A high-precision real-time demodulation method based on 5G NR

By using the combination technology of FPGA and DSP chips on boards that support 4G LTE, variable sampling and high-precision demodulation are achieved from 153.6MHz to 122.88MHz, solving the problem of low demodulation accuracy of 5G NR and achieving high precision Real-time demodulation and stable operation.

CN114205198BActive Publication Date: 2024-11-05CLP KESI INSTR TECH (ANHUI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111419917.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-26
Publication Date
2024-11-05
Estimated Expiration
2041-11-26

AI Technical Summary

Technical Problem

The prior art is difficult to add 5G NR function to boards that support 4G LTE, and the demodulation accuracy is low and real-time demodulation cannot be achieved.

Method used

Using FPGA chip, DSP chip and industrial control board, variable sampling and high-precision demodulation from 153.6MHz to 122.88MHz are used to achieve variable sampling and high-precision demodulation through FPGA variable sampling, SRIO real-time transmission, DSP high-precision demodulation and EMAC self-start. It is compatible with 4G LTE , and optimize time-frequency offset calculation using multi-core coprocessor and DMRS related calculation.

Benefits of technology

It realizes high-precision real-time demodulation of 5G NR, and the signal quality meets the requirements of 3GPP protocol. It has low equipment upgrade cost, stable operation, and large-capacity storage is convenient for remote upgrades.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114205198B_ABST
    Figure CN114205198B_ABST
Patent Text Reader

Abstract

The present invention discloses a high-precision real-time demodulation method based on 5G NR, which adopts FPGA chip, DSP chip and industrial control board, and includes the following steps: FPGA changes sampling from 153.6MHz to 122.88MHz, and the data after FPGA sampling is transmitted to DSP in real time through SRIO, and stored in DDR3 plug-in of DSP, and DSP uses GPIO interrupt response and ping-pong switching to realize real-time reading of data from DDR3 to L2 RAM; DSP uses multi-core and coprocessor to realize real-time demodulation of L2 RAM data, and optimizes the calculation of time offset and frequency offset to realize high-precision demodulation; DSP transmits the demodulation result of physical layer to industrial control board by TCP; DSP realizes self-starting by EMAC mode. The present invention realizes high-precision real-time demodulation of 5G NR, and the signal quality index meets the requirements of 3GPP protocol for 5G NR demodulation, and also has the advantages of low equipment upgrade cost, stable operation, large-capacity storage, and convenient remote upgrade.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of mobile communication technology, and in particular relates to a high-precision real-time demodulation method based on 5G NR. Background Art

[0002] With the popularization of 4G LTE (4th Generation Long Term Evolution) and the rise of 5GNR (5th Generation New Radio), how to add 5G NR functions to the original boards that support 4G LTE has become a problem that needs to be solved urgently.

[0003] Currently, there are few devices that support 5G NR demodulation, the accuracy is low, and real-time demodulation cannot be achieved. Summary of the invention

[0004] The purpose of the present invention is to provide a high-precision real-time demodulation method based on 5G NR, which has a reasonable design, solves the shortcomings of the prior art, and has good effects.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] A high-precision real-time demodulation method based on 5G NR, using an FPGA chip, a DSP chip and an industrial control board, includes the following steps:

[0007] S1, FPGA sampling change: FPGA sampling changes from 153.6MHz to 122.88MHz, without changing the board, to meet the 100MHz bandwidth requirement of 5G NR and be compatible with 4G LTE;

[0008] S2 and SRIO real-time transmission; the data after FPGA sampling is transmitted to DSP in real time through SPIO and stored in DDR3 plug-in of DSP, realizing large-capacity real-time storage of original IQ data; DSP uses GPIO interrupt response and ping-pong switching to realize real-time reading of data from DDR3 to L2 RAM;

[0009] S3, DSP high-precision demodulation; DSP uses multi-core and coprocessor to achieve real-time demodulation of L2 RAM data, and optimizes the calculation of time offset and frequency offset to achieve high-precision demodulation;

[0010] S4; DSP transmits the physical layer demodulation result to the industrial control board using TCP;

[0011] S5 and DSP use EMAC to achieve self-starting.

[0012] Furthermore, in step S1, after FPGA inserts 0 four times for the input data x, according to the properties of convolution, deletes the items with 0, constructs a low-pass filter coefficient h of 122.88 / (153.6*4), takes 1 point every 4 points, divides it into 4 groups of data, marked as h0, h1, h2, h3, convolves the input data x with h0, h1, h2, h3 respectively, obtains y0, y1, y2, y3, sums them up to obtain y, and extracts 1 every 5 points for y to achieve variable sampling from 153.6MHz to 122.88MHz.

[0013] Furthermore, in step S2, the DSP uses the EDMA3 controller to perform ping-pong switching on the data.

[0014] Furthermore, in step S3, for the time offset calculation, on the basis of using power triggering and CP correlation to find the synchronization point, DMRS correlation calculation is added to determine the frame header.

[0015] Furthermore, in step S3, for frequency offset calculation, the traditional CP is first used to calculate the initial frequency offset value and perform compensation, and then the channel estimation value of the DMRS is used to perform correlation calculation, so as to obtain an accurate final frequency offset value.

[0016] Furthermore, in step S5, the DSP program is converted into an eth file using the TI tool chain and stored in the industrial control board. During the startup of the main control program, the eth file is sent to the DSP using the pcsendpkt tool. After the DSP receives the file, it starts the DSP main program from the L2 RAM.

[0017] The beneficial technical effects brought by the present invention are:

[0018] The present invention realizes high-precision real-time demodulation of 5G NR, and the signal quality index meets the requirements of 3GPP protocol for 5G NR demodulation. It also has the advantages of low equipment upgrade cost, stable operation, large-capacity storage, and easy remote upgrade. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below in conjunction with the accompanying drawings:

[0020] Figure 1 It is a working principle diagram of the method of the present invention;

[0021] Figure 2 is the EVM index of the signal received in the present invention; DETAILED DESCRIPTION

[0022] The present invention proposes a high-precision real-time demodulation method based on 5G NR. In order to make the advantages and technical solutions of the present invention clearer and more specific, the present invention is described in detail below in conjunction with specific embodiments.

[0023] like Figure 1 As shown, a high-precision real-time demodulation method based on 5G NR uses an FPGA chip, a DSP chip, and an industrial control board, including the following steps:

[0024] S1, FPGA variable sampling;

[0025] Specifically, the AD chip used in the present invention is ADC12D500RF, and the sampling rate of the initial reception of the FPGA is 153.6MHz, and the 4G LTE sampling rate of 30.72MHz is directly downsampled by 5 times. The maximum single-carrier bandwidth of 5G NR in sub 6GHz is 100MHz, so the sampling rate needs to be converted to 122.88MHz. In order to save costs and facilitate upgrading, it is necessary to perform variable sampling of inserting 4 and extracting 5 without changing the board. However, the maximum frequency of ADC12D500RF is 500MHz, and 153.6MHz becomes 614.4MHz after inserting 4, which exceeds the maximum frequency and cannot be achieved by FPGA. After inserting 0 four times into the input data x, according to the nature of convolution, the items with 0 can be directly deleted to construct a low-pass filter coefficient h of 122.88 / (153.6*4), that is, 153.6*4. The effective bandwidth in MHz is 122.88MHz, and the rest are filtered out. Take 1 of every 4 data of h, and divide it into 4 groups of data, marked as h0, h1, h2, h3. Convolve the input data x with h0, h1, h2, h3 respectively to get y0, y1, y2, y3, and sum them to get y. Then, take 1 of every 5 data of y to achieve variable sampling from 153.6MHz to 122.88MHz.

[0026] S2, SRIO real-time transmission;

[0027] Specifically, the data after FPGA sampling is transmitted to DSP (the DSP chip model used in this embodiment is TMS320TCI6614) in real time through SRIO, and stored in DDR3, which can store up to 2G bytes of raw IQ data in real time. FPGA operates a specific GPIO port, and DSP uses EDMA3 controller to ping-pong data according to the interrupt response of the GPIO port edge, thereby realizing real-time reading from DDR3 to L2 RAM for DSP to achieve high-precision demodulation.

[0028] S3, DSP high-precision demodulation; DSP uses multi-core and coprocessor to achieve real-time demodulation of L2 RAM data, and optimizes the calculation of time offset and frequency offset to achieve high-precision demodulation;

[0029] Specifically, the DSP performs physical layer demodulation on the data of the L2 RAM, and makes full use of the multi-core, FFTC and other coprocessors of the DSP to realize real-time demodulation. The DSP calculates and compensates for the time deviation and frequency deviation indicators, so that the demodulation result can achieve high precision. For the time deviation calculation, power triggering and CP correlation are traditionally used to find the synchronization point, but for the uplink signal of the full subframe, power triggering cannot be used, and CP correlation can only find the beginning of the symbol, and the frame header cannot be located. The present invention adds DMRS correlation calculation on the basis of the traditional method to determine the frame header. For the frequency deviation calculation, CP correlation is traditionally used to calculate the frequency deviation value, but the number of CP points is too small. It is acceptable for correcting large frequency deviations, but it is inaccurate for small frequency deviations. The present invention first uses traditional CP to calculate the rough frequency deviation value, and compensates it, and then uses the channel estimation value of DMRS to perform correlation calculation to obtain an accurate frequency deviation value.

[0030] S4; TCP stable transmission;

[0031] Specifically, after the DSP physical layer demodulation is completed, the demodulation result can be transmitted to the industrial control board by sending a broadcast packet through the EMAC network port. However, considering the stability of network port transmission and the difficulty of developing the main control program, the DSP uses the sgmii and mdio modules to implement the EMAC bottom-level driver, and then uses TI's real-time operating system sysbios and ndk library to achieve TCP stable transmission. 7*24h continuous stress testing is carried out in the environmental laboratory, and TCP transmission is very stable.

[0032] S5, EMAC self-start;

[0033] DSP uses EMAC to achieve self-startup, which is convenient for remote upgrade and in line with the idea of ​​industrialization. The specific operation is to convert the DSP program into an eth file using the TI tool chain and store it in the industrial control board. During the startup of the main control program, this eth file is sent to the DSP using the pcsendpkt tool. After the DSP receives it, it starts the DSP main program from L2 RAM.

[0034] Figure 2 The EVM index of the signal in the present invention is given, wherein the EVM reflects the modulation accuracy of the signal and is an important indicator for measuring signal quality. The EVM value in the embodiment is 0.72%, which fully meets the requirements of the 3GPP protocol.

[0035] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A high-precision real-time demodulation method based on 5G NR, characterized in that: Using FPGA chip, DSP chip and industrial control board, including the following steps: S1, FPGA sampling change: FPGA sampling changes from 153.6MHz to 122.88MHz, which meets the 100MHz bandwidth requirement of 5GNR without changing the board and is compatible with 4G LTE; S2 and SRIO real-time transmission; the data after FPGA sampling is transmitted to DSP in real time through SPIO and stored in DDR3 plug-in of DSP, realizing large-capacity real-time storage of original IQ data; DSP uses GPIO interrupt response and ping-pong switching to realize real-time reading of data from DDR3 to L2 RAM; S3, DSP high-precision demodulation; DSP uses multi-core and coprocessor to achieve real-time demodulation of L2 RAM data, and optimizes the calculation of time offset and frequency offset to achieve high-precision demodulation; S4; DSP transmits the physical layer demodulation result to the industrial control board using TCP; S5, DSP uses EMAC to achieve self-start; In step S3, for the time offset calculation, on the basis of using power triggering and CP correlation to find the synchronization point, DMRS correlation calculation is added to determine the frame header; For frequency offset calculation, the traditional CP is first used to calculate the initial frequency offset value and compensate it, and then the channel estimation value of DMRS is used for correlation calculation to obtain the final frequency offset value; In step S1, the AD chip used is ADC12D500RF. After FPGA inserts 0 four times into the input data x, according to the nature of convolution, deletes the items with 0, constructs a low-pass filter coefficient h of 122.88 / (153.6*4), takes 1 point every 4 points, divides it into 4 groups of data, marked as h0, h1, h2, h3, convolves the input data x with h0, h1, h2, h3 respectively, obtains y0, y1, y2, y3, sums them up to get y, and extracts 1 every 5 points of y to achieve variable sampling from 153.6MHz to 122.88MHz.

2. According to a high-precision real-time demodulation method based on 5G NR according to claim 1, it is characterized in that: In step S2, the DSP uses the EDMA3 controller to ping-pong the data.

3. The high-precision real-time demodulation method based on 5G NR according to claim 1, characterized in that: In step S5, the DSP program is converted into an eth file using the TI tool chain and stored in the industrial control board. During the startup of the main control program, the eth file is sent to the DSP using the pcsendpkt tool. After the DSP receives the file, it starts the DSP main program from the L2 RAM.