PRECODER MATRIX QUANTIZATION FOR COMPRESSED CSI FEEDBACK

    公开(公告)号:US20240333356A1

    公开(公告)日:2024-10-03

    申请号:US18740247

    申请日:2024-06-11

    CPC classification number: H04B7/0626 H04B7/0417 H04B7/0634 H04B7/0658

    Abstract: Certain aspects of the present disclosure provide techniques for precoder matrix quantization for compressed channel state information (CSI) feedback. A method for wireless communications by a user equipment (UE) includes receiving a CSI report configuration for frequency domain compressed precoder matrix feedback. The CSI report configuration configures the UE to report, for a plurality of selected beams at a plurality of time domain taps, a frequency domain compression basis vector and a plurality of linear combination coefficients associated with the frequency domain compression. For each of the beams, the UE groups the time domain taps into at least first and second groups. The groups each can have zero, one, or more than one time domain taps. The UE quantizes the corresponding linear combination coefficients and/or frequency domain compression basis vectors based on the grouping.

    FORWARD ERROR CORRECTION CODE RATE SELECTION FOR A FOUNTAIN CODE SYSTEM

    公开(公告)号:US20240014921A1

    公开(公告)日:2024-01-11

    申请号:US18252648

    申请日:2021-01-19

    CPC classification number: H04L1/0002 H04L1/0041 H04L1/0026

    Abstract: Aspects of the present disclosure relate to selection of a coding rate for forward error correction coding associated with Fountain code encoding. According to some aspects, a wireless device may determine a plurality of block error rates associated with a channel between the wireless device and a second device, the plurality of block error rates being associated with a plurality of coding rates. The wireless device may select a coding rate from the plurality of coding rates for forward error correction coding based on the plurality of block error rates and the plurality of coding rates. The wireless device may encode a source data packet using a Fountain code to produce a codeword, and perform the forward error correction coding on the codeword based on the selected coding rate to produce a transmit signal. The wireless device may transmit the transmit signal to the second device.

    GRADIENT DATASET AWARE CONFIGURATION FOR OVER-THE-AIR (OTA) MODEL AGGREGATION IN FEDERATED LEARNING

    公开(公告)号:US20230413070A1

    公开(公告)日:2023-12-21

    申请号:US18030261

    申请日:2020-12-14

    CPC classification number: H04W24/02 G06N3/098 H04W72/04

    Abstract: A method performed by a user equipment (UE) generates local gradients for a federated learning task. The method calculates a gradient Sum-Power level based on the local gradients. The method receives a mapping between gradient Sum-Power levels and scaling factors for channel inversion coefficients to process a data block into an unencoded uplink signal. The method also determines a channel inversion coefficient based on a scaling factor obtained from the mapping and the calculated gradient Sum-Power level. The method applies analog modulation and the channel inversion coefficient to the data block to form the unencoded uplink signal. The method further transmits, to a network, the unencoded uplink signal on shared uplink resources for an over-the-air computation of global gradients for the federated learning task.

    WIRELESS COMMUNICATIONS BASED ON RATELESS CODES AND LAYERED MODULATION

    公开(公告)号:US20230396355A1

    公开(公告)日:2023-12-07

    申请号:US18249922

    申请日:2020-12-14

    CPC classification number: H04L1/0003 H04L1/0061 H04L1/0009

    Abstract: Methods, systems, and devices for wireless communications are described. A rateless code may be used to generate n encoded packets based on a set of source packets. A modulation technique may be used to generate first symbol packets from the encoded packets. And a layered modulation technique may be used to generate coded symbol packets from the first symbol packets. The layered modulation technique may include generating a coded symbol packet from a group of first symbol packets. Different weights may be applied to the different first symbol packets included in the group of first symbol packets before the first symbol packets in the group are combined together. A receiving device may demodulate the transmitted coded symbol packets to reconstruct the group of first symbol packets, use the reconstructed group of first symbol packets to recover the encoded packets, and use the encoded packets to reconstruct the source packets.

    CONSTELLATION SHAPING CONFIGURATION AND FEEDBACK

    公开(公告)号:US20230361913A1

    公开(公告)日:2023-11-09

    申请号:US18245144

    申请日:2020-11-18

    CPC classification number: H04L1/0042 H04L1/0075

    Abstract: Methods, systems, and devices for wireless communications are described. A user equipment (UE) may receive, from a base station, a configuration indicating a set of encoding parameters corresponding to a constellation shaping encoding process. The UE may calculate a number of bits of a first subset of a source bit stream, and the first subset of the source bit stream may include or encompass an input to a distribution matcher. The UE may apply, to the first subset, a distribution matcher parameter based on the set of encoding parameters. The UE may calculate a number of bits of a second subset of the source bit stream, and the second subset of the source bit stream may be concatenated with an output bit stream of the distribution matcher as an input to a channel encoder. The UE may signal the source bit stream to the base station.

Patent Agency Ranking