Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04W—WIRELESS COMMUNICATION NETWORKS
  • H04W28/00—Network traffic management; Network resource management
  • H04W28/02—Traffic management, e.g. flow control or congestion control
  • H04W28/06—Optimizing the usage of the radio link, e.g. header compression, information sizing, discarding information
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/045—Combinations of networks
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/045—Combinations of networks
  • G06N3/0455—Auto-encoder networks; Encoder-decoder networks
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/0464—Convolutional networks [CNN, ConvNet]
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/0495—Quantised networks; Sparse networks; Compressed networks
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T9/00—Image coding
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T9/00—Image coding
  • G06T9/002—Image coding using neural networks
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
  • H04L69/04—Protocols for data compression, e.g. ROHC
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04W—WIRELESS COMMUNICATION NETWORKS
  • H04W72/00—Local resource management
  • H04W72/20—Control channels or signalling for resource management
  • H04W72/23—Control channels or signalling for resource management in the downlink direction of a wireless link, i.e. towards a terminal
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/048—Activation functions
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04W—WIRELESS COMMUNICATION NETWORKS
  • H04W24/00—Supervisory, monitoring or testing arrangements
  • H04W24/08—Testing, supervising or monitoring using real traffic

Definitions

• the following relates to wireless communications, including encoding techniques for neural network architectures.
• Wireless communications systems are widely deployed to provide various types of communication content such as voice, video, packet data, messaging, broadcast, and so on. These systems may be capable of supporting communication with multiple users by sharing the available system resources (e.g., time, frequency, and power).
• Examples of such multiple- access systems include fourth generation (4G) systems such as Long Term Evolution (LTE) systems, LTE-Advanced (LTE-A) systems, or LTE-A Pro systems, and fifth generation (5G) systems which may be referred to as New Radio (NR) systems.
• 4G systems such as Long Term Evolution (LTE) systems, LTE-Advanced (LTE-A) systems, or LTE-A Pro systems
• 5G systems which may be referred to as New Radio (NR) systems.
• the differential encoding operation may include an encoding of an amount of data of the compressed dataset (e.g., after the additional encoding operation) based on previous values for the amount of data, such as an initial value, an initial reconstructed value, previous reconstructed values, previous values for a same data from a previous time instance, etc.
• the apparatus may include means for receiving an indication of one or more encoding operations to use for encoding a compressed dataset, the one or more encoding operations including a differential encoding operation or an entropy encoding operation or both; means for encoding, by a neural network, a dataset to generate the compressed dataset; means for quantizing the compressed dataset encoded by the neural network; means for encoding the quantized and compressed dataset based on receiving the indication of the one or more encoding operations; and means for transmitting the encoded, quantized, and compressed dataset to a second device after encoding the compressed dataset based on the one or more encoding operations.
• encoding the quantized and compressed dataset may include operations, features, means, or instructions for determining, after quantizing the compressed dataset encoded by the neural network, a differential value between a first reconstruction value of a data in the quantized and compressed dataset at a first time instance and a second reconstruction value of the data at a second time instance after the first time instance, where the differential value may be determined based on the indication of the one or more encoding operations and the quantized and compressed dataset may be encoded based on the differential value.
• the entropy encoding operation may include an encoding of the compressed dataset using one or more symbols having lengths that vary based on a probability that a symbol occurs.
• a non-transitory computer-readable medium storing code for wireless communications at a device is described.
• the code may include instructions executable by a processor to transmit, to a UE, an indication of one or more encoding operations for the UE to use for encoding a compressed dataset, the one or more encoding operations including a differential encoding operation or an entropy encoding operation or both; receive, from the UE, an encoded, quantized, and compressed dataset after the compressed dataset has been encoded following a quantization operation based on the one or more encoding operations; decode, based on the one or more encoding operations, the encoded, quantized, and compressed dataset to generate a compressed dataset; and decode, by a neural network, the compressed dataset to generate a dataset based on decoding the encoded, quantized, and compressed dataset.
• Time intervals of a communications resource may be organized according to radio frames each having a specified duration (e.g., 10 milliseconds (ms)). Each radio frame may be identified by a system frame number (SFN) (e.g., ranging from 0 to 1023).
• SFN system frame number
• a subframe, a slot, a mini-slot, or a symbol may be the smallest scheduling unit (e.g., in the time domain) of the wireless communications system 100 and may be referred to as a transmission time interval (TTI).
• TTI duration e.g., the quantity of symbol periods in a TTI
• the smallest scheduling unit of the wireless communications system 100 may be dynamically selected (e.g., in bursts of shortened TTIs (sTTIs)).
• a carrier may support multiple cells, and different cells may be configured according to different protocol types (e.g., MTC, narrowband IoT (NB-IoT), enhanced mobile broadband (eMBB)) that may provide access for different types of devices.
• protocol types e.g., MTC, narrowband IoT (NB-IoT), enhanced mobile broadband (eMBB)
• Some UEs 115 may be low cost or low complexity devices and may provide for automated communication between machines (e.g., via Machine-to-Machine (M2M) communication).
• M2M communication or MTC may refer to data communication technologies that allow devices to communicate with one another or a base station 105 without human intervention.
• M2M communication or MTC may include communications from devices that integrate sensors or meters to measure or capture information and relay such information to a central server or application program that makes use of the information or presents the information to humans interacting with the application program.
• Some UEs 115 may be designed to collect information or enable automated behavior of machines or other devices. Examples of applications for MTC devices include smart metering, inventory monitoring, water level monitoring, equipment monitoring, healthcare monitoring, wildlife monitoring, weather and geological event monitoring, fleet management and tracking, remote security sensing, physical access control, and transaction- based business charging.
• groups of the UEs 115 communicating via D2D communications may utilize a one-to-many (1 :M) system in which each UE 115 transmits to every other UE 115 in the group.
• a base station 105 facilitates the scheduling of resources for D2D communications.
• D2D communications are carried out between the UEs 115 without the involvement of a base station 105.
• the D2D communication link 135 may be an example of a communication channel, such as a sidelink communication channel, between vehicles (e.g., UEs 115).
• vehicles may communicate using vehicle-to-everything (V2X) communications, vehicle-to-vehicle (V2V) communications, or some combination of these.
• V2X vehicle-to-everything
• V2V vehicle-to-vehicle
• a vehicle may signal information related to traffic conditions, signal scheduling, weather, safety, emergencies, or any other information relevant to a V2X system.
• vehicles in a V2X system may communicate with roadside infrastructure, such as roadside units, or with the network via one or more network nodes (e.g., base stations 105) using vehicle-to-network (V2N) communications, or with both.
• V2N vehicle-to-network
• the control plane entity may manage non-access stratum (NAS) functions such as mobility, authentication, and bearer management for the UEs 115 served by the base stations 105 associated with the core network 130.
• NAS non-access stratum
• User IP packets may be transferred through the user plane entity, which may provide IP address allocation as well as other functions.
• the user plane entity may be connected to IP services 150 for one or more network operators.
• the IP services 150 may include access to the Internet, Intranet(s), an IP Multimedia Subsystem (IMS), or a Packet- Switched Streaming Service.
• IMS IP Multimedia Subsystem
• Packet- Switched Streaming Service Packet- Switched Streaming Service
• the wireless communications system 100 may operate using one or more frequency bands, typically in the range of 300 megahertz (MHz) to 300 gigahertz (GHz).
• the region from 300 MHz to 3 GHz is known as the ultra-high frequency (UHF) region or decimeter band because the wavelengths range from approximately one decimeter to one meter in length.
• UHF waves may be blocked or redirected by buildings and environmental features, but the waves may penetrate structures sufficiently for a macro cell to provide service to the UEs 115 located indoors.
• the transmission of UHF waves may be associated with smaller antennas and shorter ranges (e.g., less than 100 kilometers) compared to transmission using the smaller frequencies and longer waves of the high frequency (HF) or very high frequency (VHF) portion of the spectrum below 300 MHz.
• HF high frequency
• VHF very high frequency
• the wireless communications system 100 may also operate in a super high frequency (SHF) region using frequency bands from 3 GHz to 30 GHz, also known as the centimeter band, or in an extremely high frequency (EHF) region of the spectrum (e.g., from 30 GHz to 300 GHz), also known as the millimeter band.
• SHF super high frequency
• EHF extremely high frequency
• the wireless communications system 100 may support millimeter wave (mmW) communications between the UEs 115 and the base stations 105, and EHF antennas of the respective devices may be smaller and more closely spaced than UHF antennas. In some examples, this may facilitate use of antenna arrays within a device.
• mmW millimeter wave
• the propagation of EHF transmissions may be subject to even greater atmospheric attenuation and shorter range than SHF or UHF transmissions.
• the techniques disclosed herein may be employed across transmissions that use one or more different frequency regions, and designated use of bands across these frequency regions may differ by country or regulating body.
• the wireless communications system 100 may utilize both licensed and unlicensed radio frequency spectrum bands.
• the wireless communications system 100 may employ License Assisted Access (LAA), LTE-Unlicensed (LTE-U) radio access technology, or NR technology in an unlicensed band such as the 5 GHz industrial, scientific, and medical (ISM) band.
• LAA License Assisted Access
• LTE-U LTE-Unlicensed
• NR NR technology
• an unlicensed band such as the 5 GHz industrial, scientific, and medical (ISM) band.
• devices such as the base stations 105 and the UEs 115 may employ carrier sensing for collision detection and avoidance.
• operations in unlicensed bands may be based on a carrier aggregation configuration in conjunction with component carriers operating in a licensed band (e.g., LAA).
• Operations in unlicensed spectrum may include downlink transmissions, uplink transmissions, P2P transmissions, or D2D transmissions, among other examples.
• a base station 105 or a UE 115 may be equipped with multiple antennas, which may be used to employ techniques such as transmit diversity, receive diversity, multiple-input multiple-output (MIMO) communications, or beamforming.
• the antennas of a base station 105 or a UE 115 may be located within one or more antenna arrays or antenna panels, which may support MIMO operations or transmit or receive beamforming.
• one or more base station antennas or antenna arrays may be co-located at an antenna assembly, such as an antenna tower.
• antennas or antenna arrays associated with a base station 105 may be located in diverse geographic locations.
• a base station 105 may have an antenna array with a quantity of rows and columns of antenna ports that the base station 105 may use to support beamforming of communications with a UE 115.
• a UE 115 may have one or more antenna arrays that may support various MIMO or beamforming operations.
• an antenna panel may support radio frequency beamforming for a signal transmitted via an antenna port.
• Beamforming which may also be referred to as spatial filtering, directional transmission, or directional reception, is a signal processing technique that may be used at a transmitting device or a receiving device (e.g., a base station 105, a UE 115) to shape or steer an antenna beam (e.g., a transmit beam, a receive beam) along a spatial path between the transmitting device and the receiving device.
• Beamforming may be achieved by combining the signals communicated via antenna elements of an antenna array such that some signals propagating at particular orientations with respect to an antenna array experience constructive interference while others experience destructive interference.
• the adjustment of signals communicated via the antenna elements may include a transmitting device or a receiving device applying amplitude offsets, phase offsets, or both to signals carried via the antenna elements associated with the device.
• the adjustments associated with each of the antenna elements may be defined by a beamforming weight set associated with a particular orientation (e.g., with respect to the antenna array of the transmitting device or receiving device, or with respect to some other orientation).
• a base station 105 or a UE 115 may use beam sweeping techniques as part of beam forming operations.
• a base station 105 may use multiple antennas or antenna arrays (e.g., antenna panels) to conduct beamforming operations for directional communications with aUE 115.
• Some signals e.g., synchronization signals, reference signals, beam selection signals, or other control signals
• the base station 105 may transmit a signal according to different beamforming weight sets associated with different directions of transmission.
• the Radio Resource Control (RRC) protocol layer may provide establishment, configuration, and maintenance of an RRC connection between a UE 115 and a base station 105 or a core network 130 supporting radio bearers for user plane data.
• RRC Radio Resource Control
• transport channels may be mapped to physical channels.
• a device may support same-slot HARQ feedback, where the device may provide HARQ feedback in a specific slot for data received in a previous symbol in the slot. In other cases, the device may provide HARQ feedback in a subsequent slot, or according to some other time interval.
• Machine learning has become a popular tool in wireless communications systems to promote more efficient communications.
• machine learning models deployed at a UE 115 may enable the UE 115 to make decisions or perform actions (e.g., using prediction or regression or other targets) without additional signaling from a base station 105 (e.g., the UE 115 can make an inference about an action to perform based on an input or detected event).
• the machine learning models may enable the UE 115 to prepare transmissions for more efficient communications (e.g., using classification or compression or other targets). Before the machine learning models can be deployed on devices, the machine learning models may be prepared and trained (e.g., a using a dataset).
• Machine learning models using neural networks may be deployed on devices (e.g., UEs 115) for different applications (e.g., prediction, classification, compression, regression, or other targets).
• the machine learning model may include one or more parameters (e.g., prepared datasets) that, when identified by the devices, enable or support a corresponding application at the devices.
• a network device may collect different datasets to identify implications or effects of the datasets for the different applications. That is, a machine learning model may be trained on one or more datasets using the neural networks, and when real-world data is input to the machine learning model, the machine learning model may generate an output based on the dataset.
• the quantities measured by the UE 115 may be dependent on multiple parameters.
• the multiple parameters that can affect the measurements may include antenna design and placement measurements at the UE 115 and their time-varying blockage, environment parameters (e.g., location, shadowing, presence and movement of reflectors in a neighborhood of the UE 115 and/or the base station 105, etc., where the placement of reflectors can also give rise to inter-tap correlation, such as a resolution of broad paths/beams being split into many multiple paths), a loading on a cell (e.g., resulting in handoffs), a movement of the UE 115 in question (e.g., a change in its orientation), etc.
• the UE 115 may use or may be implemented by neural networks that learn the dependence of the measured quantities on individual parameters, isolate those measured quantities through various layers, and compress the measurements, while reducing (e.g., minimizing) the compression loss. For example, the UE 115 may compress the measurements using neural networks to reduce a size of a transmission carrying the measurements.
• Wireless communications system 100 may support adding additional layers to a compression operation when a UE 115 is compressing data in a transmission prior to sending the transmission to a base station 105.
• the UE 115 may measure one or more channel conditions and then use a neural network to compress the measurement data.
• the UE 115 may quantize the output of the neural network to provide data that is more capable of being communicated.
• the UE 115 may add the additional layers to further compress the data, where the UE 115 may use a differential encoding, entropy encoding, or both to further compress and encode the measurement data prior to sending the transmission to the base station.
• the differential encoding may be example of encoding differences between two measurements, rather than encoding absolute values of measurements.
• the differential encoding may use previous values, initial values, reconstruction values, initial reconstruction values, or a combination thereof to indicate differential values.
• the base station 105 may then use a differential decoding, an entropy decoding, or both when decoding the encoded, quantized, and compressed transmission from the UE 115.
• the base station 105 may transmit indications of the differential encoding and the entropy encoding (e.g., and parameters for each encoding) for the UE 115 to use.
• FIG. 2 illustrates an example of a wireless communications system 200 that supports encoding techniques for neural network architectures in accordance with aspects of the present disclosure.
• wireless communications system 200 may implement aspects of or may be implemented by aspects of wireless communications system 100.
• wireless communications system 200 may include a base station 105-a and a UE 115-a, which may be examples of corresponding base stations 105 and UEs 115, respectively, as described with reference to FIG. 1.
• base station 105-a and UE 115-a may communicate on resources of a carrier 205 (e.g., for downlink communications) and of a carrier 210 (e.g., for uplink communications).
• carrier 205 and carrier 210 may include same or different resources (e.g., time and frequency resources) for the corresponding transmissions.
• UE 115-a may then perform an extraction that contains additional stacked layers.
• the additional stacked layers may include convolution layers, fully connected layers, or other layers with or without activations (e.g., residual neural network (ResNet) layers).
• a neural network employed on UE 115-a may compress the feature.
• the neural network may use a convolutional or fully connected layer or another type of layer for this compression.
• UE 115-a may repeat this process for subsequent features.
• UE 115-a may use one or more additional compression layers (e.g., convolutional, fully connected, or another type of layer).
• additional compression layers e.g., convolutional, fully connected, or another type of layer.
• UE 115-a may then perform additional encoding for a final compression, such as a differential encoding, an entropy encoding, or both.
• the encoding parameters 215 may indicate how UE 115-a is to perform a corresponding encoder operation.
• base station 105-a may indicate which previous values UE 115-a is to use to perform the differential encoder operation.
• UE 115-a may use an initial value for a data (e.g., at an intra-coded frame (I-frame)) to determine a differential value for that data at a later time instance (e.g., at predictive frames (P-frames)). Additionally or alternatively, UE 115-a may use previous reconstructed values for a data to determine a differential value for that data at a later time instance.
• I-frame intra-coded frame
• P-frames predictive frames
• any encoding device e.g., a base station 105, a TRP, another type of UE 115, etc.
• any decoding device may perform the techniques described herein to compress and encode a dataset before transmitting the dataset to an additional device.
• base station 105-a is shown receiving and decoding compressed and encoded dataset 235 (e.g., an encoded, quantized, and compressed dataset)
• any decoding device e.g., a UE 115, a TRP, another type of base station 105, etc.
• FIGs. 3A and 3B illustrate examples of a compression procedure 300 and a compression procedure 301, respectively, that support encoding techniques for neural network architectures in accordance with aspects of the present disclosure.
• compression procedure 300 and compression procedure 301 may implement aspects of or may be implemented by aspects of wireless communications system 100, wireless communications system 200, or both.
• an encoding device e.g., a UE 115 or an additional encoding device
• samples e.g., data
• a decoding device e.g., a base station 105 or an additional decoding device
• the encoding device may compress one or more features that have been extracted using a feature compression 320.
• feature compression 320 e.g., a compression operation
• a bit count of an output may be less than a bit count of an input.
• the decoding device may perform operations in an order that is opposite to operations performed by the encoding device. For example, if the encoding device follows operations (A, B, C, D ), the decoding device may follow inverse operations ( D , C, B, A). Additionally, the decoding device may perform operations that are fully symmetric to operations of the encoding device. This use of symmetric operations may reduce a quantity of bits used for neural network configuration at the encoding device. Additionally or alternatively, the decoding device may perform additional operations (e.g., convolution operations, fully connected operation, ResNet operations, etc.) in addition to operations performed by the encoding device. That is, the decoding device may perform operations that are asymmetric to operations of the encoding device.
• additional operations e.g., convolution operations, fully connected operation, ResNet operations, etc.
• the encoding device may transmit measurements (e.g., channel state feedback) with a reduced payload.
• measurements e.g., channel state feedback
• This reduced payload may conserve network resources that may otherwise have been used to transmit a full data set as sampled by the encoding device.
• the encoding device may use additional layers to perform additional encoding operations to further compress data to be transmitted to the decoding device.
• the encoding device may apply or use a differential encoder 360, an entropy encoder 365, or both on an output of a single shot encoder to further compress a feedback 370 transmitted to the decoding device.
• the feedback 370 may be an example of the information transmitted over an air interface (or some other medium) between a transmitting device and a receiving device.
• the decoding device may also use an entropy decoder 375, a differential decoder 380, or both before performing the operations of a single shot decoder described with reference to compression procedure 300 in FIG. 3A.
• the encoding device may take input 305 and put input 305 through an encoder neural network 350.
• encoder neural network 350 may correspond to spatial feature extraction 310, tap domain feature extraction 315 and feature compression 320 as described with reference to FIG. 3 A (e.g., the single shot encoder).
• the encoding device may use a quantizer 355 to perform a quantization operation (e.g., quantization 325 as described with reference to FIG. 3 A). Subsequently, the encoding device may then use or apply differential encoder 360, entropy encoder 365, or both to the quantized output of encoder neural network 350.
• the encoding device may perform one or more ResNet operations using one or more corresponding ResNet blocks, such as a ResNet block 415 and a ResNet block 420.
• the one or more ResNet operations may further refine the spatial feature and/or the temporal feature.
• a ResNet operation may include multiple operations associated with a feature.
• a ResNet operation may include multiple (e.g., 3) one-dimensional convolution operations, a skip connection (e.g., between input of the ResNet and output of the ResNet to avoid application of the one dimensional convolution operations), a summation operation of a path through the multiple one-dimensional convolution operations and a path through the skip connection, or additional operations.
• the multipleonel-dimensinoal convolution operations may include a Wx256 convolution operation with kernel size 3 with output that is input to a batch normalization (BN) layer followed by a rectified linear unit (ReLU) activation (e.g., such as a LeakyReLU activation) that produces an output data set of dimension 256x64, a 256x512 convolution operation with kernel size 3 with output that is input to a BN layer followed by an ReLU activation that produces an output data set of dimension 512x64, and a 512xW convolution operation with kernel size 3 that outputs a BN data set of dimension Wx64.
• Output from the one or more ResNet operations may be a Wx64 matrix.
• the encoding device may use an additional convolutional layer 425 to perform a WxV convolution operation on output from the one or more ResNet operations.
• the WxV convolution operation may include a pointwise (e.g., tap-wise) convolution operation.
• the WxV convolution operation may compress spatial features into a reduced dimension for each tap.
• the WxV convolution operation may have an input of W features and an output of V features.
• Output from the WxV convolution operation may be a Vx64 matrix.
• a decoding device e.g., a base station 105 or an additional decoding device
• compression and encoding configuration 500 may represent a compression and encoding procedure that includes a differential encoder 510, an entropy encoder 515, or both at an output of an encoder (e.g., single shot encoder, encoder neural network, etc.).
• differential encoder 510 By using or applying differential encoder 510 at the output of the encoder, the encoding device may decrease or limit error propagation.
• E l /D l may be multiplied by ⁇ E ⁇ c ⁇ ⁇ 2 ⁇ to determine a reconstructed value for the initial value for the data, where the reconstructed value is given by x 0.
• the encoding device may determine x 0 by performing similar actions as the decoding device or may receive an indication of the reconstructed value from the decoding device.
• the encoding device may use a reconstructed value for a previous value to encode differential values for values at the given P-frame. For example, for a P-frame N-l (e.g., at a time N-l), the encoding device may determine a differential value for a data at time N-l based on Equation 3.
• the encoding device may then perform similar operations as described previously (e.g., divide the differential value for a data given by Equation 3 by ⁇ JE ⁇ A ⁇ 2 ⁇ ).
• the decoding device may receive a representation of a data given by E P /D P and may multiply the received representation of the data (e.g., encoded version of the data) by ⁇ JE ⁇ A ⁇ 2 ⁇ to generate a reconstructed version of Equation 3 (e.g., (x W -i — x w _ 2 )) ⁇
• the decoding device may then add a reconstructed value for a previous value of the data (e.g., x /v-2 ) determined by the decoding device to determine and obtain a reconstructed value for the given value of the data at time N-l (e.g., x «_i).
• the P-frames may be constructed based on x 0 rather than an immediately preceding reconstruction value for a data.
• the decoding device may receive an encoded, quantized, and compressed dataset from the encoding device and may use an entropy decoder 520 and a differential decoder 525 to partially decode the encoded, quantized, and compressed dataset. Additionally, the decoding device may use one or more fully connected layers, one or more convolutional layers, and one or more ResNet blocks as described with reference to FIG. 3 A to determine and obtain output 530 corresponding to input 505.
• FIG. 6 illustrates an example of a compression and encoding configuration 600 that supports encoding techniques for neural network architectures in accordance with aspects of the present disclosure.
• compression and encoding configuration 600 may implement aspects of or may be implemented by aspects of wireless communications system 100, wireless communications system 200, or both.
• an encoding device e.g., a UE 115 or an additional encoding device
• UE 115-b may receive an indication of one or more encoding operations to use for encoding a compressed dataset, the one or more encoding operations including a differential encoding operation or an entropy encoding operation or both.
• UE 115-b may receive one or more parameters corresponding to the one or more encoding operations, each of the one or more parameters corresponding to a respective encoding operation of the one or more encoding operations, where the quantized and compressed dataset is encoded based on the one or more parameters.
• UE 115-b may encode the quantized and compressed dataset based on receiving the indication of the one or more encoding operations.
• UE 115-b may encode the quantized and compressed dataset using the entropy encoding operation.
• the entropy encoding operation may include an encoding of the compressed dataset using one or more symbols having lengths that vary based on a probability that a symbol occurs.
• FIG. 9 shows a block diagram 900 of a device 905 that supports encoding techniques for neural network architectures in accordance with aspects of the present disclosure.
• the device 905 may be an example of aspects of a device 805 or a UE 115 as described herein.
• the device 905 may include a receiver 910, a transmitter 915, and a communications manager 920.
• the device 905 may also include a processor. Each of these components may be in communication with one another (e.g., via one or more buses).
• the differential decoder component 1445 may be configured as or otherwise support a means for receiving the encoded, quantized, and compressed dataset including differential values for data in the dataset that are based on previous reconstruction values for the data.
• a method for wireless communications at a UE comprising: receiving an indication of one or more encoding operations to use for encoding a compressed dataset, the one or more encoding operations comprising a differential encoding operation or an entropy encoding operation or both; encoding, by a neural network, a dataset to generate the compressed dataset; quantizing the compressed dataset encoded by the neural network; encoding the quantized and compressed dataset based at least in part on receiving the indication of the one or more encoding operations; and transmitting the encoded, quantized, and compressed dataset to a second device after encoding the compressed dataset based at least in part on the one or more encoding operations.
• Aspect 3 The method of any of aspects 1 through 2, wherein encoding the quantized and compressed dataset comprises: encoding the quantized and compressed dataset using the differential encoding operation after encoding the dataset using the neural network.
• Aspect 7 The method of any of aspects 1 through 6, wherein encoding the quantized and compressed dataset comprises: determining, after quantizing the compressed dataset encoded by the neural network, an initial reconstruction value for a data in the quantized and compressed dataset at an initial time instance associated with encoding the dataset; and determining, after the quantizing, a differential value between an additional reconstruction value of the data at an additional time instance after the initial time instance and the initial reconstruction value for the data, wherein the differential value is determined based at least in part on the indication of the one or more encoding operations and the quantized and compressed dataset is encoded based at least in part on the differential value.
• Aspect 8 The method of any of aspects 1 through 7, wherein the differential encoding operation comprises an encoding of an amount of data of the compressed dataset based at least in part on previous values for the amount of data.
• Aspect 11 The method of aspect 10, wherein transmitting the indication of the one or more encoding operations comprises: transmitting one or more parameters corresponding to the one or more encoding operations, each of the one or more parameters corresponding to a respective encoding operation of the one or more encoding operations, wherein the quantized and compressed dataset is encoded based at least in part on the one or more parameters.
• Aspect 12 The method of any of aspects 10 through 11, wherein decoding the encoded, quantized, and compressed dataset comprises: decoding the encoded, quantized, and compressed dataset using a differential decoding operation before decoding the dataset using the neural network.
• Aspect 14 The method of any of aspects 10 through 13, wherein receiving the encoded, quantized, and compressed dataset comprises: receiving the encoded, quantized, and compressed dataset comprising differential values for data in the dataset that are based at least in part on initial values for the data.
• Aspect 16 The method of any of aspects 10 through 15, wherein receiving the encoded, quantized, and compressed dataset comprises: receiving the encoded, quantized, and compressed dataset comprising differential values for data in the dataset that are based at least in part on initial reconstruction values for the data.
• Aspect 17 The method of any of aspects 10 through 16, wherein the differential decoding operation comprises an encoding of an amount of data of the compressed dataset based at least in part on previous values for the amount of data.
• Aspect 18 The method of any of aspects 10 through 17, wherein the entropy decoding operation comprises an encoding of the compressed dataset using one or more symbols having lengths that vary based at least in part on a probability that a symbol occurs.
• Aspect 19 An apparatus for wireless communications at a UE, comprising a processor; memory coupled with the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform a method of any of aspects 1 through 9.
• Aspect 20 An apparatus for wireless communications at a UE, comprising at least one means for performing a method of any of aspects 1 through 9.
• Aspect 21 A non-transitory computer-readable medium storing code for wireless communications at a UE, the code comprising instructions executable by a processor to perform a method of any of aspects 1 through 9.
• Aspect 22 An apparatus for wireless communications at a device, comprising a processor; memory coupled with the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform a method of any of aspects 10 through 18.
• Aspect 23 An apparatus for wireless communications at a device, comprising at least one means for performing a method of any of aspects 10 through 18.
• LTE, LTE-A, LTE-A Pro, or NR may be described for purposes of example, and LTE, LTE-A, LTE-A Pro, or NR terminology may be used in much of the description, the techniques described herein are applicable beyond LTE, LTE-A, LTE-A Pro, or NR networks.
• the described techniques may be applicable to various other wireless communications systems such as Ultra Mobile Broadband (UMB), Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20, Flash-OFDM, as well as other systems and radio technologies not explicitly mentioned herein.
• UMB Ultra Mobile Broadband
• IEEE Institute of Electrical and Electronics Engineers
• Wi-Fi Wi-Fi
• WiMAX IEEE 802.16
• IEEE 802.20 Flash-OFDM
• non-transitory computer-readable media may include RAM, ROM, electrically erasable programmable ROM (EEPROM), flash memory, compact disk (CD) ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that may be used to carry or store desired program code means in the form of instructions or data structures and that may be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Also, any connection is properly termed a computer-readable medium.
• RAM random access memory
• ROM read only memory
• EEPROM electrically erasable programmable ROM
• CD compact disk
• magnetic disk storage or other magnetic storage devices or any other non-transitory medium that may be used to carry or store desired program code means in the form of instructions or data structures and that may be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor.
• any connection is properly termed a computer-readable medium.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Artificial Intelligence (AREA)
• Evolutionary Computation (AREA)
• General Health & Medical Sciences (AREA)
• General Engineering & Computer Science (AREA)
• Biophysics (AREA)
• Computational Linguistics (AREA)
• Data Mining & Analysis (AREA)
• Life Sciences & Earth Sciences (AREA)
• Health & Medical Sciences (AREA)
• Molecular Biology (AREA)
• Computing Systems (AREA)
• Biomedical Technology (AREA)
• Software Systems (AREA)
• Mathematical Physics (AREA)
• Signal Processing (AREA)
• Computer Networks & Wireless Communication (AREA)
• Multimedia (AREA)
• Computer Security & Cryptography (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)
• Mobile Radio Communication Systems (AREA)

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=80445982

Family Applications (1)

Country Status (8)

Families Citing this family (7)

Citations (1)

Family Cites Families (4)

• 2021
  • 2021-03-05 US US17/194,077 patent/US12361277B2/en active Active
• 2022
  • 2022-02-04 BR BR112023017266A patent/BR112023017266A2/pt unknown
  • 2022-02-04 JP JP2023544730A patent/JP2024509502A/ja active Pending
  • 2022-02-04 CN CN202280017590.1A patent/CN116917902A/zh active Pending
  • 2022-02-04 EP EP22705667.8A patent/EP4302232A1/en active Pending
  • 2022-02-04 KR KR1020237029435A patent/KR20230154014A/ko active Pending
  • 2022-02-04 WO PCT/US2022/015285 patent/WO2022186947A1/en not_active Ceased
  • 2022-02-07 TW TW111104256A patent/TW202239152A/zh unknown

Patent Citations (1)

Also Published As

Similar Documents

Legal Events