An uplink data processing system

By re-dividing the processing architecture in the uplink data processing of 5G NR base stations, only key data is transmitted and calculation methods are implemented in the processor, the problem of high interface throughput pressure is solved and data processing capabilities and flexibility is improved.

CN114696978BActive Publication Date: 2024-12-03GUANGZHOU HUIRUI SITONG INFORMATION SCI & TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202011617407.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-12-31
Publication Date
2024-12-03
Estimated Expiration
2040-12-31

AI Technical Summary

Technical Problem

In the uplink data processing of 5G NR base stations, the interface throughput pressure between the hardware processing part and the software processing part is relatively high, resulting in insufficient data processing capabilities and inability to effectively respond to future update needs.

Method used

The processing architecture of the High-PHY part is re-divided. Only the power delay distribution results, delay data, filter coefficients and demapping position information are transmitted between the hardware accelerator and the processor, which reduces the data transmission pressure and flexibly places the calculation method. The processor is implemented through software algorithms for easy updates.

Benefits of technology

It reduces the data transmission pressure between the hardware accelerator and the processor, enhances the data processing capability of the High-PHY part, and improves processing flexibility and adaptability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114696978B_ABST
    Figure CN114696978B_ABST
Patent Text Reader

Abstract

The present disclosure relates to the field of data processing, and in particular to an uplink data processing system, including a hardware accelerator and a processor; the hardware accelerator is used to receive uplink data and reference signal position information transmitted by the processor, and transmit the power delay distribution result generated according to the uplink data and the reference signal position information to the processor; the processor is used to transmit the delay data and filter coefficient generated according to the power delay distribution result to the hardware accelerator; and is used to self-generate demapping position information and the reference signal position information and transmit them to the hardware accelerator; the hardware accelerator is also used to generate transmission bits according to the uplink data, the delay data, the filter coefficient and the demapping position information. The present disclosure is used to solve the problem of high interface throughput pressure caused by data exchange between the hardware processing part and the software processing part when the physical layer of the high-speed port processes uplink data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of data processing, and in particular to an uplink data processing system. Background Art

[0002] In the base station implementation of the global 5G standard (5G New Radio, 5G NR) based on the new air interface design of Orthogonal Frequency Division Multiplexing (OFDM), the high-speed port physical layer (High-PHY) that processes uplink data is the most complex part. For the currently proposed processing architecture, if the architecture is implemented entirely through hardware chips, it will affect flexibility. The main reason is that with the evolution of protocols and algorithm improvements, for example, there is a high possibility of continuous improvement of some calculations in channel estimation, and being solidified in hardware chips will make it impossible to cope with future update needs. If the architecture is implemented entirely using software-defined radio (Software Defined Radio, SDR), it will place huge demands on software processors. The main reason is that the data processing volume of 5G NR is very large, and only using software processors such as digital signal processing (DSP) chips cannot solve the complex algorithm support for full specifications on the network side. Summary of the invention

[0003] The inventors found that in the related art, a commonly used method is to implement the channel estimation and protocol analysis related parts in software, and implement relatively stable algorithms, such as bit-level processing modules, multiple-in-multiple-out (MIMO) modules, etc. in hardware accelerators. However, because the data processing volume of 5G NR is very large, if all the data output by the channel estimation is to be transmitted from the software processor to the hardware chip for further processing, it will put great pressure on the interface throughput between the software and hardware.

[0004] The present invention provides an uplink data processing system for solving the problem of high interface throughput pressure caused by data exchange between a hardware processing part and a software processing part when a high-speed port physical layer processes uplink data.

[0005] In a first aspect, an embodiment of the present disclosure provides an uplink data processing system, comprising a hardware accelerator and a processor; the hardware accelerator is used to receive uplink data and reference signal position information transmitted by the processor, generate a power delay distribution result according to the uplink data and the reference signal position information, and transmit the power delay distribution result to the processor; the processor is used to generate delay data and filter coefficients according to the power delay distribution result, and transmit the delay data and the filter coefficients to the hardware accelerator; and is used to self-generate demapping position information and the reference signal position information, and transmit the demapping position information and the reference signal position information to the hardware accelerator; the hardware accelerator is also used to generate transmission bits according to the uplink data, the delay data, the filter coefficient and the demapping position information.

[0006] Optionally, the hardware accelerator includes a first accelerator, a second accelerator, a third accelerator and a fourth accelerator; the first accelerator is used to generate a channel estimation result after frequency offset estimation and the power delay distribution result according to the uplink data and the reference signal position information, and transmit the channel estimation result after frequency offset estimation to the second accelerator and the third accelerator respectively, and transmit the power delay distribution result to the processor; the second accelerator is used to generate uplink data after frequency offset correction according to the delay data and the channel estimation result after frequency offset estimation, and transmit the uplink data after frequency offset correction to the fourth accelerator; the third accelerator is used to generate a second interpolation result and a noise autocorrelation matrix according to the delay data and the filter coefficient, and transmit the second interpolation result and the noise autocorrelation matrix to the fourth accelerator; the fourth accelerator is used to generate the transmission bit according to the demapping position information, the uplink data after frequency offset correction, the second interpolation result and the noise autocorrelation matrix.

[0007] Optionally, the processor includes a memory, which stores a corresponding computer program; the processor is used to execute the program stored in the memory to implement the following steps: generate the delay data according to the power delay distribution result, and transmit the delay data to the second accelerator and the third accelerator, and generate correlation coefficients and Doppler expansion results; generate the filter coefficients according to the correlation coefficients and the Doppler expansion results, and transmit the filter coefficients to the third accelerator; self-generate the demapping position information and the reference signal position information, and transmit the demapping position information and the reference signal position information to the hardware accelerator.

[0008] Optionally, the filter coefficient includes a time domain filter coefficient and a frequency domain filter coefficient; generating the filter coefficient according to the correlation coefficient and the Doppler expansion result, and transmitting the filter coefficient to the third accelerator includes: generating the frequency domain filter coefficient according to the correlation coefficient, and transmitting the frequency domain filter coefficient to the third accelerator; generating the time domain filter coefficient according to the Doppler expansion result, and transmitting the time domain filter coefficient to the third accelerator.

[0009] Optionally, the demapping position information and the reference signal position information are self-generated, and the demapping position information and the reference signal position information are transmitted to the hardware accelerator, including: self-generating the reference signal position information and transmitting it to the first accelerator; self-generating the demapping position information and transmitting it to the fourth accelerator.

[0010] Optionally, the first accelerator includes a reference signal extraction module, a reference signal local sequence generation module, a channel estimation module, a frequency offset estimation module and a power delay distribution calculation module; the reference signal extraction module is used to extract reference signal data according to the uplink data and the reference signal position information, and transmit the reference signal data to the channel estimation module; the reference signal local sequence generation module is used to self-generate a local sequence of the reference signal, and transmit the local sequence to the channel estimation module; the channel estimation module is used to generate a channel estimation result according to the reference signal data and the local sequence, and transmit the channel estimation result to the frequency offset estimation module; the frequency offset estimation module is used to perform frequency offset estimation according to the channel estimation result, generate a frequency offset estimation result, and transmit the frequency offset estimation result to the second accelerator, and transmit the channel estimation result after the frequency offset estimation to the power delay distribution calculation module and the third accelerator respectively; the power delay distribution calculation module is used to generate a power delay distribution result according to the channel estimation result after the frequency offset estimation, and transmit the power delay distribution result to the measurement module.

[0011] Optionally, the third accelerator includes a frequency domain filtering module, a time domain filtering module, a frequency domain interpolation module, a time domain interpolation module and a noise autocorrelation matrix calculation module; the frequency domain filtering module is used to generate a first filtering result according to the channel estimation result after the frequency offset estimation, the delay data and the frequency domain filtering coefficient, and transmit the first filtering result to the time domain filtering module; the time domain filtering module is used to generate a second filtering result according to the first filtering result and the time domain filtering coefficient, and transmit the second filtering result to the frequency domain interpolation module and the noise autocorrelation matrix calculation module respectively; the frequency domain interpolation module is used to complete frequency domain interpolation according to the second filtering result, generate a first interpolation result, and transmit the first interpolation result to the time domain interpolation module; the time domain interpolation module is used to complete time domain interpolation according to the first interpolation result, generate a second interpolation result, and transmit the second interpolation result to the fourth accelerator; the noise autocorrelation matrix calculation module is used to generate the noise autocorrelation matrix according to the second filtering result, and transmit the noise autocorrelation matrix to the fourth accelerator.

[0012] Optionally, the fourth accelerator includes a demapping module, a balancing module, a demodulation module and a bit-level processing module; the demapping module is used to generate demapped data according to the uplink data after frequency offset correction, the second interpolation result and the demapping position information, and transmit the demapped data to the balancing module; the balancing module is used to generate balanced multi-layer data according to the demapped data and the noise autocorrelation matrix, and transmit the balanced multi-layer data to the demodulation module; the demodulation module is used to generate demodulated data according to the balanced multi-layer data, and transmit the demodulated data to the bit-level processing module; the bit-level processing module is used to generate the transmission bit according to the demodulated data.

[0013] Optionally, the processor is a digital signal processing chip.

[0014] Optionally, the hardware accelerator is a field programmable gate array integrated circuit.

[0015] Compared with the related art, the above technical solution provided by the embodiment of the present disclosure has the following advantages: the system provided by the embodiment of the present disclosure redivides the processing architecture of the High-PHY part that processes uplink data. In the redivided processing architecture, the data that needs to be transmitted between the hardware accelerator and the processor that implements the software algorithm only include the power delay distribution results, delay data, filter coefficients, demapping location information and the reference signal location information. It is no longer necessary to transmit all the uplink data or channel estimation output data between the hardware accelerator and the processor, which greatly reduces the data transmission pressure of the interface between the hardware accelerator and the processor, and further enhances the ability of the High-PHY part to process data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] Figure 1 Schematic diagram of the connection structure of the uplink data processing system provided in the embodiment of the present disclosure Figure 1 ;

[0019] Figure 2 A schematic diagram of the internal structure connection of the hardware accelerator provided in an embodiment of the present disclosure;

[0020] Figure 3 A schematic diagram of the internal structure connection of a processor provided in an embodiment of the present disclosure;

[0021] Figure 4 A schematic diagram of the internal structure connection of a coefficient calculation module provided in an embodiment of the present disclosure;

[0022] Figure 5 A schematic diagram of the internal structure connection of the location information generation module provided in an embodiment of the present disclosure;

[0023] Figure 6 A schematic diagram of the internal structure connection of the first accelerator provided in an embodiment of the present disclosure;

[0024] Figure 7 A schematic diagram of the internal structure connection of the third accelerator provided in an embodiment of the present disclosure;

[0025] Figure 8A schematic diagram of the internal structure connection of a fourth accelerator provided in an embodiment of the present disclosure;

[0026] Fig. 9 Schematic diagram of the connection structure of the uplink data processing system provided in the embodiment of the present disclosure Figure 2 . DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.

[0028] The uplink data processing system provided in the embodiments of the present disclosure re-divides the processing architecture of the High-PHY part. The uplink data processing system is applied to the High-PHY part of the 5G NR base station side, and obtains transmission bits after demodulating the uplink data. The new processing architecture of the system reduces the pressure on the interface between the hardware accelerator and the processor, and enhances the data processing capability of the High-PHY part.

[0029] In one embodiment, if Figure 1 As shown, the uplink data processing system includes a hardware accelerator 101 and a processor 102. The hardware accelerator 101 implements a fixed calculation process in a solidified form. Once the calculation method in the hardware accelerator 101 is solidified, it cannot be modified; the processor 102 implements the calculation process through a software algorithm, and the software algorithm stored in the processor 102 can be modified according to actual conditions and needs.

[0030] In this embodiment, the hardware accelerator 101 is used to receive uplink data and reference signal position information transmitted by the processor 102, generate a power delay profile (PDP) result according to the uplink data and the reference signal position information, and transmit the power delay profile result to the processor 102.

[0031] Processor 102 is used to generate timing delay (TD) data and filter coefficients according to the power delay distribution result, and transmit the delay data and filter coefficients to the hardware accelerator 101; and to self-generate demapping position information and reference signal (RS) position information, and transmit the demapping position information and reference signal position information to the hardware accelerator 101.

[0032] The hardware accelerator 101 is further configured to generate transmission bits according to the uplink data, the delay data, the filter coefficients and the demapping position information.

[0033] In this embodiment, the uplink data is IQ data, where I refers to in-phase and Q refers to quadrature. The IQ data is communication data formed by two orthogonal signals.

[0034] In this embodiment, the hardware accelerator 101 is used to receive a large amount of uplink data, and complete most of the processing of the uplink data in the hardware accelerator. The data exchanged with the processor only includes the power delay distribution result, delay data, filter coefficient, demapping position information and the reference signal position information. The data volume of these data is relatively small, which greatly reduces the data transmission pressure of the interface between the hardware accelerator and the processor. At the same time, the process of processing the power delay distribution result to generate delay data and filter coefficients, as well as the process of self-generation of demapping position information and reference signal position information, has a relatively flexible calculation method, which is placed in the processor for calculation through software algorithms, which is convenient for updating the subsequent calculation method.

[0035] In this embodiment, a bit is the smallest unit of data in a computer, and a "bit" has a value, including 0 or 1. A transmission bit refers to data transmitted in the form of a bit.

[0036] In one embodiment, if Figure 2 As shown, the hardware accelerator 101 includes a first accelerator 1011 , a second accelerator 1012 , a third accelerator 1013 and a fourth accelerator 1014 .

[0037] The first accelerator 1011 is used to generate a channel estimation result and a power delay distribution result after frequency offset estimation according to the uplink data and the reference signal position information, and transmit the channel estimation result after frequency offset estimation to the second accelerator 1012 and the third accelerator 1013 respectively, and transmit the power delay distribution result to the processor 102.

[0038] The second accelerator 1012 is used to generate uplink data after frequency offset correction according to the delay data and the channel estimation result after frequency offset estimation, and transmit the uplink data after frequency offset correction to the fourth accelerator 1014.

[0039] The third accelerator 1013 is used to generate a second interpolation result and a noise autocorrelation matrix according to the time delay data and the filter coefficient, and transmit the second interpolation result and the noise autocorrelation matrix to the fourth accelerator 1014.

[0040] The fourth accelerator 1014 is used to generate transmission bits according to the demapping position information, the uplink data after the frequency offset correction, the second interpolation result and the noise autocorrelation matrix.

[0041] In one embodiment, the processor 102 includes a memory, and the memory stores a corresponding computer program. The processor is used to execute the program stored in the memory to implement the following steps:

[0042] Generate time delay data according to the power delay distribution result, transmit the time delay data to the second accelerator 1012 and the third accelerator 1013, and generate correlation coefficient and Doppler spread result;

[0043] Generate a filter coefficient according to the correlation coefficient and the Doppler spread result, and transmit the filter coefficient to the third accelerator 1013;

[0044] The demapping position information and the reference signal position information are generated by the hardware accelerator 101 , and the demapping position information and the reference signal position information are transmitted to the hardware accelerator 101 .

[0045] In one embodiment, when the processor executes the computer program in the memory, the processor is functionally divided, such as Figure 3 As shown, the processor 102 includes a measurement module 1021, a coefficient calculation module 1022 and a position information generation module 1023;

[0046] The measurement module 1021 is used to generate delay data according to the power delay distribution result, and transmit the delay data to the second accelerator 1012 and the third accelerator 1013, and to generate correlation coefficients and Doppler spread results, and transmit the correlation coefficients and Doppler spread results to the coefficient calculation module 1022.

[0047] The coefficient calculation module 1022 is used to generate filter coefficients according to the correlation coefficients and the Doppler spread result, and transmit the filter coefficients to the third accelerator 1013 .

[0048] The position information generating module 1023 is used to self-generate demapping position information and reference signal position information, and transmit the demapping position information and reference signal position information to the hardware accelerator 101 .

[0049] In one embodiment, the filter coefficients include time domain filter coefficients and frequency domain filter coefficients. The filter coefficients are generated according to the correlation coefficients and the Doppler spread result, and the filter coefficients are transmitted to the third accelerator. The specific process is as follows: the frequency domain filter coefficients are generated according to the correlation coefficients, and the frequency domain filter coefficients are transmitted to the third accelerator 1013; the time domain filter coefficients are generated according to the Doppler spread result, and the time domain filter coefficients are transmitted to the third accelerator 1013.

[0050] Similarly, the processor is functionally divided again according to this part of the content, such as Figure 4As shown, the coefficient calculation module 1022 includes a frequency domain coefficient calculation module 1022-1 and a time domain coefficient calculation module 1022-2; the filter coefficient includes a time domain filter coefficient and a frequency domain filter coefficient.

[0051] The frequency domain coefficient calculation module 1022 - 1 is used to generate frequency domain filter coefficients according to the correlation coefficients, and transmit the frequency domain filter coefficients to the third accelerator 1013 .

[0052] The time domain coefficient calculation module 1022 - 2 is used to generate a time domain filter coefficient according to the Doppler expansion result, and transmit the time domain filter coefficient to the third accelerator 1013 .

[0053] In one embodiment, demapping position information and reference signal position information are self-generated and transmitted to the hardware accelerator. The specific process is as follows: reference signal position information is self-generated and transmitted to the first accelerator 1011; demapping position information is self-generated and transmitted to the fourth accelerator 1014.

[0054] Similarly, the processor is functionally divided again according to this part of the content, such as Figure 5 As shown, the location information generating module 1023 includes a reference signal location calculating module 1023-1 and a demapping location calculating module 1023-2;

[0055] The reference signal position calculation module 1023 - 1 is used to self-generate reference signal position information and transmit it to the first accelerator 1011 .

[0056] The demapping position calculation module 1023 - 2 is used to self-generate demapping position information and transmit it to the fourth accelerator 1014 .

[0057] In one embodiment, if Figure 6 As shown, the first accelerator 1011 includes a reference signal extraction module 1011-1, a reference signal local sequence generation module 1011-2, a channel estimation module 1011-3, a frequency offset estimation module 1011-4 and a power delay distribution calculation module 1011-5;

[0058] The reference signal extraction module 1011-1 is used to extract reference signal data according to the uplink data and the reference signal position information, and transmit the reference signal data to the channel estimation module 1011-3;

[0059] A reference signal local sequence generation module 1011-2, configured to self-generate a local sequence of a reference signal and transmit the local sequence to a channel estimation module 1011-3;

[0060] The channel estimation module 1011-3 is used to generate a channel estimation result according to the reference signal data and the local sequence, and transmit the channel estimation result to the frequency offset estimation module 1011-4;

[0061] The frequency offset estimation module 1011-4 is used to perform frequency offset estimation according to the channel estimation result, generate the frequency offset estimation result, and transmit the frequency offset estimation result to the second accelerator 1012, and transmit the channel estimation result after the frequency offset estimation to the power delay distribution calculation module 1011-5 and the third accelerator 1013 respectively;

[0062] The power delay distribution calculation module 1011 - 5 is used to generate a power delay distribution result according to the channel estimation result after the frequency offset estimation, and transmit the power delay distribution result to the measurement module 1021 .

[0063] In one embodiment, if Figure 7 As shown, the third accelerator 1013 includes a frequency domain filtering module 1013-1, a time domain filtering module 1013-2, a frequency domain interpolation module 1013-4, a time domain interpolation module 1013-5 and a noise autocorrelation matrix calculation module 1013-3;

[0064] The frequency domain filtering module 1013-1 is used to generate a first filtering result according to the channel estimation result after the frequency offset estimation, the delay data and the frequency domain filtering coefficient, and transmit the first filtering result to the time domain filtering module 1013-2;

[0065] The time domain filtering module 1013-2 is used to generate a second filtering result according to the first filtering result and the time domain filtering coefficient, and transmit the second filtering result to the frequency domain interpolation module 1013-4 and the noise autocorrelation matrix calculation module 1013-3 respectively;

[0066] The frequency domain interpolation module 1013-4 is used to complete the frequency domain interpolation according to the second filtering result, generate a first interpolation result, and transmit the first interpolation result to the time domain interpolation module 1013-5;

[0067] A time domain interpolation module 1013-5, configured to complete time domain interpolation according to the first interpolation result, generate a second interpolation result, and transmit the second interpolation result to a fourth accelerator;

[0068] The noise autocorrelation matrix calculation module 1013 - 3 is used to generate a noise autocorrelation matrix according to the second filtering result, and transmit the noise autocorrelation matrix to the fourth accelerator.

[0069] In one embodiment, if Figure 8 As shown, the fourth accelerator 1014 includes a demapping module 1014-1, an equalization module 1014-2, a demodulation module 1014-3 and a bit-level processing module 1014-4;

[0070] The demapping module 1014-1 is used to generate demapped data according to the uplink data after the frequency offset correction, the second interpolation result and the demapping position information, and transmit the demapped data to the equalization module 1014-2;

[0071] The equalization module 1014-2 is used to generate equalized multi-layer data according to the demapped data and the noise autocorrelation matrix, and transmit the equalized multi-layer data to the demodulation module 1014-3;

[0072] The demodulation module 1014-3 is used to generate demodulated data according to the equalized multi-layer data, and transmit the demodulated data to the bit level processing module 1014-4;

[0073] The bit level processing module 1014 - 4 is used to generate transmission bits according to the demodulated data.

[0074] In one embodiment, if Fig. 9 As shown, the fourth accelerator further includes a decoder 1014 - 5 . The decoder 1014 - 5 receives the transmission bits transmitted by the bit-level processing module 1014 - 4 , decodes the transmission bits, and transmits them to the processor 102 .

[0075] In this embodiment, the processor 103 includes a Family Application Programming Interface (FAPI interface) 1024. The FAPI interface 1024 is used to package the decoded transmission bits into FAPI data for transmission, so as to complete other processing processes on the base station side.

[0076] In one embodiment, the processor 102 is a digital signal processing (DSP) chip. The processor 102 may also be implemented in other forms according to actual conditions and needs, and the protection scope of the present disclosure is not limited to the specific implementation form of the processor.

[0077] In one embodiment, the hardware accelerator 101 is a field programmable gate array (FPGA) integrated circuit. The hardware accelerator 101 may also be implemented in other forms according to actual conditions and needs, and the protection scope of the present disclosure is not limited to the specific implementation form of the hardware accelerator.

[0078] In one embodiment, the channel estimation is a channel estimation method based on the Least Squares (LS) criterion.

[0079] In one embodiment, when the uplink data processing system processes uplink data, during the process of the hardware accelerator 101 processing data, each accelerator included is scheduled by the processor 102. The specific process is as follows:

[0080] The first accelerator 1011 mainly implements LS channel estimation for each user. Since LS channel estimation is implemented independently between RS symbols, the processing of the current symbol can be started after each RS symbol arrives. The processor 102 identifies the RS symbol of each user. When it is determined that the current RS symbol is the RS symbol of the user to be processed, it immediately configures the parameters of the first accelerator 1011 of the user and sends a start signal to the first accelerator 1011, and the first accelerator 1011 will start working. If the processor 102 finds that the current RS symbol is the last RS symbol of the user in this time slot when generating the start signal, it should also inform the first accelerator 1011 through the configuration parameters.

[0081] The first accelerator 1011 saves the requested pending tasks in the form of a queue. When each requested task is processed, if it is not the last RS symbol of the user in this time slot, there is no need to notify the processor 102 in the form of a completion interrupt. If it is recognized that the current task is the last RS symbol of the user in this time slot, the first accelerator 1011 will perform a PDP merge operation to generate a PDP result, and after completion, send the first accelerator 1011 completion interrupt to the processor 102, and attach the user number corresponding to the completed task in the return parameter.

[0082] The second accelerator 1012 completes the frequency offset correction for the user data portion. Since the frequency offset estimation can only be performed after the LS channel estimation of all RS symbols is completed, the processor 102 generates a start signal and sends it to the second accelerator 1012 when the task of the first accelerator 1011 of the last RS symbol of the current user is completed and the first accelerator 1011 completion interrupt is received to start the second accelerator 1012.

[0083] The frequency offset correction of the second accelerator 1012 includes the frequency offset correction of the RS part and the frequency offset correction of the uplink data part. The correction processing time for the uplink data part will be relatively long, and the subsequent channel estimation does not depend on the uplink data after the correction, so the processing time of the frequency offset correction of the uplink data part is allowed to be delayed. After the second accelerator 1012 completes the frequency offset correction of the RS part, it sends a semi-complete interrupt of the second accelerator 1012 to the processor 102, so that the processor 102 starts the third accelerator 1013. However, the frequency offset correction of the uplink data part has no direct timing relationship with the processing process of the third accelerator 1013. After the frequency offset correction of the uplink data part is completed, the fourth accelerator 1014 can start working. Therefore, after the frequency offset correction of the uplink data part is completed, the second accelerator 1012 will send a completion interrupt to the processor 102 to start the fourth accelerator 1014.

[0084] The third accelerator 1013 mainly completes the noise suppression of channel estimation through filtering. The main steps include time domain channel denoising, time offset correction, RS position frequency domain filtering noise suppression, RS position time domain filtering noise suppression, data position frequency domain interpolation, data position time domain interpolation and noise autocorrelation matrix calculation, where the noise autocorrelation matrix is ​​also called noise covariance matrix.

[0085] The third accelerator 1013 can only be started after the processor 102 completes the calculation of the filter coefficient. The filter coefficient is configured by the processor 102 to the third accelerator 1013, which is the part with the largest amount of calculation in the channel estimation. The third accelerator 1013 will output the second interpolation result and the noise autocorrelation matrix that are filtered and interpolated in the time domain and frequency domain respectively. After completion, the completion interrupt of the third accelerator 1013 will be sent to the processor 102.

[0086] The fourth accelerator 1014 includes several modules such as a demapping module 1014-1, an equalization module 1014-2, a demodulation module 1014-3 and a bit-level processing module 1014-4, which are connected in series. The work of each module inside the fourth accelerator 1014 no longer requires the processor 102 to participate in the scheduling. Among them, the demapping module 1014-1 requires the processor 102 to configure the startup parameters. After the parameters are configured, if the processor 102 confirms that the completion interrupt of the third accelerator 1013 of the current user has been received, and the completion interrupt of the second accelerator 1012 has also been received, a start signal is generated and transmitted to the fourth accelerator 1014 to start the fourth accelerator 1014. After the processing of the fourth accelerator 1014 is completed, the completion interrupt flag of the fourth accelerator 1014 is generated and transmitted to the processor 102. At this time, the processor 102 can package and send data of the FAPI interface.

[0087] The uplink data processing system provided by the present disclosure redistributes the processing architecture of the High-PHY part that processes uplink data. In the redistributed processing architecture, the data that needs to be transmitted between the hardware accelerator and the processor that implements the software algorithm only include the power delay distribution results, delay data, filter coefficients, demapping position information and the reference signal position information. It is no longer necessary to transmit all the uplink data or channel estimation output data between the hardware accelerator and the processor, which greatly reduces the data transmission pressure of the interface between the hardware accelerator and the processor. At the same time, the process of processing the power delay distribution results to generate delay data and filter coefficients, as well as the process of self-generation of demapping position information and reference signal position information, has a more flexible calculation method, which is placed in the processor for calculation through software algorithms, which is convenient for updating subsequent calculation methods. As a result, the ability of the High-PHY part to process uplink data is enhanced.

[0088] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0089] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An uplink data processing system, characterized in that: Includes hardware accelerators and processors; The hardware accelerator is used to receive uplink data and reference signal position information transmitted by the processor, generate a power delay distribution result according to the uplink data and the reference signal position information, and transmit the power delay distribution result to the processor; The processor is used to generate delay data and filter coefficients according to the power delay distribution result, and transmit the delay data and the filter coefficients to the hardware accelerator; and is used to self-generate demapping position information and the reference signal position information, and transmit the demapping position information and the reference signal position information to the hardware accelerator; The hardware accelerator is further used to generate transmission bits according to the uplink data, the delay data, the filter coefficient and the demapping position information; The hardware accelerator includes a first accelerator, a second accelerator, a third accelerator and a fourth accelerator; The first accelerator is used to generate a channel estimation result after frequency offset estimation and the power delay distribution result according to the uplink data and the reference signal position information, and transmit the channel estimation result after frequency offset estimation to the second accelerator and the third accelerator respectively, and transmit the power delay distribution result to the processor; The second accelerator is used to generate uplink data after frequency offset correction according to the delay data and the channel estimation result after the frequency offset estimation, and transmit the uplink data after frequency offset correction to the fourth accelerator; The third accelerator is used to generate a second interpolation result and a noise autocorrelation matrix according to the time delay data and the filter coefficient, and transmit the second interpolation result and the noise autocorrelation matrix to the fourth accelerator; The fourth accelerator is used to generate the transmission bits according to the demapping position information, the uplink data after frequency offset correction, the second interpolation result and the noise autocorrelation matrix.

2. The uplink data processing system according to claim 1, characterized in that: The processor includes a memory storing a corresponding computer program; The processor is used to execute the program stored in the memory to implement the following steps: Generating the delay data according to the power delay distribution result, transmitting the delay data to the second accelerator and the third accelerator, and generating a correlation coefficient and a Doppler spread result; generating the filter coefficient according to the correlation coefficient and the Doppler spread result, and transmitting the filter coefficient to the third accelerator; The demapping position information and the reference signal position information are self-generated, and the demapping position information and the reference signal position information are transmitted to the hardware accelerator.

3. The uplink data processing system according to claim 1, characterized in that: The processor includes a measurement module, a coefficient calculation module and a position information generation module; The measuring module is used to generate the delay data according to the power delay distribution result, and transmit the delay data to the second accelerator and the third accelerator, and to generate a correlation coefficient and a Doppler spread result, and transmit the correlation coefficient and the Doppler spread result to the coefficient calculation module; The coefficient calculation module is used to generate the filter coefficient according to the correlation coefficient and the Doppler spread result, and transmit the filter coefficient to the third accelerator; The position information generating module is used to self-generate the demapping position information and the reference signal position information, and transmit the demapping position information and the reference signal position information to the hardware accelerator.

4. The uplink data processing system according to any one of claims 2 or 3, characterized in that: The filter coefficients include time domain filter coefficients and frequency domain filter coefficients; The step of generating the filter coefficient according to the correlation coefficient and the Doppler spread result, and transmitting the filter coefficient to the third accelerator comprises: generating the frequency domain filter coefficient according to the correlation coefficient, and transmitting the frequency domain filter coefficient to the third accelerator; The time domain filter coefficient is generated according to the Doppler expansion result, and the time domain filter coefficient is transmitted to the third accelerator.

5. The uplink data processing system according to claim 4, characterized in that: The method comprises: generating the demapping position information and the reference signal position information by itself, and transmitting the demapping position information and the reference signal position information to the hardware accelerator, comprising: Self-generating the reference signal position information and transmitting it to the first accelerator; The demapped position information is generated automatically and transmitted to the fourth accelerator.

6. The uplink data processing system according to claim 5, characterized in that: The first accelerator includes a reference signal extraction module, a reference signal local sequence generation module, a channel estimation module, a frequency offset estimation module and a power delay distribution calculation module; The reference signal extraction module is used to extract reference signal data according to the uplink data and the reference signal position information, and transmit the reference signal data to the channel estimation module; The reference signal local sequence generation module is used to self-generate a local sequence of a reference signal and transmit the local sequence to the channel estimation module; The channel estimation module is used to generate a channel estimation result according to the reference signal data and the local sequence, and transmit the channel estimation result to the frequency offset estimation module; The frequency offset estimation module is used to perform frequency offset estimation according to the channel estimation result, generate a frequency offset estimation result, and transmit the frequency offset estimation result to the second accelerator, and transmit the channel estimation result after the frequency offset estimation to the power delay distribution calculation module and the third accelerator respectively; The power delay distribution calculation module is used to generate a power delay distribution result according to the channel estimation result after the frequency offset estimation, and transmit the power delay distribution result to the measurement module.

7. The uplink data processing system according to claim 5, characterized in that: The third accelerator includes a frequency domain filtering module, a time domain filtering module, a frequency domain interpolation module, a time domain interpolation module and a noise autocorrelation matrix calculation module; The frequency domain filtering module is used to generate a first filtering result according to the channel estimation result after the frequency offset estimation, the delay data and the frequency domain filtering coefficient, and transmit the first filtering result to the time domain filtering module; The time domain filtering module is used to generate a second filtering result according to the first filtering result and the time domain filtering coefficient, and transmit the second filtering result to the frequency domain interpolation module and the noise autocorrelation matrix calculation module respectively; The frequency domain interpolation module is used to perform frequency domain interpolation according to the second filtering result, generate a first interpolation result, and transmit the first interpolation result to the time domain interpolation module; The time domain interpolation module is used to complete the time domain interpolation according to the first interpolation result, generate a second interpolation result, and transmit the second interpolation result to the fourth accelerator; The noise autocorrelation matrix calculation module is used to generate the noise autocorrelation matrix according to the second filtering result, and transmit the noise autocorrelation matrix to the fourth accelerator.

8. The uplink data processing system according to claim 5, characterized in that: The fourth accelerator includes a demapping module, an equalization module, a demodulation module and a bit-level processing module; The demapping module is used to generate demapped data according to the uplink data after the frequency offset correction, the second interpolation result and the demapping position information, and transmit the demapped data to the equalization module; The equalization module is used to generate equalized multi-layer data according to the demapped data and the noise autocorrelation matrix, and transmit the equalized multi-layer data to the demodulation module; The demodulation module is used to generate demodulated data according to the equalized multi-layer data, and transmit the demodulated data to the bit-level processing module; The bit-level processing module is used to generate the transmission bit according to the demodulated data.

9. The uplink data processing system according to any one of claims 1 to 8, characterized in that: The processor is a digital signal processing chip.

10. The uplink data processing system according to any one of claims 1 to 8, characterized in that: The hardware accelerator is a field programmable logic gate array integrated circuit.