Multi-process and multi-thread universal data synchronization method of radar on domestic platform

Abstract

The invention belongs to the technical field of signal processing multi-process and multi-thread collaborative data calculation of a radar on a domestic platform, and particularly relates to a multi-process and multi-thread general data synchronization method of the radar on the domestic platform, which is suitable for practical engineering application. Compared with the prior art, the method is more matched with the application environment of domestic multi-core and multi-board cards. In order to adapt to radar localization big data processing, the resource utilization rate is increased for signal processing calculation in a multi-process / thread mode, and multi-core and multi-board cards are flexibly connected in series by applying the alignment method. Through the stability of multi-core multi-board card data alignment transmission, the computing power of radar signal processing software is improved, and a basis capable of being stably implemented is provided for extension of a computing method for radar signal processing.

Description

Technical Field

[0001] The present invention belongs to the field of data calculation technology of multi-process and multi-thread collaboration of radar signal processing on a domestic platform, and in particular relates to a general data synchronization method of multi-process and multi-thread of radar on a domestic platform, which is suitable for practical engineering applications. Background Art

[0002] For a long time, radar signal processing on imported boards has adopted a serial data processing method for large data volume calculations, that is, the data must complete one calculation before the next calculation can be taken. For example, the next step of anti-interference calculation can only be carried out after pulse compression is completed. This requires that all data of the same calculation method (such as pulse compression algorithm) must be processed sequentially within the process of a board (board 1). This serial method can be implemented on imported boards.

[0003] However, due to the performance limitations of domestic boards, synchronization problems such as the following occurred during the serial calculation of signal processing.

[0004] When the amount of data increases to a certain extent, it will exceed the computing power of the domestic platform board 1 hardware platform. In order not to affect the signal processing function and performance of the radar, the method of separating the data to board 2 for calculation is adopted. Since all pulse compression data are required for the next step of anti-interference calculation, the data of board 1 and board 2 need to be transferred to board 3 for data alignment in preparation for anti-interference calculation. This creates the need to synchronize and align data from different sources. If the data is not synchronized and aligned, the signal processing will not be able to complete its own function. Therefore, a stable method of synchronizing and aligning data is particularly important.

[0005] Therefore, it is necessary to develop a new general data synchronization method based on multi-process and multi-thread of domestic platform to solve the above problems existing in the general data synchronization method of multi-process and multi-thread of domestic platform for existing radar. Summary of the invention

[0006] The purpose of the present invention is to provide a general data synchronization method for radar multi-process and multi-thread on a domestic platform, so as to solve the problem that when the amount of data processed by radar signal exceeds the capability of the hardware platform and needs to be subpackaged for calculation data, or when there are multiple input data in the thread, the data from multiple sources are synchronized and aligned before the next step of calculation is performed, so as to achieve the purpose of stably improving the upper limit and flexibility of the working capability of radar signal processing software.

[0007] The present invention achieves the above-mentioned purpose through the following technical solutions: A general data synchronization method for radar multi-process and multi-thread on a domestic platform, characterized in that the synchronization method comprises the following steps: Step 1: Condition constraints.

[0008] In the signal processing software program, the source process / thread is used as the network sender, and the destination process / thread is used as the network receiver. Two ring memories a and b of size k are opened in the destination process / thread. The message in the packet header received by the network must contain a packet number that is incremented by 1, and the maximum number is greater than k.

[0009] Step 2, receive data.

[0010] The destination process / thread receives the data packets 8 from the source process / thread via the network at radar cycles (i.e., fixed time intervals).

[0011] Step 3: Data alignment.

[0012] The data packet number 8 in the header message of data packet 8 is stored in the first position of the ring memory a, and the number of data packets is accumulated in the first position of the ring memory b of the same size. It is determined whether the number of data packets with the same header number 8 is equal to the number of data packets that need to be aligned. The judgment principle is detailed in step 5.

[0013] Step 4: Align the protection measures.

[0014] Due to the reuse characteristics of the ring memory, it is necessary to add a count clearing protection measure before accumulating the number of packets numbered 8. The protection measures are detailed as follows: The first cycle of ring memory a stores numbers 0 to k-1 (k=8). When the number of the packet received by the network is 8, which is not equal to the number 0 of the packet stored in the first position of ring memory a, the first position of ring memory b is cleared before accumulation. This ensures that when multiple packets with the same number of 8 are aligned, they all start counting from 0.

[0015] Step 5: Alignment conditions When the count value of the first position of the ring memory b is equal to the number of data packets with the requirement number 8, the data has been aligned, the first position of the ring memory b can be cleared, and the next operation is performed. Otherwise, repeat steps 2 to 5.

[0016] The beneficial effects of the present invention compared with the prior art are: Compared with the prior art, the present invention is more suitable for the application environment of domestic multi-core and multi-board cards. In order to adapt to the big data processing of domestic radar, the signal processing calculation is increased in resource utilization through multi-process / thread mode, and the alignment method is applied to flexibly connect multi-core and multi-board cards in series. Through the stability of multi-core and multi-board data alignment transmission, the computing power of radar signal processing software is improved, providing a stable implementation basis for the expansion of radar signal processing calculation methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1This is a flow chart of a general data synchronization method for multi-process and multi-thread of a radar on a domestic platform of the present invention, and the figure takes k=8 as an example; DETAILED DESCRIPTION

[0018] The present invention is based on a domestically produced CPU board and is completed under the QTCreator software in a Linux environment. The illustrated method is a universal synchronization method, which is applicable to a variety of signal processing software, including the QTCreator software.

[0019] The radar has a general data synchronization method for multi-process and multi-thread on a domestic platform, characterized in that the synchronization method comprises the following steps: Step 1: Condition constraints.

[0020] In the signal processing software program, the source process / thread is used as the network sender, and the destination process / thread is used as the network receiver. Two ring memories a and b of size k are opened in the destination process / thread. The message in the packet header received by the network must contain a packet number that is incremented by 1, and the maximum number is greater than k.

[0021] k is any integer value greater than 1 and less than or equal to 64.

[0022] Step 2, receive data.

[0023] The destination process / thread receives the data packets 8 from the source process / thread via the network at radar cycles (i.e., fixed time intervals).

[0024] Step 3: Data alignment.

[0025] The data packet number 8 in the header message of data packet 8 is stored in the first position of the ring memory a, and the number of data packets is accumulated in the first position of the ring memory b of the same size. It is determined whether the number of data packets with the same header number 8 is equal to the number of data packets that need to be aligned. The judgment principle is detailed in step 5.

[0026] Step 4: Align the protection measures.

[0027] Due to the reuse characteristics of the ring memory, it is necessary to add a count clearing protection measure before accumulating the number of packets numbered 8. The protection measures are detailed as follows: The first cycle of ring memory a stores numbers 0 to k-1 (k=8). When the number of the packet received by the network is 8, which is not equal to the number 0 of the packet stored in the first position of ring memory a, the first position of ring memory b is cleared before accumulation. This ensures that when multiple packets with the same number of 8 are aligned, they all start counting from 0.

[0028] Step 5: Alignment conditions When the count value of the first position of the ring memory b is equal to the number of data packets with the requirement number 8, the data has been aligned, the first position of the ring memory b can be cleared, and the next operation is performed. Otherwise, repeat steps 2 to 5.

[0029] This general data synchronization method based on multi-process and multi-thread of domestic platform can synchronize and align data after data segmentation; it can align and synchronize data when data is input from multiple sources; it improves the computing power of radar signal processing software and is one of the foundations for the implementation of signal processing software on domestic platform.

[0030] Compared with the prior art, the present invention is more suitable for the application environment of domestic multi-core and multi-board cards. In order to adapt to the big data processing of domestic radar, the signal processing calculation is increased in resource utilization through multi-process / thread mode, and the alignment method is applied to flexibly connect multi-core and multi-board cards in series. Through the stability of multi-core and multi-board data alignment transmission, the computing power of radar signal processing software is improved, providing a stable implementation basis for the expansion of radar signal processing calculation methods.

Claims

1. A general data synchronization method for radar multi-process and multi-thread on a domestic platform, characterized by: The synchronization method comprises the following steps: Step 1, conditional constraints; In the signal processing software program, the source process / thread is used as the network sending end, and the destination process / thread is used as the network receiving end; two ring memories a and b of size k are opened in the destination process / thread, and the message in the packet header received by the network must contain a packet number that is a value that increases by 1, and the maximum number is greater than k; Step 2, receiving data; The destination process / thread receives the data packet 8 through the network at a radar cycle (i.e., a fixed time interval); Step 3, data alignment; Store the data packet number 8 in the header message of data packet 8 into the first position of the ring memory a, and accumulate the number of data packets in the first position of the ring memory b of the same size; Determine whether the number of data packets with the same header number of 8 is equal to the number of data packets that need to be aligned. For details on the determination principle, see step 5. Step 4, aligning protection measures; Due to the reuse characteristics of the ring memory, it is necessary to add a count clearing protection measure before accumulating the number of packets numbered 8; Step 5: Alignment conditions When the count value of the first position of the ring memory b is equal to the number of data packets with the requirement number 8, the data has been aligned, the first position of the ring memory b can be cleared, and the next step is performed; otherwise, steps 2 to 5 are repeated.

2. According to claim 1, a general data synchronization method for radar multi-process and multi-thread on a domestic platform is characterized in that: The k is any integer value greater than 1 and less than or equal to 64.

3. According to claim 1, a general data synchronization method for radar multi-process and multi-thread on a domestic platform is characterized in that: The counting clearing protection measure is as follows: the ring memory a stores numbers 0 to k-1 in the first cycle. When the data packet numbered 8 received by the network is not equal to the data packet numbered 0 stored in the first position of the ring memory a, the first position of the ring memory b is cleared before accumulation; to ensure that when multiple packets with the same number of 8 are aligned, the counting starts from 0.

Images