EP-4738735-A1 - DATA COMPRESSION METHOD AND APPARATUS, AND READABLE STORAGE MEDIUM

EP4738735A1EP 4738735 A1EP4738735 A1EP 4738735A1EP-4738735-A1

Abstract

This application may be applied to the field of communication technologies, and provides a data compression method, an apparatus, and a readable storage medium, to reduce a calculation amount during compression on input data. The method includes: obtaining a compressed bitstream based on a first mapping relationship, where the first mapping relationship corresponds to a bit width and a first parameter, the first mapping relationship includes a plurality of value ranges of a first element and a plurality of bit values of a data bitstream that correspond to each other, the first element is determined based on the input data, the data bitstream is a binary bitstream corresponding to a value range of the first element, a value of the bit width is related to a quantity of bits of a bit value of the data bitstream, and the first parameter is used to determine the value range of the first element.

Inventors

LI, JIAHUI
MA, MENGYAO
LI, YANCHUN

Assignees

Huawei Technologies Co., Ltd.

Dates

Publication Date: 20260506
Application Date: 20230830

Claims (20)

A data compression method, wherein the method comprises: obtaining input data to be compressed; and obtaining a compressed bitstream based on a first mapping relationship, wherein the first mapping relationship corresponds to a bit width and a first parameter, the first mapping relationship comprises a plurality of value ranges of a first element and a plurality of bit values of a data bitstream that correspond to each other, the first element is determined based on the input data, the data bitstream is a binary bitstream corresponding to a value range of the first element, a value of the bit width is related to a quantity of bits of a bit value of the data bitstream, and the first parameter is used to determine the value range of the first element.
The method according to claim 1, wherein that the first element is determined based on the input data comprises: the first element is the input data; or the first element is a real part and an imaginary part of the input data; or the first element is an element in an absolute value of the input data; or the first element is an element in a vector obtained after the input data is normalized.
The method according to claim 1 or 2, wherein the first mapping relationship is determined from a second mapping relationship based on the bit width and/or the first parameter, and the second mapping relationship comprises boundaries of all the value ranges of the first element in the first mapping relationship.
The method according to any one of claims 1 to 3, wherein the first mapping relationship further comprises the plurality of bit values of the data bitstream and a plurality of first restored values that correspond to each other, and the first restored value is a result obtained by dequantizing the data bitstream.
The method according to any one of claims 1 to 4, wherein obtaining the compressed bitstream based on the first mapping relationship comprises: blocking the input data to obtain a blocked result; determining, based on the first mapping relationship, a data bitstream corresponding to the blocked result; and performing packet assembly on the data bitstream and a frame header to obtain the compressed bitstream, wherein the frame header comprises information indicating the first parameter.
The method according to claim 5, wherein after obtaining the blocked result, the method further comprises: normalizing the blocked result based on a normalization coefficient to obtain a normalized result; and determining, based on the first mapping relationship, the data bitstream corresponding to the blocked result comprises: determining, based on the first mapping relationship, a data bitstream corresponding to the normalized result.
The method according to claim 6, wherein the frame header further comprises information indicating the normalization coefficient, and the information indicating the normalization coefficient indicates a uniformly quantized value or a non-uniformly quantized value of the normalization coefficient.
The method according to any one of claims 5 to 7, wherein the information indicating the first parameter indicates a quantized value of the first parameter.
The method according to any one of claims 1 to 8, wherein the first parameter is determined based on an optimization indicator, and the optimization indicator comprises at least one of a normalized mean square error NMSE, a generalized cosine similarity GCS, a transmission throughput, and a bit error rate that correspond to a restored value obtained by dequantizing the data bitstream.
The method according to any one of claims 1 to 9, wherein the method further comprises: determining compression parameter information based on a first transmission configuration, wherein the first transmission configuration is related to a signal bandwidth and/or a quantity of signal transmission streams, and the compression parameter information comprises at least one of the following: a block size, a target compression ratio, the bit width, a downsampling ratio, a transform mode, the first parameter, a boundary of the value range of the first element, or a boundary of a value range of the normalization coefficient.
The method according to any one of claims 1 to 9, wherein the method further comprises: obtaining a third mapping relationship, wherein the third mapping relationship comprises a second transmission configuration and compression parameter information that correspond to each other, the second transmission configuration is related to a modulation and coding scheme of a signal, and the compression parameter information comprises at least one of the following: a block size, a target compression ratio, the bit width, a downsampling ratio, a transform mode, the first parameter, a boundary of the value range of the first element, or a boundary of a value range of the normalization coefficient.
The method according to any one of claims 1 to 9, wherein the method further comprises: receiving compression parameter information, wherein the compression parameter information comprises at least one of the following: a block size, a target compression ratio, the bit width, a downsampling ratio, a transform mode, the first parameter, a boundary of the value range of the first element, or a boundary of a value range of the normalization coefficient.
The method according to any one of claims 1 to 11, wherein the method further comprises: sending the compression parameter information, wherein the compression parameter information comprises the at least one of the following: the block size, the target compression ratio, the bit width, the downsampling ratio, the transform mode, the first parameter, the boundary of the value range of the first element, or the boundary of the value range of the normalization coefficient.
A data compression method, wherein the method comprises: receiving a compressed bitstream, wherein the compressed bitstream comprises a frame header and a data bitstream; and obtaining, based on a fourth mapping relationship, reconstructed data corresponding to the compressed bitstream, wherein the fourth mapping relationship corresponds to the bit width and the first parameter, the fourth mapping relationship comprises a plurality of bit values of a data bitstream and a plurality of first restored values that correspond to each other, and the first restored value is a result obtained by dequantizing the data bitstream.
The method according to claim 14, wherein the fourth mapping relationship further comprises a plurality of value ranges of a first element and the plurality of bit values of the data bitstream that correspond to each other, the first element is related to input data, and the compressed bitstream is determined based on the input data.
The method according to claim 15, wherein the fourth mapping relationship is determined from a second mapping relationship based on the bit width and/or the first parameter, and the second mapping relationship comprises boundaries of all the value ranges of the first element in the fourth mapping relationship.
The method according to claim 14, wherein the frame header comprises information indicating a normalization coefficient; and obtaining, based on the fourth mapping relationship, the reconstructed data corresponding to the compressed bitstream comprises: obtaining, based on the fourth mapping relationship, a restored value corresponding to the data bitstream in the compressed bitstream; and de-normalizing the restored value based on the normalization coefficient to obtain the reconstructed data.
The method according to claim 17, wherein the information indicating the normalization coefficient indicates a uniformly quantized value or a non-uniformly quantized value of the normalization coefficient.
The method according to any one of claims 14 to 18, wherein the frame header comprises information indicating the first parameter, and the information indicating the first parameter indicates a quantized value of the first parameter.
The method according to any one of claims 14 to 19, wherein the first parameter is determined based on an optimization indicator, and the optimization indicator comprises at least one of a normalized mean square error NMSE, a generalized cosine similarity GCS, a transmission throughput, and a bit error rate that correspond to the restored value obtained by dequantizing the data bitstream.

Description

TECHNICAL FIELD This application relates to the field of communication technologies, and in particular, to a data compression method, an apparatus, and a readable storage medium. BACKGROUND In an antenna and remote radio scenario, an antenna, a radio frequency unit, and the like of a base station are deployed at a remote end, and a baseband and a central compute node are deployed at a local end. A problem of data fronthaul between the remote end and the local end is considered. During uplink transmission, obtained sampling data needs to be transferred to the central compute node through an optical fiber or an Ethernet for subsequent operations such as baseband processing and data fusion. During downlink transmission, data sent from the central compute node needs to be obtained and processed by the radio frequency unit and then transmitted through the antenna. However, as a quantity of antennas and a quantity of distributed units increase, an amount of data requiring fronthaul surges sharply. To reduce a bandwidth requirement of a fronthaul link and system costs, fronthaul data needs to be compressed. However, a current solution for compressing the fronthaul data needs to perform a large amount of calculation at a transmit end and a receive end, requires a large quantity of calculation resources, and is highly complex. SUMMARY Embodiments of this application provide a data compression method, an apparatus, and a readable storage medium, to lower complexity of compressing data. To achieve the foregoing objective, this application uses the following technical solutions. According to a first aspect, a data compression method is provided. The method may be performed by a first unit, or may be performed by a module (for example, a processor, a chip, or a chip system) used in the first unit. Descriptions are provided below by using an example in which the first unit performs the method. The method includes: The first unit obtains input data to be compressed. The first unit obtains a compressed bitstream based on a first mapping relationship, where the first mapping relationship corresponds to a bit width and a first parameter, the first mapping relationship includes a plurality of value ranges of a first element and a plurality of bit values of a data bitstream that correspond to each other, the first element is determined based on the input data, the data bitstream is a binary bitstream corresponding to a value range of the first element, a value of the bit width is related to a quantity of bits of a bit value of the data bitstream, and the first parameter is used to determine the value range of the first element. Based on the data compression method provided in embodiments of this application, the first unit can directly determine, based on the first mapping relationship, a bit value, of a data bitstream, corresponding to a value of a first element, so that a calculation amount required for some operations (for example, quantization) in a process of compressing the input data can be omitted, and the compressed bitstream can be obtained, thereby lowering complexity. With reference to the first aspect, in a possible design, that the first element is determined based on the input data includes: The first element is the input data; or the first element is a real part and an imaginary part of the input data; or the first element is an element in an absolute value of the input data; or the first element is an element in a vector obtained after the input data is normalized. Based on this solution, the first unit can perform specific processing on the input data to obtain the first element, and can determine a corresponding data bitstream based on a value of the first element and the first mapping relationship. With reference to the first aspect, in a possible design, the first mapping relationship is determined from a second mapping relationship based on the bit width and/or the first parameter, and the second mapping relationship includes boundaries of all the value ranges of the first element in the first mapping relationship. Based on this solution, the first mapping relationship may be included in the second mapping relationship, so that storage resources can be saved. With reference to the first aspect, in a possible design, the first mapping relationship further includes the plurality of bit values of the data bitstream and a plurality of first restored values that correspond to each other, and the first restored value is a result obtained by dequantizing the data bitstream. Based on this solution, the first mapping relationship may be further used to dequantize the data bitstream. With reference to the first aspect, in a possible design, that the first unit obtains the compressed bitstream based on the first mapping relationship includes: The first unit blocks the input data to obtain a blocked result. The first unit determines, based on the first mapping relationship, a data bitstream corresponding to the block