Implementation method and device for PRACH preamble sequence detection based on FPGA
By implementing asynchronous RAM and signal processing technology on FPGA, combined with threshold determination, the PRACH preamble sequence detection efficiency and accuracy problems between the UE and the base station in 5G communication are solved, and efficient and stable detection effects are achieved.
Patent Information
- Application Number
- CN202410930176.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-26
AI Technical Summary
In 5G communication, during the random access between the UE and the base station, it is necessary to realize PRACH preamble sequence detection with large capacity, high speed and high stability, and the prior art is difficult to meet the performance requirements in industrial scenarios.
A PRACH preamble sequence detection method and device based on FPGA is designed, and the cross-clock domain conversion of front and back end data is completed through asynchronous RAM, signal processing is used by FFT and IFFT, and frequency domain ZC sequence generation is realized in FPGA, combined with threshold determination. technology to improve the accuracy and efficiency of detection.
It realizes PRACH preamble sequence detection with low complexity, high stability and fast response, reduces false alarm rate and missed detection probability, and is suitable for FPGA chips of different specifications, meeting the requirements of industrial production.
Smart Images

Figure CN119030843A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of mobile communication technology, and in particular to a method and device for realizing PRACH preamble sequence detection based on FPGA. Background Art
[0002] In 5G (5th Generation) mobile communications, UE (User Equipment) must go through processes such as cell search, cell system information acquisition, and random access to access the wireless network.
[0003] After the UE is turned on, the cell search process begins. When the cell search is successful, the UE can achieve downlink synchronization with the cell. The UE can obtain cell system information through downlink transmission. The cell system information includes the PRACH (Physical Random Access Channel) time-frequency resources available for the cell, which in 5G is mainly determined by the prach-ConfigIndex and prach-FrequencyOffset fields in SIB1. These two fields determine the format of the preamble and the optional time-frequency resources for all UEs that need to access the cell.
[0004] The initial step of the random access process is for the UE to send a random access preamble. The UE can generate the preamble by generating a time domain ZC (ZaddoffChu) sequence, FFT (Fast Fourier Transform), mapping time-frequency domain resources, IFFT (Inverse Fast Fourier Transform) and other steps. The preamble is transmitted on the PRACH. Each cell has 64 available preamble sequences, and the UE will select one of them (or specified by the base station) to transmit on the PRACH. By receiving and detecting the preamble, the base station can learn that there is a UE in the cell applying for network access, and can estimate the transmission delay between it and the UE, so that the base station can calibrate the uplink transmission delay. The calibration information is notified to the UE through the TA (Timing Advance) command.
[0005] The reception and detection of preamble at the base station belongs to the physical layer in 5G communication, which can be divided into two parts: the low physical layer (front end) and the high physical layer (back end). The main method is to correlate the locally generated ZC sequence with the received preamble data, and finally perform a threshold determination of the correlation peak in the back end. By determining the correlation peak, the base station can obtain information such as the ID (IDentify), TA, and SINR (Signal to Interference plus Noise Ratio) of the preamble.
[0006] In the 5G era, the requirements of comprehensive network coverage, large UE access volume, high reliability and low latency require that 5G base stations should be small in size while having large capacity, fast response and high stability. In order to meet these performance requirements, in industrial scenarios, signal processing of some channels such as PRACH and PUSCH (Physical Uplink Shared Channel) is often deployed on FPGA to achieve hardware acceleration. Therefore, designing a PRACH preamble sequence detection method and device that is efficient, reliable and has low computational complexity and can be implemented on FPGA (Field Programmable Gate Array) is crucial for performance improvement and overhead optimization of 5G base station systems. Summary of the invention
[0007] The purpose of the present invention is to solve the above problems. A method and device for implementing PRACH preamble sequence detection based on FPGA are designed. The method and device can quickly and effectively complete the preamble detection steps in the random access process, and solve the technical difficulties of random access between UE and base station in 5G communication, such as large capacity, high rate and high stability.
[0008] A first aspect of the present invention provides a method for implementing PRACH preamble sequence detection based on FPGA, and the method for implementing PRACH preamble sequence detection based on FPGA comprises the following steps:
[0009] The time domain IQ data y of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain RAM0 and RAM1 are two RAMs, p is the index of the corresponding symbol, and m is the subcarrier index;
[0010] Asynchronous RAM is used to complete the cross-clock domain conversion of front-end and back-end data, and the PRACH effective subcarrier IQ data in the frequency domain After asynchronous RAM0, RAM1 and RAM2 write and read, the back-end input is obtained. Wherein, v is the index of the corresponding symbol, and m is the subcarrier index;
[0011] Process the input data obtained through the front-end and back-end transfer to obtain the preamble detection result preamble_indication, and output it in the form of a 32-bit wide register group;
[0012] The configuration parameter prach_config of PRACH is processed by the CPU and sent to the FPGA via PCIE. The detection result preamble_indication obtained by the correlation peak detection is processed across the clock domain and reported to the CPU via PCIE.
[0013] Optionally, in a first implementation of the first aspect of the present invention, the time domain IQ data y of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain include:
[0014] The time domain IQ data of the received signal y receive (t) When writing to RAM, the write address will move the spectrum according to the subcarrier index k. When k<0.5·(N FFT -N padding ), write address is addr w =k+0.5·(N FFT -N padding ), when k≥0.5·(N FFT +N padding ), the write address is addr w =k-0.5·(N FFT +N padding ), where N FFT is the number of FFT points, N padding is the number of points of the guard interval filled with 0;
[0015] Extract the PRACH valid subcarrier according to the subcarrier offset. r ≤N ZC +N offset -1, the read enable is assigned a value of 1, where N offset =prach freq_offset ·12+prach k , prach freq_offset is the number of resource blocks of PRACH subcarrier frequency offset, prach k For subcarriers in prach freq_offsetThe number of resource units to offset from the base.
[0016] Optionally, in a second implementation of the first aspect of the present invention, the input data obtained through the front-end and back-end transfer is processed to obtain the preamble detection result preamble_indication, and output it in the form of a register group with a 32-bit width, including:
[0017] Divide the PDP into multiple windows, search for the peak value in each window, remove the peak point and the adjacent left and right points on the PDP, calculate the average power of the remaining points, and obtain the average noise power;
[0018] The signal-to-interference-to-noise ratio is calculated by the peak and average noise power, and the strong peak and weak peak are determined. The strong peak is the preamble detected, and the weak peak is determined again to remove the false alarm and detect the preamble.
[0019] According to the time domain results of each window Power i (m),i=0,1,...,N win -1 and preamble detection status preamble status (i) Output timing advance estimate TA estimation , and generate preamble_indication.
[0020] Optionally, in a third implementation of the first aspect of the present invention, dividing the PDP into a plurality of windows, searching for a peak value in each window, removing the peak point and one point on the left and right adjacent to the peak point on the PDP, and calculating the average power of the remaining points to obtain the average noise power includes:
[0021] According to the cyclic shift value N cs and the number of windows N win , calculate the root sequence i th The start and end positions of the window i ,end i ) and length L i , so win i =[start i ,end i ], Power(n) is divided into multiple windows, and its corresponding i th The PDP of the window is Power i ,(start i ,end i ) is calculated as follows:
[0022]
[0023] Where Nzc =139or839 corresponds to the number of points N for expansion z ' c =256or1024, when N cs =139or839, end i =255or1023,i=0;
[0024] Traverse i th The value in the window, find i th The maximum value of the window M i , with the maximum value M i Corresponding position index i ;
[0025] i th The average noise power of the window is calculated as:
[0026]
[0027] Compute the average noise power of all windows on the root sequence:
[0028]
[0029] Optionally, in a fourth implementation manner of the first aspect of the present invention, calculating the signal to interference noise ratio by using the peak value and the average noise power, and determining the strong peak and the weak peak, includes:
[0030] According to the average noise power P and i th The maximum value of the window M i , calculate the linear sinr i :sinr i =M i / P, traverse all windows and find sinr i The maximum value of sinr max , use the threshold set (T1, T2) to obtain i th Detection status of preamble in the window:
[0031]
[0032] When FPGA makes threshold determination, there are three conditions: Condition 1: if sinr i >8·T1, condition 2: ifsinr i ∈[T1,8·T1], Condition 3: if sinr i >sinr max T2;
[0033] First, condition 1 is judged. If condition 1 is met, the correlation peak will be judged as a strong peak, that is, the preamble is detected. If condition 2 is met, the correlation peak will be judged as a weak peak. If both conditions are not met, it is considered that there is no correlation peak. The weak peak will be judged again by condition 3. If condition 3 is met, the weak peak is indeed the sent preamble, otherwise it is a false alarm caused by noise or interference from other root sequences. When the preamble status (i)>0, then it is considered that th A preamble was detected within the window.
[0034] Optionally, in a fifth implementation of the first aspect of the present invention, the detection result preamble_indication includes at least the preamble ID, the timing advance TA estimation (i) Average noise power P and its corresponding frequency domain and time domain scenarios.
[0035] The second aspect of the present invention provides a method and device for implementing PRACH preamble sequence detection based on FPGA, wherein the implementation device for PRACH preamble sequence detection based on FPGA includes a PRACH front-end processing module, a front-end and back-end transfer module, a PRACH back-end processing module, a cross-clock domain register and a configuration port module, wherein
[0036] PRACH front-end processing module, used to process the time domain IQ data of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain And transmit the output data to the front-end and back-end transfer modules, where RAM0 and RAM1 are 2 RAMs, p is the index of the corresponding symbol, and m is the subcarrier index;
[0037] The front-end and back-end transfer module is used to complete the cross-clock domain conversion of the front-end and back-end data using asynchronous RAM, and the output of the PRACH front-end processing module After asynchronous RAM0, RAM1 and RAM2 write and read, the back-end input is obtained. Wherein, v is the index of the corresponding symbol, and m is the subcarrier index;
[0038] The PRACH back-end processing module is used to process the input data obtained from the front-end and back-end transfer modules, obtain the preamble detection result preamble_indication, and output it in the form of a 32-bit wide register group;
[0039] The cross-clock domain register and configuration port module is used for the configuration parameter prach_config of PRACH to be processed by the CPU and then sent to the FPGA via PCIE. The detection result preamble_indication obtained by the correlation peak detection is processed across the clock domain and reported to the CPU via PCIE.
[0040] Optionally, in a first implementation of the second aspect of the present invention, the PRACH front-end processing module includes a frame and time slot determination submodule, a cyclic prefix removal submodule, a data FIFO cache submodule, and an FFT and effective subcarrier extraction submodule.
[0041] Optionally, in a second implementation of the second aspect of the present invention, the PRACH back-end processing module includes a local sequence generation preparation submodule, a local sequence generation submodule, a correlation and IFFT submodule for received data and local sequence, and a correlation peak detection submodule.
[0042] The present invention provides a method and device for implementing PRACH preamble sequence detection based on FPGA. Compared with other methods and devices for implementing preamble sequence detection, the present invention has the following advantages:
[0043] 1. A complete set of PRACH preamble sequence detection devices has been built at the FPGA engineering level, which has low complexity, high stability and feasibility. It is suitable for FPGA chips of various specifications and can be specifically applied to the development scenarios of wireless communication base stations such as hardware acceleration;
[0044] 2. Before performing PRACH preamble sequence detection, the program in the FPGA will identify the current frame number and time slot number according to the 3GPP technical standards and the configuration parameters sent by the CPU. The subsequent process will be carried out only when it is determined that the PRACH signal is transmitted in the current frame and time slot. This can reduce the workload of the FPGA program processing to a certain extent, improve the efficiency of the system, and reduce the false alarm rate;
[0045] 3. When generating a local frequency domain ZC sequence, the existing IP in the FPGA (the two IP cores, FFT and DFT) cannot support 139-point and 839-point Fourier transforms. According to the definition of the ZC sequence and the characteristics of the periodic cycle, the DFT operation is transformed. Thus, the present invention realizes the generation of a relatively simple frequency domain ZC sequence in the FPGA. The basic method is to transform the root u and u with respect to the modulus N ZC The multiplicative inverse element q of is stored in the form of a table in advance. When generating the sequence, the root value index in the configuration parameter is used to retrieve (u,q). Then, according to the sequence generation formula X u (k) = x u * (qk)Xu (0) Generate a local frequency domain ZC sequence, which not only solves the problem that the FPGA cannot perform FFT / IFFT operations on prime numbers such as 139 and 839, but also reduces the complexity of the calculation;
[0046] 4. In the threshold determination of correlation peak detection, the processing logic in the FPGA will perform strong and weak peak classification and secondary threshold determination, which can reduce the probability of false alarms and missed detections and greatly improve the accuracy of preamble detection;
[0047] 5. PRACH preamble sequence detection is divided into two independent parts, the front end and the back end. Therefore, in actual application, the front end and the back end can be deployed in a distributed manner. For example, when the CPU performance is strong, the front end can be deployed on the FPGA and the back end can be deployed on the CPU. That is, the front end part of the PRACH preamble sequence detection is hardware accelerated by the FPGA alone, so as to reasonably utilize hardware resources.
[0048] The present invention can effectively detect the PRACH preamble code from the received signal data, and the design is suitable for FPGA chips of different specifications, with low complexity, high stability and fast response. While ensuring high accuracy, it requires less computing resources and can meet the requirements of industrial production. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiment.The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention.
[0050] Figure 1 A schematic diagram of a first embodiment of a method and device for implementing PRACH preamble sequence detection based on FPGA provided in an embodiment of the present invention;
[0051] Figure 2 A schematic diagram of a second embodiment of a method and device for implementing PRACH preamble sequence detection based on FPGA provided in an embodiment of the present invention;
[0052] Figure 3 A schematic diagram of an embodiment of correlation peak detection provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0053] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0054] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 A schematic diagram of a first embodiment of a method and device for implementing PRACH preamble sequence detection based on FPGA provided in an embodiment of the present invention, the device specifically includes:
[0055] PRACH front-end processing module M1: time domain IQ (In-phase and Quadrature-phase) data of the received signal receive (t) After being processed by the front-end module, the output data is the PRACH effective subcarrier IQ data in the frequency domain RAM0 and RAM1 refer to two RAMs (Random Access Memory), p is the index of the corresponding symbol, and m is the subcarrier index (the short format contains 139 subcarriers and the long format contains 839 subcarriers). The output of the front end will eventually be given to the front-end and back-end transfer modules.
[0056] Front-end and back-end transfer module M2: Since the PRACH front-end and back-end are relatively independent, the front-end and back-end modules in the FPGA project can also be distributed and arranged on different chip processing boards, resulting in different clock frequencies of the front-end and back-end modules. Therefore, it is necessary to convert the front-end and back-end data across clock domains, which is completed using asynchronous RAM to convert the frequency domain IQ data in the front-end clock domain to the frequency domain IQ data in the back-end clock domain. The output Y′ of the front end RAMθ (p,m),θ=0,1After passing through the transfer module, after being written and read by asynchronous RAM0, RAM1 and RAM2, the input of the back end is obtained v is the index of the corresponding symbol, and m is the subcarrier index (the short format contains 139 subcarriers, and the long format contains 839 subcarriers).
[0057] PRACH back-end processing module M3: After the input from the transfer module is processed by the back-end module, the preamble detection result preamble_indication is output in the form of a 32-bit register group, which contains preamble ID, TA, SINR and other information, named PREAMBLE_INDICATION_X, and X is numbered from 0 to 49, as shown in Table 1.
[0058] Cross-clock domain register and configuration port module M4: The configuration parameter prach_config of PRACH is issued in this module via PCIE (Peripheral Component Interconnect Express, peripheral device high-speed connection standard). Because the configuration parameters are fixed after the cell initialization is completed, there is no need for cross-clock domain processing. After the various information obtained by the correlation peak detection, namely preamble_indication, is processed across the clock domain, it is reported to the CPU (Central Processing Unit, core processor) via PCIE.
[0059] In this embodiment, in the PRACH front-end processing module, the discrimination between frames and time slots is determined by the prach_frame_x, prach_frame_y and prach_sfn_bit_map fields in the configuration parameter prach_config. By mapping to a table with a maximum width of 80 bits, prach_sfn_bit_map can specifically describe which time slots have sent PRACH signals. By discriminating between frames and time slots, the present invention can clearly distinguish whether a PRACH signal is sent in a certain frame and a certain time slot, avoiding the situation of performing PRACH preamble detection on each time slot on each frame, reducing the workload of FPGA program processing, and suppressing the occurrence of false alarms to a certain extent.
[0060] In this embodiment, in the PRACH front-end processing module, FFT and effective subcarrier extraction adopt a dual RAM ping-pong relay processing method. For example, when RAM No. 0 is still accessing the data of a certain symbol, if the data of the next symbol has arrived, RAM No. 1 can immediately perform the access operation. This ping-pong processing method can effectively avoid the situation of data timing errors during high-speed processing. When performing FFT on preambles in the formats of 0, 1, 2, and 3, because there are too many points, it is necessary to perform multi-level downsampling and low-pass filtering operations first to reduce the number of points to a level that can be used for FFT operations.
[0061] In this embodiment, in the front-end and back-end transfer modules, three RAMs are used to access the output data of the front end. This is to prepare for the symbol reorganization operation in the back end. For example, in the B4 format, In the example, the symbol indexes in RAM0 are: 0, 2, 4, 6, 8, 10; the symbol indexes in RAM1 are: 1, 3, 5, 7, 9, 11. After accessing the three RAMs of the transfer module, we get The symbol indexes in RAM0 are: 0, 3, 6, 9; the symbol indexes in RAM1 are: 1, 4, 7, 10; the symbol indexes in RAM2 are: 2, 5, 8, 11. Through this operation, the program can perform symbol reorganization in a parallel manner in the back-end processing module.
[0062] In this embodiment, in the PRACH backend processing module, the local sequence generation preparation is mainly to obtain the root value u=1,2,...,N of the generated root sequence according to the logical index of the ZC sequence root value. ZC -1, and u modulo N ZC The multiplicative inverse element q of ZC -1. It is worth noting that in this patent, in order to reduce the complexity of correlating the received data with the local sequence, a correlation operation is performed in the frequency domain. Therefore, it is necessary to generate a local frequency domain ZC sequence. However, the number of points of the ZC sequence is 139 points or 839 points. In FPGA, this type of prime number cannot be directly fast Fourier transformed through the FFT IP (Intellectual Property) core, and the proprietary DFT (Discrete Fourier transform) IP core does not support 139-point and 839-point Fourier transforms. Therefore, it is impossible to obtain a frequency domain ZC sequence by generating a time domain ZC sequence and then performing a Fourier transform. Of course, it can be expanded by zero padding to convert it into a number of points that can be used for FFT in FPGA, but this method will increase the complexity of the program while reducing the accuracy of the value of the generated local frequency domain ZC sequence. Therefore, this patent adopts another way to generate a frequency domain ZC sequence. If the time domain ZC sequence x u The expression of (n) is Then x u (n) The corresponding frequency domain transform X u The expression of (k) is X u (k) = x u * (qk)X u (0),k=0,1,...,N ZC-1. The u value and the corresponding q value will be pre-stored. By generating the frequency domain ZC sequence in this way, the workload of FPGA processing can be greatly reduced, and a more accurate local sequence and related results can be obtained, thereby improving the accuracy of program calculation.
[0063] In this embodiment, in the correlation peak detection of the PRACH backend processing module, the program will be based on the cyclic shift value N in the prach_config configuration parameter. CS , divide the relevant power value into multiple windows, search for the peak value in each window and calculate the noise floor; the ratio of the peak value to the noise floor, i.e., SINR, and the threshold set of (threshold 1, threshold 2) are subjected to two threshold determinations. After the first threshold determination, two types of relevant peaks can be obtained: strong peaks and weak peaks (which may include false alarms). The second threshold determination will perform a threshold determination on the weak peak, combining the maximum SINR obtained by traversing the relevant power value with threshold 2 to eliminate the interference caused by false alarms. The method of two threshold determinations in this patent can greatly reduce false alarms caused by interference such as noise or preambles on other root sequences, and can accurately detect preambles transmitted in PRACH. It is worth noting that for preambles of different formats, the preset threshold sets (threshold 1, threshold 2) are also different. In the FPGA program of this patent, the corresponding preset threshold set will be selected according to the format of the preamble to improve the pertinence and accuracy of the threshold determination.
[0064] In this embodiment, the number of root sequences required, prach_root_num, is indicated in the configuration parameter prach_config sent, which is determined by the maximum number of preambles sent, prach_preamble_num, and the number of preambles that can be configured on each root sequence, prach_window_num_per_root. If multiple root sequences are required, the local sequence generation preparation in the back-end module, the generation of local sequences, the correlation and IFFT of the received data and the local sequence, and the correlation peak detection submodules need to be looped multiple times.
[0065] See also Figure 2 , a schematic diagram of a second embodiment of a method and device for implementing PRACH preamble sequence detection based on FPGA provided in an embodiment of the present invention, the method comprising:
[0066] Submodule in the front-end module: frame and time slot discrimination. Its input is the IQ data y of the time domain received signal of 1 time slot receive (t), the width is 32 bits, the upper 16 bits are the real part, and the lower 16 bits are the imaginary part. The output is the IQ data y of the time domain received signal containing the PRACH signal prach(t), the data format is the same as the input. According to the 3GPP protocol standard, PRACH signals are only sent on specified frames and time slots, which can be obtained by looking up Table 6.3.3.2-2, Table 6.3.3.2-3 and Table 6.3.3.2-4 of TS38.211. In this patent, the fields prach_frame_x, prach_frame_y and prach_sfn_bit_map in the prach_config parameter define the frames and time slots for sending PRACH signals:
[0067]
[0068] Among them, n f is the frame number of the currently received signal, x, y are prach_frame_x, prach_frame_y respectively. slot is the time slot number of the current received signal, It is the specified time slot number in prach_sfn_bit_map. For example, in a PRACH signal with a subcarrier spacing of 30kHz, if prach_config_index = 218 and it is in FDD (Frequency Division Duplexing) mode, then by querying Table 6.3.3.2-2 of TS38.211, it can be obtained that the subframe numbers for sending PRACH signals are 1, 3, 5, 7, 9, and PRACH signals are sent in both time slots within a subframe, then the corresponding prach_sfn_bit_map is 0011_0011_0011_0011_0011. Judgment The method is: shift prach_sfn_bit_map to the left, and the shift value is n slot , if the value of the highest bit of prach_sfn_bit_map is 1 after shifting, then
[0069] Submodule 2 in the front-end module: removes the cyclic prefix. Its input is the IQ data y of the time domain received signal containing the PRACH signal prach (t), the output is the time domain signal IQ data y(t) after removing the CP (Circular Prefix). The received PRACH signal includes a cyclic prefix T cp and one or more sequences T seq , the removal of CP is as follows:
[0070] y(t)=y receive (t+T cp ),t=0,1,..,T seq ·N sym-1
[0071] Among them, N sym is the number of symbols, such as N in B4 format sym = 12, which can be found in Table 5.7.1-1 of TS36.211 and Table 6.3.3.1-1 and 6.3.3.1-2 of TS38.211. The length of CP is T cp It is pre-calculated by the CPU and sent by PCIE, and does not require FPGA to calculate locally, which can reduce the calculation length and save resource overhead.
[0072] Submodule 3 in the front-end module: data FIFO buffer. Its input is the IQ data y(t) of the time domain signal after CP removal, and its output is the continuous time domain IQ data y in 2048 clock units. fifo (t). Due to air interface reception and crossing clock domains, y receive (t) is not necessarily continuous, which leads to y(t) is not necessarily continuous, so it is accessed through the synchronous FIFO, cached and converted into y continuous within 2048 clock units. fifo (t).
[0073] Submodule 4 in the front-end module: FFT and effective subcarrier extraction. Its input is the time domain IQ data y fifo (t), the output is the IQ data of the frequency domain PRACH signal Get the number of FFT points N according to the prach_config parameter and perform FFT on the time domain signal. It is worth noting that in the long format, multi-level downsampling filtering is required before FFT.
[0074] Y(s,k)=FFT(y fifo (t))
[0075] s=0,1,...,S-1,k=0,1,...,N-1
[0076] Among them, s is the symbol index, S is the number of symbols in the valid sequence under the preamble format, k is the index of the subcarrier, and N is the number of FFT points.
[0077] The extraction of PRACH effective subcarriers is carried out by RAM writing and reading, and the steps are divided into two steps: 1. Shifting the spectrum; 2. Extracting the prach effective subcarriers according to the subcarrier offset.
[0078] In step 1, when IQ data is written to RAM, the write address will perform different operations according to the subcarrier index k:
[0079] When k<0.5·(N FFT -N padding), write address is addr w =k+0.5·(N FFT -N padding ), when k≥0.5·(N FFT +N padding ), the write address is addr w =k-0.5·(N FFT +N padding ). Where N FFT is the number of FFT points, N padding is the number of points of the guard interval filled with 0. After the above operation, the guard interval is removed while the FFT spectrum is shifted.
[0080] In step 2, the following operations will also occur when data is read out of RAM:
[0081] When the address addr is read r ≤N ZC +N offset -1, the read enable is assigned a value of 1. offset =prach freq_offset ·12+prach k , prach freq_offset is the number of RBs (ResourceBlocks) of PRACH subcarrier frequency offset, prach k For subcarriers in prach freq_offset The number of REs (ResourceElements) offset based on the value.
[0082] In addition, a ping-pong operation is adopted in the PRACH effective subcarrier extraction submodule, and two RAMs (RAM0 and RAM1) are set up in a relay manner for the extraction of PRACH frequency domain IQ data. For example, when the format is B4 and the start symbol is 0, RAM0 is responsible for writing and reading IQ data with symbol index s=0,2,4,6,8,10, and RAM1 is responsible for writing and reading IQ data with symbol index s=1,3,5,7,9,11, and the data of consecutive symbols are processed in a relay manner. This can avoid timing disorder when the IQ data flows in too fast.
[0083] After spectrum shifting and removing the guard interval:
[0084] Y′(s,k′)=FFT shift (Y(s,k))
[0085] k′=0,1,...,N FFT -N padding -1
[0086] After PRACH subcarrier extraction, the signal is represented as:
[0087]
[0088] Where k' is the subcarrier index, p indicates the symbol index of the PRACH subcarrier read from RAM0 and RAM1 in this module, m is the PRACH subcarrier index, N ZC is the ZC sequence length of PRACH.
[0089] The input of the front-end and back-end transfer modules is PRACH subcarrier data Output is PRACH subcarrier data Since the front-end and back-end modules are not in the same clock domain, asynchronous RAM is required for cross-clock domain processing in the front-end and back-end transfer modules. At the same time, 1) the PRACH subcarrier data read out from RAM0 and RAM1 of the front-end module needs to be read according to the timing of the back-end local sequence generation; 2) the frequency domain IQ data in the back-end module needs symbol reorganization; therefore, RAM0, RAM1, and RAM2 are used in the back-end to divide the data of different symbols into 3 groups; the following is explained in B4 and C2 formats:
[0090]
[0091] There are two cases in the C2 format, that is, when the number of time domain PRACH occasions is 1 or 2, S=4 or 8 is corresponding.
[0092] According to the above formula, according to different preamble formats, and Re-partition by symbol index into and v is the symbol index and m is the PRACH subcarrier index.
[0093] Submodule 1 in the backend module: local sequence generation preparation. Its input is the root index i of the local sequence, i=i init ,i init +1,...,i init +N root -1, the output is the root u of the local sequence and u modulo N ZC The multiplicative inverse q, that is, (q*u)modN ZC ≡1. Where i init is the initial index value, N root is the number of root sequences to be configured; the rooted u and the corresponding q values have been pre-generated and stored. According to the rooted index i, u and q are generated from the pre-stored method.
[0094] Submodule 2 in the backend module: Generate local sequence. Its input is the root u and corresponding q value of the local sequence, and the output is the local ZC sequence X in the frequency domain u (k). If the time domain ZC sequence is x u (n), as follows:
[0095]
[0096] x u The frequency domain transformation of (n) is X u (k) is still a ZC sequence, using the time domain ZC sequence x u The periodic characteristics of (n) can be used to describe the DFT[x u (n)] to achieve the purpose of simplifying the calculation. The derivation process is as follows:
[0097] Let u·q=m·N ZC +1, where q∈[1,N ZC -1] and is an integer, and m is an integer greater than or equal to 0, then we can get:
[0098]
[0099] Substitute the above formula into X u (k) = DFT[x u (k)], we can get:
[0100]
[0101] in:
[0102]
[0103] Therefore, in FPGA, when the root is u, q is u modulo N ZC When the multiplicative inverse element of the frequency domain ZC sequence X u (k) can be generated using the following formula:
[0104]
[0105] Submodule 3 in the backend module: Correlation and IDFT of received data and local sequence. Its input is PRACH subcarrier data and the frequency domain local ZC sequence X u (k), the output is the power form of the time domain correlation result corresponding to the root u, namely PDP (Power Delay Profile): Power u (n).
[0106] First, symbol reorganization is performed according to the preamble format. Here, only B4 and C2 formats are considered:
[0107]
[0108] Perform frequency domain correlation:
[0109]
[0110] Where v is the symbol index, m is the PRACH subcarrier index, m = 0, 1, ..., N ZC -1, u is the root of the corresponding ZC sequence, () * represents the conjugate operation, and ()·() is the vector dot product operation.
[0111] Next, the frequency domain correlation results are IFFT-ed to obtain the time domain correlation results, because N ZC =139 or 839 is a prime number. In FPGA, it is not possible to directly use the FFT IP core for IFFT, so it is necessary to fill it with 0 at the end and expand it to the nearest integer power of 2, expanding 139 or 839 points to 256 or 1024 points. It should be noted that in the subsequent correlation peak detection, the position offset of the detection window caused by the expansion must be taken into account.
[0112]
[0113] After calculating the power of the time domain correlation result, the PDP of multiple symbols is obtained:
[0114] Power u (v,n)=|c u ′(v,n) 2
[0115] The PDP of multiple symbols is averaged to obtain the final PDP:
[0116]
[0117] Among them, v max is the maximum symbol index, determined by the preamble format. In B4 format, v max =3, when the number of domain occasions is 1 in C2 format, v max =1, when the number of domain occasions is 2, v max =3.
[0118] Submodule 4 in the backend module: Correlation peak detection. Its input is Power u (n), the output is the detection result of the correlation peak preamble_indication. The basic principle is to search for the correlation peak in each search window on the PDP, and make a judgment through the threshold set and the maximum SINR. When the judgment is passed, it is considered that the preamble is detected; the correlation peak detection is as follows Figure 3 As shown, it is divided into 4 steps: peak search and noise power estimation, correlation peak detection, TA estimation, and generation of preamble_indication.
[0119] Divide the PDP into multiple windows, search for the peak value in each window, remove the peak point and the adjacent left and right points on the entire PDP (including all windows), and calculate the average power of the remaining points, which is the average noise power. The signal-to-interference-noise ratio is calculated by the peak value and the average noise power. First, perform threshold judgment 1 to obtain two types of correlation peaks: strong peaks and weak peaks, and set the status to 2 and 1. Strong peaks are directly regarded as valid correlation peaks, that is, preambles are detected, while weak peaks need to perform threshold judgment 2 again. After performing threshold judgment 2 and removing false alarms, weak peaks are considered to have detected preambles.
[0120] Because the correlation peak detection step is performed in the PDP on one root sequence, the repeated discussion of different root sequences u is omitted in the following process. u (n) is equivalent to Power(n).
[0121] Peak search and noise power estimation. Its input is Power(n), cyclic shift value N cs and the number of windows N win , the output is the estimated noise power of each window on the root sequence: P i .
[0122] First, according to N cs With N win , and the expansion case (from 139 or 839 points to 256 or 1024 points), calculate the root sequence i th The start and end positions of the window i ,end i ) and length L i . i =[start i ,end i ], Power(n) is divided into multiple windows, and its corresponding i th The PDP of the window is Power i . (start i ,end i ) is calculated as follows:
[0123]
[0124] Where N zc =139or839 corresponds to the number of points N for expansion z ' c=256or1024, it is worth noting that when N cs =139or839, end i =255or1023,i=0.
[0125] Traverse i th The value in the window, find i th The maximum value of the window M i , and its corresponding position index Index i .
[0126] Next, th The average noise power of the window can be calculated as:
[0127]
[0128] Finally, calculate the average noise power of all windows on the root sequence:
[0129]
[0130] The correlation peaks in each search window are determined. The input is i th The maximum value of the window M i and the estimated noise power P, the output is i th Window preamble detection status (values are 0, 1, 2): preamble status (i), state 0 represents no effective correlation peak, state 1 represents a weak peak, and state 2 represents a strong peak.
[0131] According to the calculated average noise power P and i th The maximum value of the window M i , calculate the linear sinr i :sinr i =M i / P. Traverse all windows and find sinr i The maximum value of sinr max . Use the threshold set (T1, T2) to obtain i th Detection status of preamble in the window:
[0132]
[0133] When FPGA uses the above formula to make threshold judgment, there will be 3 conditions: Condition 1: if sinr i >8·T1, condition 2: if sinr i ∈[T1,8·T1], Condition 3: if sinr i >sinr max ·T2.
[0134] The judgment process is as follows: first, condition 1 will be judged. If condition 1 is met, the correlation peak will be judged as a strong peak, that is, the preamble is detected. If condition 2 is met, the correlation peak will be judged as a weak peak. If both conditions are not met, it is considered that there is no correlation peak. For weak peaks, condition 3 will be judged again. If condition 3 is met, the weak peak is indeed the sent preamble. Otherwise, it is a false alarm caused by noise or interference from other root sequences. This process is called 2-threshold judgment in this patent. Finally, when the preamble status (i)>0, then it is considered that th The preamble is detected in the window. Note that the threshold set (T1, T2) is set differently in different preamble formats.
[0135] TA estimation: Its input is the time domain correlation result Power of each window i (m),i=0,1,...,N win -1 and preamble detection status preamble status (i), the output is the timing advance estimate TA estimation For i th If a preamble is detected, its peak position is used to calculate TA:
[0136]
[0137] Where i satisfies the preamble status (i)>0.
[0138] Generate preamble_indication: its input is preamble status Window index i of the correlation peak where (i)>0 th That is, the preamble ID and the timing advance TA estimation (i), peak value M i , the estimated noise power P, and the corresponding frequency domain occasion freq_occ_id(i) and time domain occasion time_occ_id(i), are output as preamble_indication_x, x=0,1,...,49. The preamble detection result, frame number, and time slot number are reported in the form of a 32-bit wide register, as shown in Table 1;
[0139] Table 1 Definition of Preamble detection result reporting register
[0140]
[0141]
[0142] In Table 1, preamble_num is the number of preambles detected, N detected_preamble , which is defined as follows:
[0143]
[0144] prach_notify_done is a counter. When the counter increases by 1, it indicates that the preamble detection of a slot carrying a PRACH signal has been completed.
[0145] The input of the cross-clock domain register and configuration port module is reg_wr_en, reg_wr_data, reg_addr, reg_rd_en and preamble_indication_x, x=0,1,...,49, and the output is prach_config_x, x=0,1,...,12, reg_rd_data. The data transmission between this module and PCIE is carried out in the form of a 32-bit wide register. By matching the register address (reg_addr) with the write timing (reg_wr_en), the configuration parameter prach_config is correctly obtained from PCIE. Because the configuration parameters are fixed, cross-clock domain processing is not required. Preamble_indication_x is the result reported after preamble detection. The detection results of different slots will change, and cross-clock domain processing is required. After processing, cdc_preamble_indication_x is obtained, and the clock frequency is crossed from the backend domain to the PCIE domain. By matching the register address (reg_addr) with the register read enable (reg_rd_en), cdc_preamble_indication_x is assigned to reg_rd_data, and the preamble_indication is finally reported.
[0146] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. A method for implementing PRACH preamble sequence detection based on FPGA, characterized in that: The implementation method of the PRACH preamble sequence detection based on FPGA comprises the following steps: The time domain IQ data y of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain RAM0 and RAM1 are two RAMs, p is the index of the corresponding symbol, and m is the subcarrier index; Asynchronous RAM is used to complete the cross-clock domain conversion of front-end and back-end data, and the PRACH effective subcarrier IQ data in the frequency domain After asynchronous RAM0, RAM1 and RAM2 write and read, the back-end input is obtained. Wherein, v is the index of the corresponding symbol, and m is the subcarrier index; Process the input data obtained through the front-end and back-end transfer to obtain the preamble detection result preamble_indication, and output it in the form of a 32-bit wide register group; The configuration parameter prach_config of PRACH is processed by the CPU and sent to the FPGA via PCIE. The detection result preamble_indication obtained by the correlation peak detection is processed across the clock domain and reported to the CPU via PCIE.
2. The method for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 1, characterized in that: The time domain IQ data y of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain include: The time domain IQ data of the received signal y receive (t) When writing to RAM, the write address will move the spectrum according to the subcarrier index k. When k<0.5·(N FFT -N padding ), write address is addr w =k+0.5·(N FFT -N padding ), when k≥0.5·(N FFT +N padding ), the write address is addr w =k-0.5·(N FFT +N padding ), where N FFT is the number of FFT points, N padding is the number of points of the guard interval filled with 0; Extract the PRACH valid subcarrier according to the subcarrier offset. r ≤N ZC +N offset -1, the read enable is assigned a value of 1, where N offset =prach freq_offset ·12+prach k , prach freq_offset is the number of resource blocks of PRACH subcarrier frequency offset, prach k For subcarriers in prach freq_offset The number of resource units to offset from the base.
3. The method for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 1, characterized in that: The input data obtained through the front-end and back-end transfer is processed to obtain the preamble detection result preamble_indication, and output it in the form of a 32-bit wide register group, including: Divide the PDP into multiple windows, search for the peak value in each window, remove the peak point and the adjacent left and right points on the PDP, calculate the average power of the remaining points, and obtain the average noise power; The signal-to-interference-to-noise ratio is calculated by the peak and average noise power, and the strong peak and weak peak are determined. The strong peak is the preamble detected, and the weak peak is determined again to remove the false alarm and detect the preamble. According to the time domain results of each window Power i (m),i=0,1,...,N win -1 and preamble detection status preamble status (i) Output timing advance estimate TA estimation , and generate preamble_indication.
4. The method for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 3, characterized in that: The PDP is divided into a plurality of windows, a peak value in each window is searched, and after removing the peak point and one point on the left and right adjacent to the PDP, the average power of the remaining points is calculated to obtain the average noise power, including: According to the cyclic shift value N cs and the number of windows N win , calculate the root sequence i th The start and end positions of the window i ,end i ) and length L i , so win i =[start i ,end i ], Power(n) is divided into multiple windows, and its corresponding i th The PDP of the window is Power i ,(start i ,end i ) is calculated as follows: Where N zc =139 or 839 corresponding to the number of points N' zc =256 or 1024, when N cs =139 or 839, end i =255 or 1023, i=0; Traverse i th The value in the window, find i th The maximum value of the window M i , with the maximum value M i Corresponding position index i ; i th The average noise power of the window is calculated as: Compute the average noise power of all windows on the root sequence:
5. The method for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 3, characterized in that: The signal to interference noise ratio is calculated by peak and average noise power, and strong peaks and weak peaks are determined, including: According to the average noise power P and i th The maximum value of the window M i , calculate the linear sinr i :sinr i =M i / P, traverse all windows and find sinr i The maximum value of sinr max , use the threshold set (T1, T2) to obtain i th Detection status of preamble in the window: When FPGA makes threshold determination, there are three conditions: Condition 1: if sinr i >8·T1, condition 2: if sinr i ∈[T1,8·T1], Condition 3: if sinr i >sinr max T2; First, condition 1 is judged. If condition 1 is met, the correlation peak will be judged as a strong peak, that is, the preamble is detected. If condition 2 is met, the correlation peak will be judged as a weak peak. If both conditions are not met, it is considered that there is no correlation peak. The weak peak will be judged again by condition 3. If condition 3 is met, the weak peak is indeed the sent preamble, otherwise it is a false alarm caused by noise or interference from other root sequences. When the preamble status (i)>0, then it is considered that th A preamble was detected within the window.
6. A method for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 1, characterized in that: The detection result preamble_indication includes at least the preamble ID and the timing advance TA estimation (i) Average noise power P and its corresponding frequency domain and time domain scenarios.
7. A device for implementing PRACH preamble sequence detection based on FPGA, characterized in that: The FPGA-based PRACH preamble sequence detection implementation device includes a PRACH front-end processing module, a front-end and back-end transfer module, a PRACH back-end processing module, a cross-clock domain register and a configuration port module, wherein PRACH front-end processing module, used to process the time domain IQ data of the received signal receive (t) is processed and the output data is the PRACH effective subcarrier IQ data in the frequency domain And transmit the output data to the front-end and back-end transfer modules, where RAM0 and RAM1 are 2 RAMs, p is the index of the corresponding symbol, and m is the subcarrier index; The front-end and back-end transfer module is used to complete the cross-clock domain conversion of the front-end and back-end data using asynchronous RAM, and the output of the PRACH front-end processing module After asynchronous RAM0, RAM1 and RAM2 write and read, the back-end input is obtained. Wherein, v is the index of the corresponding symbol, and m is the subcarrier index; The PRACH back-end processing module is used to process the input data obtained from the front-end and back-end transfer modules, obtain the preamble detection result preamble_indication, and output it in the form of a 32-bit wide register group; The cross-clock domain register and configuration port module is used for the configuration parameter prach_config of PRACH to be processed by the CPU and then sent to the FPGA via PCIE. The detection result preamble_indication obtained by the correlation peak detection is processed across the clock domain and reported to the CPU via PCIE.
8. The device for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 7, characterized in that: The PRACH front-end processing module includes a frame and time slot determination submodule, a cyclic prefix removal submodule, a data FIFO buffer submodule, and an FFT and effective subcarrier extraction submodule.
9. The device for implementing PRACH preamble sequence detection based on FPGA as claimed in claim 7, characterized in that: The PRACH back-end processing module includes a local sequence generation preparation submodule, a local sequence generation submodule, a correlation and IFFT submodule for received data and local sequence, and a correlation peak detection submodule.
Citation Information
Cited By
Receiver IQ skew measurement method and system based on frequency domain recombination ZC sequence
CN120811492A
