# US 12,015,947 B2 · Generalized neural network architectures based on frequency and/or time division > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,015,947 B2 - Document ID: US12015947B2 - Status: Granted - Kind: B2 - Application number: 17/365,510 - Filed: 2021-07-01 - Published: 2023-01-05 - Granted: 2024-06-18 - Earliest claimed: 2021-07-01 - Assignees: QUALCOMM Incorporated - Applicants: QUALCOMM Incorporated - Inventors: Pavan Kumar Vitthaladevuni; Taesang Yoo; Naga Bhushan - Web page: https://patentgrep.com/patent/US12015947B2 ## CPC - H04W 28/06: Optimizing {the usage of the radio link}, e.g. header compression, information sizing {, discarding information (system modifying transmission characteristic according to link quality by modifying frame length H04L1/0007; dynamic adaptation of the packet size for flow control or congestion control H04L47/365)} - G06N 3/045: Combinations of networks - G06N 3/0455: Auto-encoder networks; Encoder-decoder networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/0495: Quantised networks; Sparse networks; Compressed networks - G06N 20/20: Ensemble learning - H03M 7/6011: {Encoder aspects} - H04W 24/10: Scheduling measurement reports {; Arrangements for measurement reports} - H03M 7/3059: {Digital compression and data reduction techniques where the original information is represented by a subset or similar information, e.g. lossy compression} - H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - H03M: CODING; DECODING; CODE CONVERSION IN GENERAL (using fluidic means F15C4/00; optical analogue/digital converters G02F7/00; coding, decoding or code conversion, specially adapted for particular applications, see the relevant subclasses, e.g. G01D, G01R, G06F, G06T, G09G, G10L, G11B, G11C, H04B, H04L, H04M, H04N; ciphering or deciphering for cryptography or other purposes involving the need for secrecy G09C) ## Abstract Certain aspects of the present disclosure provide techniques for measurement encoding and decoding using neural networks to compress and decompress measurement data. One example method generally includes: generating, via each of a plurality of neural network encoders operating on measurement data, a compressed measurement based on a respective portion of the measurement data, wherein each of the neural network encoders is based on the same neural network model; generating at least one message indicative of the measurement data based on the compressed measurements; and transmitting the at least one message. ## Claims ### Claim 1 An apparatus, comprising: a memory having executable instructions; and one or more processors configured to execute the executable instructions and cause the apparatus to: generate, via each of a plurality of neural network encoders operating on measurement data, a compressed measurement based on a respective portion of the measurement data, wherein each of the neural network encoders is based on the same neural network model, wherein the measurement data is generated using measurements performed for a bandwidth, and wherein the portion of the measurement data is associated with a subband of the bandwidth; generate at least one message indicative of the measurement data based on the compressed measurements; and transmit the at least one message. ### Claim 2 (depends on 1) The apparatus of claim 1, wherein the one or more processors are further configured to cause the apparatus to process the compressed measurements via another neural network model to generate further compressed measurements, wherein the at least one message indicates the further compressed measurements. ### Claim 3 (depends on 1) The apparatus of claim 1, wherein: the compressed measurements comprise a first compressed measurement and a second compressed measurement; the one or more processors are further configured to cause the apparatus to generate a first differential measurement to represent the second compressed measurement, the first differential measurement being generated based on a difference between the first compressed measurement and the second compressed measurement; and the at least one message is generated based on the first compressed measurement and the first differential measurement. ### Claim 4 (depends on 3) The apparatus of claim 3, wherein the first compressed measurement and the second compressed measurement are generated based on measurements performed on different subbands. ### Claim 5 (depends on 3) The apparatus of claim 3, wherein the first compressed measurement and the second compressed measurement are generated based on measurements performed at different intervals. ### Claim 6 (depends on 5) The apparatus of claim 5, wherein: the compressed measurements further include a third compressed measurement; the one or more processors are further configured to cause the apparatus to generate a second differential measurement to represent the third compressed measurement, the second differential measurement being generated based on a difference between the first compressed measurement and the third compressed measurement; the first compressed measurement and the third compressed measurement are generated based on measurements performed on different subbands; and the at least one message is generated based on the first compressed measurement, the first differential measurement, and the second differential measurement. ### Claim 7 (depends on 6) The apparatus of claim 6, wherein: the compressed measurements further include a fourth compressed measurement; the one or more processors are further configured to cause the apparatus to generate a third differential measurement for the fourth compressed measurement, the third differential measurement being generated as a function of the first differential measurement and the second differential measurement; and the at least one message is generated based on the first compressed measurement, the first differential measurement, the second differential measurement, and the third differential measurement. ### Claim 8 (depends on 7) The apparatus of claim 7, wherein the one or more processors are further configured to cause the apparatus to receive an indication of the function. ### Claim 9 (depends on 3) The apparatus of claim 3, wherein: the compressed measurements further include a third compressed measurement; the one or more processors are further configured to cause the apparatus to generate a second differential measurement for the third compressed measurement, the second differential measurement being generated based on a difference between the first compressed measurement and the third compressed measurement; and the at least one message is generated based on the first compressed measurement, the first differential measurement, and the second differential measurement. ### Claim 10 (depends on 3) The apparatus of claim 3, wherein: the compressed measurements further include a third compressed measurement; the one or more processors are further configured to cause the apparatus to generate a second differential measurement for the third compressed measurement, the second differential measurement being generated based on a difference between the second compressed measurement and the third compressed measurement; and the at least one message is generated based on the first compressed measurement, the first differential measurement, and the second differential measurement. ### Claim 11 (depends on 3) The apparatus of claim 3, wherein the one or more processors are further configured to cause the apparatus to quantize the first compressed measurement using a first granularity; and quantize the first differential measurement using a second granularity, the second granularity being coarser than the first granularity. ### Claim 12 (depends on 11) The apparatus of claim 11, wherein the one or more processors are further configured to cause the apparatus to receive an indication of the first granularity and the second granularity. ### Claim 13 (depends on 1) The apparatus of claim 1, wherein the one or more processors are further configured to cause the apparatus to receive an indication of whether to perform differential encoding of the compressed measurement. ### Claim 14 (depends on 1) The apparatus of claim 1, wherein the one or more processors are further configured to cause the apparatus to receive an indication of whether to perform differential encoding of the compressed measurements across measurements in at least one of the time domain or the frequency domain. ### Claim 15 An apparatus, comprising: a memory having executable instructions; and one or more processors configured to execute the executable instructions and cause the apparatus to: receive at least one message indicating compressed measurement data; generate, via each of a plurality of neural network decoders operating on the compressed measurement data, a decompressed measurement based on a respective portion of the compressed measurement data, wherein each of the neural network decoders is based on the same neural network model, wherein the compressed measurement data is associated with measurements performed for a bandwidth, and wherein the portion of the compressed measurement data is associated with a subband of the bandwidth; and assess channel characteristics based on the decompressed measurements. ### Claim 16 (depends on 15) The apparatus of claim 15, wherein the one or more processors are further configured to cause the apparatus to transmit an indication of whether to perform differential encoding for generation of the compressed measurement data. ### Claim 17 (depends on 15) The apparatus of claim 15, wherein the one or more processors are further configured to cause the apparatus to transmit an indication of whether to perform differential encoding for generation of the compressed measurement data across measurements in at least one of the time domain or the frequency domain. ### Claim 18 (depends on 15) The apparatus of claim 15, wherein: the compressed measurement data comprise a first compressed measurement data and a second compressed measurement data; the one or more processors are further configured to cause the apparatus to generate a first summation measurement, the first summation measurement being generated based on a sum of the first compressed measurement data and the second compressed measurement data; and the decompressed measurement is generated based on the first compressed measurement data and the first summation measurement. ### Claim 19 (depends on 18) The apparatus of claim 18, wherein the first compressed measurement data and the second compressed measurement data are associated with measurements performed on different subbands. ### Claim 20 (depends on 18) The apparatus of claim 18, wherein the first compressed measurement data and the second compressed measurement data are associated with measurements performed at different intervals. ### Claim 21 (depends on 20) The apparatus of claim 20, wherein: the compressed measurement data further include a third compressed measurement data; the one or more processors are further configured to cause the apparatus to generate a second summation measurement, the second summation measurement being generated based on a sum of the first compressed measurement data and the third compressed measurement data; the first compressed measurement data and the third compressed measurement data are associated with measurements performed on different subbands; and the decompressed measurement is generated based on the first compressed measurement data, the first summation measurement, and the second summation measurement. ### Claim 22 (depends on 21) The apparatus of claim 21, wherein: the compressed measurement data further include a fourth compressed measurement data; the one or more processors are further configured to cause the apparatus to generate a third summation measurement, the third summation measurement being generated as a function of the first summation measurement and the second summation measurement; and the decompressed measurement is generated based on the first compressed measurement data, the first summation measurement, the second summation measurement, and the third summation measurement. ### Claim 23 (depends on 22) The apparatus of claim 22, wherein the one or more processors are further configured to cause the apparatus to transmit an indication of the function. ### Claim 24 (depends on 18) The apparatus of claim 18, wherein: the compressed measurement data further include a third compressed measurement data; the one or more processors are further configured to cause the apparatus to generate a second summation measurement, the second summation measurement being generated based on a sum of the first compressed measurement data and the third compressed measurement data; and the decompressed measurement is generated based on the first compressed measurement data, the first summation measurement, and the second summation measurement. ### Claim 25 (depends on 18) The apparatus of claim 18, wherein: the compressed measurement data further include a third compressed measurement data; the one or more processors are further configured to cause the apparatus to generate a second summation measurement, the second summation measurement being generated based on a sum of the second compressed measurement data and the third compressed measurement data; and the decompressed measurement is generated based on the first compressed measurement data, the first summation measurement, and the second summation measurement. ### Claim 26 (depends on 18) The apparatus of claim 18, wherein the one or more processors are further configured to cause the apparatus to transmit an indication of a first quantization granularity for generating the first compressed measurement data and a second quantization granularity for generating the second compressed measurement data. ### Claim 27 A method for wireless communication, comprising: generating, via each of a plurality of neural network encoders operating on measurement data, a compressed measurement based on a respective portion of the measurement data, wherein each of the neural network encoders is based on the same neural network model, wherein the measurement data is generated using measurements performed for a bandwidth, and wherein the portion of the measurement data is associated with a subband of the bandwidth; generating at least one message indicative of the measurement data based on the compressed measurements; and transmitting the at least one message. ### Claim 28 A method for wireless communication, comprising: receiving at least one message indicating compressed measurement data; generating, via each of a plurality of neural network decoders operating on the compressed measurement data, a decompressed measurement based on a respective portion of the compressed measurement data, wherein each of the neural network decoders is based on the same neural network model, wherein the compressed measurement data is associated with measurements performed for a bandwidth, and wherein the portion of the compressed measurement data is associated with a subband of the bandwidth; and assessing channel characteristics based on the decompressed measurements.