WO2016107210A1

WO2016107210A1 - Redundant industrial ethernet system with multistage packet filtering and service classification control

Info

Publication number: WO2016107210A1
Application number: PCT/CN2015/089317
Authority: WO
Inventors: 周华良; 郑玉平; 杨志宏; 谢黎; 姚吉文; 赵马泉; 王凯; 姜雷
Original assignee: 国电南瑞科技股份有限公司
Priority date: 2014-12-30
Filing date: 2015-09-10
Publication date: 2016-07-07
Also published as: CN104539408A

Abstract

Disclosed is a redundant industrial Ethernet system with multistage packet filtering and service classification control, which is characterized by realizing the comprising of two paths of MAC packet processing units, a dual-network redundancy unit, a packet filtering processing unit and a packet classification forwarding unit using an FPGA, wherein the two paths of MAC packet processing units are used for receiving and sending packets and transmitting the received packets to the subsequent-stage dual-network redundancy unit to process same; the dual-network redundancy unit receives a previously received packet and discards a subsequently received packet, and then transmits same to the packet filtering processing unit; the packet filtering processing unit conducts multistage filtering on a received packet and then transmits same to the packet classification forwarding unit; and the packet classification forwarding unit selects different data transmission channels according to packet types, and transmits same to different CPU systems to process different data packets. Provided is a redundant industrial Ethernet system which greatly reduces the workload of a CPU system when processing irrelevant packets and guarantees the stability and reliability of the operation of the overall system.

Description

Redundant industrial Ethernet system with message multi-level filtering and service classification control

Technical field

The invention belongs to the field of industrial Ethernet communication of a power system automation intelligent substation system, in particular, a multi-level filtering and service classification control method based on FPGA technology for improving the reliability of the secondary equipment Ethernet communication of the intelligent substation automation system. Redundant industrial Ethernet system.

Background technique

At present, the structure of intelligent substation network communication mainly has the following four types: (1) the combination of fiber point-to-point and GOOSE network, in which the protection device in the state-owned intelligent substation is “straight-collective”, that is, point-to-point sampling and point-to-point Tripped. At the same time, there is also a protection framework for “direct mining network jump”, which is concentrated in the digital substation of the South Network; (2) a combination of fiber point-to-point, sampled value network and GOOSE network, and the protection device is a fiber point-to-point mode, and the measurement and control , metering, fault filtering is to obtain relevant information from the sampled value network; (3) using the process bus method, that is, adopting the method of AC sampling (SV) and GOOSE networking, which is divided into the common network or the network mode; (4) ) Adopt the full process bus mode, namely SV-9-2, IEEE 1588 and GOOSE unified networking.

It can be seen that there are two modes of peer-to-peer communication and networking communication in the application of intelligent substation network communication. Whether it is peer-to-peer communication or networking communication, the network packets received by the devices that form the network node are not necessarily required by themselves. In the case of large traffic packets or even network storms on the network, in order to ensure network nodes. The devices on the device operate reliably and perform functions such as protection measurement and control in real time. The current main method is to rely on the internals of the Ethernet switch. The broadcast storm is suppressed and the VLANs are divided on the Ethernet switch to reduce the impact of large traffic packets on the network on the network node protection measurement and control devices. The above method does not guarantee that the protection and monitoring devices acting as network nodes can be stably operated and the functions of protection, measurement and control are completed in real time and reliably in the presence of large traffic or network storms on the network. The technology is to filter all received Ethernet packets on the receiving device as a network node, remove the unnecessary packets of the receiving device, and control the effective packet traffic processed by the CPU, thereby ensuring the network. Each protection and control device acting as a network node in the presence of large traffic or network storms can also stably operate and perform functions such as protection, measurement and control in real time and reliably.

As a device of a network communication node in a smart substation, a general embedded CPU of a RISC architecture is generally used to process network messages. Since a single CPU is limited by the RISC architecture, its processing power is rather limited, and generally adopts a dual CPU system or a multi-core CPU system. The technology classifies and controls different network message types at the receiving end of the network node, and different CPUs or different cores of the CPU constitute a CPU system to process different types of Ethernet packets, thereby improving system-to-Ethernet packets. Processing efficiency.

The communication network in the substation directly participates in the functions of protection and measurement and control, and its reliability will directly affect the reliability of the substation automation system. In order to improve the reliability of the communication network, the commonly used method is the network redundancy design. This technology combined with substation network application uses PRP parallel redundant network technology based on IEC62439 to improve the reliability of network communication.

Summary of the invention

In order to solve the problems in the prior art, the present invention provides a message multi-level filtering and service classification that greatly reduces the processing workload of the CPU system on irrelevant messages and ensures the stability and reliability of the entire system operation. Controlled redundant industrial Ethernet system.

In order to solve the above problems, the technical solution adopted by the present invention is:

A redundant industrial Ethernet system with message multi-level filtering and service classification control, including: implementing two-way MAC packet processing unit by using FPGA, dual-network redundant unit, message filtering processing unit, and message classification Forwarding unit

The two-way MAC packet processing unit is configured to receive and send a message, and after receiving the message, transmit the message to the dual-network redundant unit of the subsequent stage to process the message;

The dual-network redundancy unit receives the first-mentioned packet, discards the received packet, and then delivers the packet to the packet filtering processing unit.

The packet filtering processing unit performs multi-level filtering on the received packet, and then transmits the packet to the packet classification forwarding unit.

The message classification and forwarding unit selects different data transmission channels according to the message type, and transmits the data to different CPU systems to process different data messages.

The foregoing redundant industrial Ethernet system with packet multi-level filtering and service classification control is characterized in that: the two-way MAC packet processing unit is configured to receive and send a packet containing a PRP identifier, and two-way packets. After receiving, the received message is classified, and the corresponding descriptor is filled in for the subsequent message classification and forwarding unit; and the time stamp is added to the end of the message for use by the CPU processing unit.

The foregoing redundant industrial Ethernet system with packet multi-level filtering and service classification control is characterized in that: the dual network redundant unit receives the packet processed by the MAC packet processing unit, and the dual network redundant unit combines The PRP identification information of the packet is received by the discarding algorithm, and the received packet is discarded. The received packet is removed from the PRP identifier and then transmitted to the packet filtering processing unit.

The foregoing redundant industrial Ethernet system with message multi-stage filtering and service classification control is characterized in that: the multi-stage filtering of the message filtering processing unit includes the purpose in the Ethernet packet Filtering of the MAC address, including hash filtering and unicast address filtering; filtering for the same CRC Ethernet packet; filtering the multi-link filtering for the feature quantity in the power system application message.

The foregoing redundant industrial Ethernet system with packet multi-level filtering and service classification control is characterized in that, when the packet is a 1588 packet, the MAC packet processing unit adds a 1588 time stamp to the end of the packet.

The foregoing redundant industrial Ethernet system with message multi-level filtering and service classification control is characterized in that: the message classification and forwarding unit sets different types of SV, GOOSE according to descriptors provided by the MAC message processing unit. The 1588 and MMS messages are transmitted to the corresponding CPU system.

The foregoing redundant industrial Ethernet system with packet multi-level filtering and service classification control is characterized in that: the multi-level filtering of the message filtering processing unit includes:

1) Perform HASH filtering on the received message multicast address: find the destination address from the packet and perform HASH calculation on it, compare the calculation result with the set allowable receiving HASH table, and accept the message according to the HASH table. Otherwise, throw away;

2) Perform unicast address filtering: the unicast address table is redistributed, and only the destination address packets that match the unicast address table are allowed to be received.

3) Storage and filtering for the same CRC Ethernet packet: If there is a packet with the same CRC check result, it can be considered as repeated packet reception, and actively remove the packet with the same CRC check result;

4) Filtering the feature quantity in the power system application message: Some messages have high similarity with the actual required message, and have the same destination address and the same type information, and are rejected by the ID layer in the application layer. These messages, which are only slightly different from the required messages, use these nuances as IDs to discard messages containing these ID information;

5) Bandwidth limit filtering for each type of packet: Set the bandwidth allowed for each packet in advance. When a large number of valid packets are flooded, the number of received packets is counted in the specified time slice. Cannot continue to receive until the next time slice is reopened.

The beneficial effects achieved by the invention:

1. Based on the general MAC function, the MAC Ethernet communication module classifies the received messages and fills in the corresponding descriptor registers for use by the FPGA class processing message module of the latter stage; Time stamp for the CPU to process messages. This improved MAC module greatly increases the efficiency of the CPU in processing Ethernet data.

2. The dual-network redundant unit (PRP) complies with the IEC62439 specification and can communicate well with other devices that comply with this specification, which is beneficial to improve the compatibility of the device. The dual-network redundant unit receives the first-come-first packet and discards it. The packet, the dual-network redundant communication mode, can realize zero-delay of network fault recovery, no frame loss when fault occurs, and improve network reliability; dual-network redundant unit provides dual-network interface, internal equivalent one-way The network port does not change the network interface of the upper layer software when implementing dual network redundant communication, and can effectively alleviate the burden of processing data by the CPU system.

3. The packet filtering processing unit Packet Filter is used for multi-level filtering of network packets, and the filtering policy is completed by the CPU through the configuration of the relevant registers through the bus. The filtering control module performs multi-level filtering from multiple aspects. By multi-level filtering of network packets, the CPU can greatly reduce the processing workload of irrelevant packets and ensure the stability and reliability of the entire system. .

4. The packet classification forwarding unit Packet Classification&transmission selects different data transmission channels according to the packet type, uses different processors to process different data packets, reduces the burden on the CPU system, reduces the usage rate of the same data transmission channel, and effectively improves Packet processing and response speed, improve system real-time, stability, the system can flexibly change and configure the channel type of the channel transmission and processor processing.

5. Support accurate network timing based on IEC61588.

DRAWINGS

1 is a block diagram showing the overall principle of a redundant industrial Ethernet technology implemented by an FPGA according to the present invention.

2 is a flow chart of message delivery and multi-stage filtration of the present invention.

detailed description

The invention is further described below in conjunction with the drawings.

As shown in FIG. 1 , a redundant industrial Ethernet system with message multi-level filtering and service classification control includes: implementing two MAC packet processing units, dual network redundant units, and messages in the FPGA. a filtering processing unit and a message classification forwarding unit;

The two-way MAC packet processing unit is configured to receive and send a packet with a PRP identifier.

(1) For receiving message transmission: the CPU system writes the message to be sent to the dual-network redundant unit through the corresponding MAC message processing unit, and the dual-network redundant unit copies the message in two identical A PRP identifier is added to the end of the packet. The identifier contains packet length information, A/B network information, packet sequence number, and PRP identification code. The target device uses this information to determine whether to receive the packet or discard the packet. . The two MAC packet processing units simultaneously send the packets containing the PRP identifier.

(2) Message reception: Two MACs in the FPGA receive the message, and classify the received message, and fill in the corresponding descriptor for the message classification and forwarding unit of the later stage; add at the end of the message The time stamp is used by the CPU system to process the message; and when the message is a 1588 message, the 1588 time stamp is added to the end of the message.

The dual-network redundant unit (PRP) receives the packet processed by the MAC packet processing unit, and the dual-network redundant unit combines the PRP identification information of the packet, and uses the discarding algorithm to receive the first-mentioned packet, and discards the packet. After the packet is removed, the received packet is removed from the PRP identifier and then transmitted to the packet filtering. Processing unit.

The multi-stage filtering of the packet filtering processing unit includes filtering the destination MAC address in the Ethernet packet, including hash filtering and unicast address filtering; filtering for the same CRC Ethernet packet; for power system application The feature quantity in the message is filtered and filtered.

The message classification and forwarding unit transmits different types of SV, GOOSE, 1588 and MMS messages to the corresponding CPU system according to the descriptors provided by the MAC message processing unit, and the data flow is more clear through multi-way message forwarding control. It can improve the reliability and scalability of the system. This improved MAC module greatly increases the efficiency of the CPU in processing Ethernet data.

The invention realizes the PRP protocol by utilizing the characteristics of parallel processing of the FPGA, so that the network fault recovers zero delay, no frame loss in case of failure, and has high network reliability and good economy. Compared with the PRP design introduced by other companies, this project has the technical advantage of short switching time.

The multi-level filtering of the message filtering processing unit includes:

1) Perform HASH filtering on the received message multicast address: find the destination address from the packet and perform HASH calculation on it, compare the calculation result with the set allowable receiving HASH table, and conform to the HASH table to receive the message, otherwise throw away;

2) Perform unicast address filtering: Because there are many cases in the HASH calculation result, the HASH calculation results may be identical for different destination addresses. Therefore, HASH filtering cannot completely remove invalid destination address packets. In the broadcast address table, only the destination address packets that match the unicast address table are allowed to be received.

4) Filtering the feature quantity in the power system application message: having the same destination address, For the same type information, these messages are rejected at the application layer using ID recognition techniques. These messages may be only slightly different from the required messages. These nuances are used as IDs to discard messages containing these ID information. The identification of the ID may be a specified character string, or may be a calculation result of a character string of a specified length from a specified offset position;

5) Filtering the bandwidth of each type of packets: Filtering the packets with too much bandwidth to ensure that other packets with higher priority are passed. Set the allowed bandwidth of each packet in advance and receive the packets within the specified time. The number of packets is counted. When the specified number is reached, the message cannot be received until the next time slice is reopened.

The invention utilizes the FPGA in the receiving device of the network node to filter unnecessary packets on the Ethernet network through the multi-stage filtering, thereby greatly reducing the processing workload of the CPU on the receiving end for the extraneous message, and ensuring that the switch does not suppress even if the switch does not inhibit the switch. In the case of network storms, the receiving devices on the network can still perform functions reliably and in real time, ensuring the stability and reliability of the entire system.

In summary, the present invention utilizes the MAC controller independently designed by the FPGA technology to receive Ethernet packets, and filters the unnecessary packets on the Ethernet network through multi-stage filtering according to the conditions set by the receiving CPU. Through the multi-level filtering function of the network packet, the processing workload of the device CPU on the irrelevant packet is greatly reduced, and the stability and reliability of the entire system operation are ensured.

The invention utilizes the characteristics of FPGA parallel processing, independently develops and implements the PRP protocol, enables the network fault to recover zero delay, does not drop frames when the fault occurs, has high network reliability and good economy.

According to the feature descriptor of the network message, the invention uses the classification control technology to classify and control different message types, and transmits the SV, GOOSE, 1588 and MMS messages to the corresponding CPU processing system, and controls the data through the multi-way message forwarding control. The flow is more clear and can improve the reliability and scalability of the system.

The basic principles, main features and advantages of the present invention have been shown and described above. It should be understood by those skilled in the art that the present invention is not limited by the foregoing embodiments, and that the present invention is only described in the foregoing description and the description of the present invention, without departing from the spirit and scope of the invention. Various changes and modifications are intended to be included within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and their equivalents.

Claims

A redundant industrial Ethernet system with packet multi-level filtering and service classification control, comprising: two MAC packet processing units, a dual network redundancy unit, a packet filtering processing unit, and a message classification and forwarding unit;

The two-way MAC packet processing unit is configured to receive and send a message, and after receiving the message, transmit the message to the dual-network redundant unit of the subsequent stage to process the message;

The dual-network redundancy unit receives the first-mentioned packet, discards the received packet, and then delivers the packet to the packet filtering processing unit.

The packet filtering processing unit performs multi-level filtering on the received packet, and then transmits the packet to the packet classification forwarding unit.

The message classification and forwarding unit selects different data transmission channels according to the message type, and transmits the data to different CPU systems to process different data messages.
The redundant industrial Ethernet system with packet multi-level filtering and service classification control according to claim 1, wherein the two-way MAC packet processing unit is configured to receive and send a packet containing a PRP identifier. After receiving the two-way message, the received message is classified, and the corresponding descriptor is filled in for the subsequent message classification and forwarding unit; and the time stamp is added to the end of the message for use by the CPU processing unit.
The redundant industrial Ethernet system with message multi-level filtering and service classification control according to claim 2, wherein the dual network redundant unit receives the message processed by the MAC packet processing unit, The network redundancy unit combines the PRP identification information of the packet with the discarding algorithm, receives the first-mentioned packet, discards the received packet, and then removes the received packet from the PRP identifier, and then transmits the packet to the packet filtering processing unit.
Redundant industry with message multi-level filtering and traffic classification control according to claim 3 An Ethernet system, characterized in that: the multi-stage filtering of the packet filtering processing unit includes filtering a destination MAC address in an Ethernet packet, including hash filtering and unicast address filtering; and targeting the same CRC Ethernet packet. The text is filtered; from the application layer, filtering is performed on the feature quantity in the power system application message.
The redundant industrial Ethernet system with packet multi-level filtering and service classification control according to claim 4, wherein when the packet is a 1588 packet, the MAC packet processing unit is at the end of the packet. Added 1588 hour markers.
The redundant industrial Ethernet system with message multi-level filtering and service classification control according to claim 5, wherein the message classification and forwarding unit sets different types according to descriptors provided by the MAC message processing unit. The SV, GOOSE, 1588 and MMS messages are transmitted to the corresponding CPU system.
The redundant industrial Ethernet system with packet multi-level filtering and service classification control according to claim 4, wherein the multi-stage filtering of the message filtering processing unit comprises:

1) Perform HASH filtering on the received message multicast address: find the destination address from the packet and perform HASH calculation on it, compare the calculation result with the set allowable receiving HASH table, and conform to the HASH table to receive the message, otherwise throw away;

2) Perform unicast address filtering: the unicast address table is redistributed, and only the destination address packets that match the unicast address table are allowed to be received.

3) Storage and filtering for the same CRC Ethernet packet: If there is a packet with the same CRC check result, it can be considered as repeated packet reception, and actively remove the packet with the same CRC check result;

4) Filtering the feature quantity in the power system application message: Some messages have high similarity with the actual required message, and have the same destination address and the same type information at the application layer. The ID identification technology is used to eliminate these messages. These messages are only slightly different from the required messages. These nuances are used as IDs to discard messages containing these ID information.

5) Bandwidth limit filtering for each type of packet: Set the bandwidth allowed for each packet in advance. When a large number of valid packets are flooded, the number of received packets is counted in the specified time slice. Cannot continue to receive until the next time slice is reopened.