Priority-based polling scheduling method for AFDX network monitoring

By adopting a priority-based polling and scheduling method in AFDX network monitoring, adjusting port priority and dynamic storage space allocation, the problems of untimely data handling and bandwidth waste in traditional methods are solved, and efficient and fair data handling and storage.

CN115766626BActive Publication Date: 2024-12-03CHINESE AERONAUTICAL RADIO ELECTRONICS RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211252562.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-13
Publication Date
2024-12-03
Estimated Expiration
2042-10-13

AI Technical Summary

Technical Problem

In traditional AFDX network monitoring methods, polling scheduling based on time slice division may cause a certain port to fail to transport data in time, resulting in full FPGA cache or wasted bandwidth, and the data transfer bandwidth cannot be effectively utilized.

Method used

Priority-based polling and scheduling method is adopted to adjust port priority in real time, dynamically allocate storage space, use DMA to carry data, and store data frames in large-capacity memory, eliminate fixed time slices, and improve data handling efficiency.

Benefits of technology

It realizes fair data handling at each port, maximizes bandwidth use, reduces CPU load and FPGA resource occupancy, avoids data discarding and bandwidth waste, and improves data handling efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115766626B_ABST
    Figure CN115766626B_ABST
Patent Text Reader

Abstract

The present invention discloses a priority-based polling scheduling method for AFDX network monitoring, which uses DMA to transfer data received from the switch monitoring port to the CPU memory according to priority, and finally stores it in a large-capacity storage device. The present invention takes into account the fairness of polling scheduling and the high efficiency of priority-based scheduling, which can not only meet the requirements of AFDX network monitoring data transfer, but also eliminate the disadvantages of traditional time division multiplexing methods.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of aviation full-duplex switched Ethernet, and relates to a priority-based polling scheduling method applied to AFDX network monitoring, which can orderly save the data entering and leaving the AFDX switch on the aircraft, store it in a large-capacity storage for ground analysis, and provide reliability guarantee for ground analysis of onboard faults. Background Art

[0002] Avionic Full-Duplex Switched Ethernet (AFDX / ARINC664) is a deterministic aircraft data network bus system suitable for avionics system information transmission. The AFDX network is based on standard IEEE802.3 technology and defines a standard for avionics electrical and protocol specifications.

[0003] As the hub of AFDX network data exchange, the AFDX switch provides safe and reliable data exchange between avionics subsystems. In order to enable the ground to review and analyze the onboard data, the data entering and leaving the switch on the aircraft needs to be saved in a large-capacity storage device, and the data can be restored on the ground for viewing and analysis.

[0004] The AFDX switch provides N monitoring ports, and the data of each monitoring port is temporarily stored in the internal cache of the FPGA through the MAC transceiver. When the MAC transceiver receives a frame of AFDX monitoring data, the EP side actively starts DMA to move the frame data to the CPU memory.

[0005] Compared with the traditional RC-side DMA startup, this method can reduce the CPU load. The EP-side DMA startup has no effect on the RC side, and the application can run other processes to improve the efficiency of the application side. At the same time, it can reduce the utilization rate of FPGA resources. There is no need to use large-capacity BRAM to cache data. The EP side can start DMA to transfer after receiving a frame of data. The internal cache of the FPGA only needs to be able to temporarily store a few frames of data.

[0006] The data frames sent from N monitoring ports to the FPGA have the same priority and should be polled and transported. Traditionally, the method based on time slice division is adopted, which divides the time dimension into N time slices. Each time slice enables a port to carry out data. When the DMA state is ready and the time slice is enabled, the monitoring data received by the port can be carried. The disadvantage of this method is that it may cause a certain port to be unable to carry data all the time, so that the internal cache of the FPGA is full and the newly received data can only be discarded. In order to avoid frame loss, a large-capacity BRAM is required, which greatly occupies the internal cache resources of the FPGA; at the same time, if a certain port has no monitoring data, the time slice is always in an invalid state, which will reduce the carrying efficiency and cause bandwidth waste. Summary of the invention

[0007] The invention aims to provide a priority-based polling scheduling method for AFDX network monitoring, which ensures that each port can fairly carry data by adjusting the priority of the port in real time, and removes the fixed time slice occupied by each port to ensure the maximum use of data carrying bandwidth. The present invention takes into account the fairness of polling scheduling and the high efficiency of priority-based scheduling, which can not only meet the requirements of AFDX network monitoring data carrying, but also eliminate the drawbacks of traditional time division multiplexing methods.

[0008] The object of the present invention is achieved through the following technical solutions:

[0009] A priority-based polling scheduling method for AFDX network monitoring includes the following steps:

[0010] Step 1: The master controller configures the MAC receivers of the N monitoring ports of the AFDX switch to enable reception;

[0011] Step 2: The master controller allocates the storage space size and writes the base address of each monitoring port into the FIFO inside the FPGA. Each base address represents a storage space.

[0012] Step 3: The MAC receiver with the receiving enable turned on receives the data frame from the corresponding monitoring port and temporarily stores the received data frame in the BRAM;

[0013] Step 4: Once the MAC receiver receives a frame of data, the monitoring port sends a write request REQ to the DMA;

[0014] Step 5: At the beginning of the time slice, if the DMA is idle, a write response ACK is given to the write request REQ of the monitoring port with the highest priority, and the priority of the monitoring port is reduced to the lowest, and the priority of other monitoring ports is increased by one level;

[0015] Step 6: After the monitoring port receives the response from DMA, it obtains the length of the DMA data to be transferred, and starts DMA to transfer the data in BRAM to the storage space corresponding to the base address according to the length of the transferred data; when the space pointed to by the base address is full, the FPGA sends a full flag and the stored frame length to the master control;

[0016] Step 7: When the master controller detects that a monitoring port is full, it moves the data frames in the storage space to the large-capacity storage.

[0017] Preferably, in step 2, 2*N space base addresses are written into the FPGA internal FIFO, and two base addresses are allocated to each monitoring port;

[0018] In step 6, when the space pointed to by one of the base addresses of the monitoring port is full, the FPGA sends a full flag and the stored frame length to the master control; at the same time, the base address of the second space is obtained as the destination base address of the next DMA transfer;

[0019] In step seven, after the data frames in the storage space are transferred to the large-capacity storage, the main control promptly writes back the base address of the space to release the storage space for FPGA to perform ping-pong storage.

[0020] Preferably, in step three, after receiving the data frame, the MAC receiver only performs length check, CRC check, and data alignment check on the received frame, and counts the byte length of the received frame; only discards overlong frames, overshort frames, and byte-unaligned frames.

[0021] Preferably, in step five, if the DMA finds that the monitoring port with the highest priority has not initiated a write request because it has not received any data, it lowers its priority to the lowest and responds to the monitoring port with the second highest priority.

[0022] The beneficial effects of the present invention are:

[0023] Through the priority-based polling scheduling algorithm, the priority of the port can be adjusted in real time to ensure that each port can transfer data fairly, eliminating the disadvantage of port blocking caused by the failure to transfer data on some ports in time; at the same time, the fixed time slice occupied by each port is removed to ensure the maximum use of data transfer bandwidth and improve the transfer efficiency; the method of starting DMA on the EP end is used to reduce the CPU load and ensure CPU work efficiency, and there is no need to use large-capacity BRAM to cache data, reducing the FPGA resource occupancy rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The figure is a flow chart of a priority-based polling scheduling method applied to AFDX network monitoring as shown in an embodiment. DETAILED DESCRIPTION

[0025] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0026] The priority-based polling scheduling method for AFDX network monitoring shown in this embodiment uses DMA to sequentially transfer data received from the switch monitoring port to the CPU memory, and finally stores it in a large-capacity storage device, taking into account the fairness of polling scheduling and the high efficiency of priority-based scheduling.

[0027] This embodiment is based on an FPGA (field programmable gate array) hardware design platform and uses a hardware description language to implement a priority-based polling scheduling function. Figure 1 As shown, the following steps are included:

[0028] Step 1: Each monitoring port of the AFDX switch has a MAC receiver. The master controller configures the MAC receivers of the N monitoring ports of the AFDX switch to enable reception. The MAC receiver can be turned on or off at any time.

[0029] Step 2: The master controller allocates the storage space size and writes 2*N space base addresses into the FPGA internal FIFO. Each base address represents a piece of storage space. Two base addresses are allocated to each monitoring port, which can perform ping-pong storage.

[0030] Step 3: The MAC receiver with reception enabled receives the data frame from the corresponding monitoring port; the logic only performs length check, CRC check, data alignment check, etc. on the received frame, and counts the byte length of the received frame; only overlong frames, undershort frames, and byte-unaligned frames are discarded, and other frames are received even if they are erroneous; N BRAM memories are instantiated to temporarily store the received data frames in the BRAM.

[0031] Step 4: Once the MAC receiver completes receiving a frame of data, the monitoring port initiates a write request REQ to the DMA, waits for the DMA to be in an idle state, and gives a write response ACK.

[0032] Step 5: At the beginning of the time slice, if the DMA is idle, a write response ACK is given to the write request REQ of the monitoring port with the highest priority, and the priority of the monitoring port is reduced to the lowest, and the priorities of other monitoring ports are increased by one level.

[0033] N monitoring ports P0~P N-1 , can initiate write requests REQ0~REQ N-1, but each time DMA is idle, it can and can only respond to one write request; in the initial state, P0 has the highest priority. If P0 and multiple other ports give write requests at the same time, DMA will only respond to P0 and start DMA transfer of P0 port data frame; once the response is successful, P0's priority drops to the lowest and P1's priority rises to the highest. The next time DMA is idle, it will give priority to responding to P1's request; if DMA finds that the monitoring port with the highest priority has not initiated a write request because it has not received data, it will lower its priority to the lowest and respond to the monitoring port with the second priority; ensure that DMA can always respond to a certain port when it is idle, and maximize the use of DMA performance, unless all monitoring ports do not receive data.

[0034] Step 6: After the monitoring port receives the response from DMA, it obtains the length of the DMA data to be transferred, and starts DMA to transfer the data in BRAM to the storage space corresponding to the base address according to the length of the transferred data. When the space pointed to by one of the base addresses of the monitoring port is full, the FPGA sends a full flag and the stored frame length to the master control; at the same time, it obtains the base address of the second space as the destination base address for the next DMA transfer.

[0035] Step 7: When the master detects a full flag of a monitoring port, it moves the data frame to a large-capacity memory and writes back the base address of the space in time to release the storage space so that the FPGA can perform ping-pong storage.

[0036] It is understandable that those skilled in the art can make equivalent substitutions or changes based on the technical solution and inventive concept of the present invention, and all these changes or substitutions should fall within the protection scope of the claims attached to the present invention.

Claims

1. A priority-based polling scheduling method for AFDX network monitoring, characterized in that The steps include: Step 1: The master controller configures the MAC receivers of the N monitoring ports of the AFDX switch to enable reception; Step 2: The master controller allocates the storage space size and writes the base address of each monitoring port into the FIFO inside the FPGA. Each base address represents a storage space. Step 3: The MAC receiver with the receiving enable turned on receives the data frame from the corresponding monitoring port and temporarily stores the received data frame in the BRAM; Step 4: Once the MAC receiver receives a frame of data, the monitoring port sends a write request REQ to the DMA; Step 5: At the beginning of the time slice, if the DMA is idle, a write response ACK is given to the write request REQ of the monitoring port with the highest priority, and the priority of the monitoring port is reduced to the lowest, and the priority of other monitoring ports is increased by one level; Step 6: After the monitoring port receives the response from DMA, it obtains the length of the DMA data to be transferred, and starts DMA to transfer the data in BRAM to the storage space corresponding to the base address according to the length of the transferred data; when the space pointed to by the base address is full, the FPGA sends a full flag and the stored frame length to the master control; Step 7: When the master controller detects that a monitoring port is full, it moves the data frames in the storage space to the large-capacity storage.

2. A priority-based polling scheduling method for AFDX network monitoring according to claim 1, characterized in that: In step 2, 2*N base addresses are written into the FPGA internal FIFO, and two base addresses are allocated to each monitoring port; In step 6, when the space pointed to by one of the base addresses of the monitoring port is full, the FPGA sends a full flag and the stored frame length to the master control; at the same time, the base address of the second space is obtained as the destination base address of the next DMA transfer; In step seven, after the data frames in the storage space are transferred to the large-capacity storage, the main control promptly writes back the base address of the space to release the storage space for FPGA to perform ping-pong storage.

3. The priority-based polling scheduling method for AFDX network monitoring according to claim 1 is characterized in that: In step three, after receiving the data frame, the MAC receiver only performs length check, CRC check, and data alignment check on the received frame, and counts the byte length of the received frame; only discards overlong frames, undershort frames, and byte-unaligned frames.

4. The priority-based polling scheduling method for AFDX network monitoring according to claim 1 is characterized in that: In step 5, if DMA finds that the monitoring port with the highest priority does not initiate a write request because it has not received any data, it will reduce its priority to the lowest and respond to the monitoring port with the second priority.