus are described that employ a partial interference alignment scheme in conjunction with an interference detection scheme to mitigate interference in quasi-static multiple input multiple output (MIMO) channels. In an aspect, sets of transmitter and receiver pairs are determined based on feasibility of alignment of the transmitters at the receivers of the sets in accordance with an interference alignment scheme and path loss associated with signal transmissions between the transmitter and receiver pairs, wherein the sets include different alignment assignments of the respective transmitters at the respective receivers of the sets. The alignment assignments can include different numbers of the transmitters aligned at the receivers of the sets. A set is then selected from the sets that contributes to an interference profile at the receivers that facilitates interference detection at the receivers.