Abstract: The disclosure provides for an apparatus for wireless communications using carrier aggregation comprised of multiple carrier components. The apparatus can include a processor configured to generate one or more instances of a codeword from a data payload. In an aspect, the apparatus also includes a modulator configured to modulate the one or more instances of the codeword onto the multiple carrier components for transmission. In an aspect, the apparatus also includes a resource manager configured to provide the processor with a virtual carrier space comprising a logical carrier having a contiguous bandwidth equivalent to the aggregated bandwidth of the multiple carrier components. In an aspect, the process may be further configured to interleave at least one of the codeword instances across the multiple carrier components. In an aspect, the modulator may be configured to modulate the codeword instance onto the multiple carrier components in accordance to the interleaving.

Abstract: A method for operating a low-bitwidth neural network includes converting a first activation to a non-negative value (e.g., absolute value). The first activation has a signed value. The sign of the activation is used to select a weight value. A product of the non-negative activation and the selected weight value is computed to determine a next activation. The next activation is quantized and supplied to a subsequent layer of the low-bitwidth neural network.

Abstract: Methods, systems, and devices for wireless communications are described. A wireless device, such as a user equipment (UE) may monitor for a decoding candidate of a codeword, wherein the codeword corresponds to a set of received bit metrics, and the decoding candidate corresponds to a plurality of information bits encoded using a polar code, determine a composite detection metric for the codeword for the decoding candidate, where the composite detection metric is derived from a subset of bit metrics for an intermediate polarization layer of the polar code, and determine a classification for performing a list decoding process on the codeword according to the decoding candidate based at least in part on the composite detection metric.

Abstract: An apparatus may utilize an air interface to transmit and/or receive a transmission during a first TTI that includes a second set of data overriding a first set of data scheduled for transmission during the first TTI. The air interface may further be utilized to transmit and/or receive a transmission during one or more additional TTIs that includes a third set of data in a data portion of a subframe, and an override indicator that is at least partially embedded in the data portion. The override indicator may be configured to indicate that the first set of data scheduled for transmission during the first TTI is overridden by the second set of data having the higher priority. The override indicator may be transmitted after the second set of data is transmitted. The one or more additional TTIs may be after the first TTI.

Abstract: Methods, systems, and devices for wireless communication employing polar code techniques are described. In some examples, polar codes with a cross-referenceable nested structure for hierarchical signaling may be used in a wireless communication system. For example, a user equipment may monitor a common portion and a dedicated portion of a control channel, the common portion assigned to a first set of control channel resources and the dedicated portion assigned to a second set of control channel resources within the control channel. The UE may concatenate, based at least in part on the monitoring, information fields of one or more symbols received via the first and second sets of control channel resources to form a polar-encoded codeword. The UE may decode the polar-encoded codeword to obtain common information from the common portion and dedicated information from the dedicated portion.

Abstract: Aspects described herein provide a method including: receiving input data at a machine learning model, comprising: a plurality of processing layers; a plurality of gate logics; a plurality of gates; and a fully connected layer; determining based on a plurality of gate parameters associated with the plurality of gate logics, a subset of the plurality of processing layers with which to process the input data; processing the input data with the subset of the plurality of processing layers and the fully connected layer to generate an inference; determining a prediction loss based on the inference and a training label associated with the input data; determining an energy loss based on the subset of the plurality of processing layers used to process the input data; and optimizing the machine learning model based on: the prediction loss; the energy loss; and a prior probability associated with the training label.

Abstract: Aspects described herein provide a method of performing phase selective convolution, including: receiving multi-phase pre-activation activation data; partitioning the multi-phase pre-activation data; applying a first activation function to the set of first phase pre-activation data to form a set of first phase activation output; convolving the set of first phase activation output with a first convolution kernel to form a first phase output feature map; negating the set of second phase activation data; applying a second activation function to the negated set of second phase pre-activation data to form a set of second phase activation output; convolving the set of second phase activation output with a second convolution kernel to form a second phase output feature map; negating the second phase output feature map; and training the neural network based on the first phase output feature map and the second phase output feature map.

Abstract: Methods, systems, and devices for configuring a machine learning network are described. A device, which may be otherwise known as user equipment (UE), may support ultra-low power sensor applications. More specifically, the device may support memory block allocation of a machine learning network based on performance levels associated with the applications. For example, the device may identify a performance level associated with an application on the device. The device may determine that the performance level satisfies a condition, and subsequently determine a memory block allocation of a machine learning network of the device based on the performance level satisfying the condition. The memory block allocation may correspond to one or more connections of the machine learning network. Based on the memory block allocation, the device may adjust a quantity of memory blocks available for the machine learning network and process the application.

Abstract: Transmission of user equipment (UE) specific control information within a resource allocation including resource blocks allocated for downlink transmissions to the UE is disclosed. Common control information may be provided in a first transmission time interval (TTI)-level control region, and UE-specific control information, specific to a particular UE, may be provided along with data in allocated downlink resources to the UE. A base station may identify a resource block (RB) for transmission of data to a UE along with UE-specific control information to be included in the RB. The control information may include, for example, parameters for use by the receiver in demodulating the RB. The base station may multiplex the control information with the data within the RB and transmit the RB and control information.

Abstract: Embodiments described herein relate to a method, comprising: receiving input data at a convolutional neural network (CNN) model; generating a factorized computation network comprising a plurality of connections between a first layer of the CNN model and a second layer of the CNN model, wherein: the factorized computation network comprises N inputs, the factorized computation network comprises M outputs, and the factorized computation network comprises at least one path from every input of the N inputs to every output of the M outputs; setting a connection weight for a plurality of connections in the factorized computation network to 1 so that a weight density for the factorized computation network is <100%; performing fast pointwise convolution using the factorized computation network to generate fast pointwise convolution output; and providing the fast pointwise convolution output to the second layer of the CNN model.

Abstract: Certain aspects of the present disclosure provide techniques for performing piecewise pointwise convolution, comprising: performing a first piecewise pointwise convolution on a first subset of data received via a first branch input at a piecewise pointwise convolution layer of a convolutional neural network (CNN) model; performing a second piecewise pointwise convolution on a second subset of data received via a second branch input at the piecewise pointwise convolution layer; determining a piecewise pointwise convolution output by summing a result of the first piecewise pointwise convolution and a result of the second piecewise pointwise convolution; and providing the piecewise pointwise convolution output to a second layer of the CNN model.

Abstract: Techniques are described for wireless communication. One method of wireless communication includes buffering a first set of coded bits including a first control message having a plurality of fields, buffering a second set of coded bits including a second control message having the plurality of fields, soft combining at least a first subset of the first set of coded bits and a second subset of the second set of coded bits in a combined set of coded bits, and decoding the first control message or the second control message based at least in part on the combined set of coded bits.

Abstract: Wireless communications systems and methods are introduced. A wireless communication device may arrange a first encoded information block including a first sub-block having a first bit location and a second sub-block having a second bit location. The second bit location is after the first bit location. The wireless communication device may also position the first location earlier in a decoding order of a receiving second wireless communication than the second bit location. The wireless communication device may transmit the first and second sub-blocks as an encoded information block to the second wireless communication device.

Abstract: A transmitter may select a control message format of a set of possible control message formats, each of the possible control message formats corresponding to a different number of information bits. The transmitter may polar encode a payload in the selected control message format to generate and transmit a polar-encoded codeword, the payload having a same number of bits for any of the set of possible control message formats. A receiver may determine the set of possible control message formats for the polar-encoded codeword, and may decode the polar-encoded codeword to identify a candidate control message. The receiver may identify a control message format in the set of possible control message formats for the candidate control message based on multiple hypotheses corresponding to the different number of information bits, and may obtain control information from the candidate control message based on the identified control message format.

Abstract: A wireless device may decode a polar coded codeword using a successive cancellation list (SCL) decoder. The decoder may implement a distributed feedback architecture, where the decoder stores one or more state maps and a set of bit arrays in memory for each layer of decoding. For different phases of decoding in a layer, the decoder may update the state maps and sets of bit arrays to limit the resources used. Additionally, when performing bit updating following the decoding of a bit of the codeword, the decoder may not update each layer of the decoding process. Instead, each sub-decoder may send a state map up to the calling layer for bit updating when the sub-decoder has completed its invocation, and may not return any intermediate state maps prior to completing invocation. Thus, each decoder and sub-decoder may perform bit updating just twice, reducing the complexity and latency of decoding.

Abstract: Aspects of the disclosure relate to wireless communication systems configured to provide piggyback downlink control information within the physical downlink shared channel (PDSCH). A first downlink control information portion may be transmitted within the physical downlink control channel (PDCCH) and may include information indicating a size of a second downlink control information portion transmitted within the PDSCH.

Abstract: Methods, systems, and devices for wireless communication are described. The examples described herein may enable a decoder to determine path metrics for various decoding paths based on identified frozen bit locations of a polar code. The path metric for a decoding path may be based on bit metrics determined for the identified frozen bit locations along the decoding path. Once the path metrics and bit metrics are determined, the decoder may compare these metrics to threshold criteria and determine whether to discard decoding paths based on the comparison. The techniques described herein for discarding decoding paths may allow the decoder to discard, prune, or disqualify certain decoding paths that are unlikely to provide an accurate representation of bits received from another device. Consequently, the decoder may be able to save power by terminating a decoding process early (i.e., early termination) if all paths are discarded, pruned, or disqualified.

Abstract: The various aspects include transmitting, by a transceiver of the wireless communications device, an indication of a primary service preference for a primary service and a secondary service preference for a secondary service, receiving, by the transceiver, a radio access technology support configuration from the network at least in part in response to the transmitted indication of the primary service preference and the secondary service preference, and supporting the primary service or the secondary service based at least in part on the channel layer configuration. In this manner, the wireless communications device provides the network with multiple ranked service preferences, so that if a network down selection is needed, the network may select a lower ranked service preference rather than generating a channel layer configuration without knowledge of which settings are most efficient for the wireless communications device.

Abstract: Methods, systems, and devices for wireless communications are described. In accordance with the described techniques, communicating devices (e.g., an encoder and decoder) may apply an orthogonal cover code to a polar codeword to reduce cross-correlation between different codewords. For example, such techniques may reduce power consumption at a decoding device by providing for earlier decoding termination (e.g., as a result of the reduced cross-correlation). Techniques for generating the cover codes (e.g., on a per-aggregation level basis) and applying the cover codes (e.g., within a search space) are described. Additionally or alternatively, the described techniques may relate to seeding of reference signals used to support decoding of the codewords. Improved orthogonality between reference signal seeds may further suppress codeword recipient ambiguity.

Abstract: Systems and methods are disclosed for minimizing latency between receipt of a NACK at a base station from a user equipment (UE) and retransmission of data to the UE. Time constraints for processing the ACK/NACK are relaxed so the base station can decode the ACK/NACK to determine whether a NACK has been received and then prepare for transmission of the appropriate data to the UE in the immediately following transmission time interval (TTI). These constraints are relaxed by separating download data indicator (DDI) from the PDCCH control data and delaying transmission of the DDI until decoding of the ACK/NACK.