What Is a Front-End Processor?

A front-end processor (FEP), also commonly referred to as a communication controller, has the main function of releasing the host to run applications. In this way, the host will not be constantly disturbed by external devices, making it more efficient to process applications.

A front-end processor (FEP), also commonly referred to as a communication controller, has the main function of releasing the host to run applications. In this way, the host will not be constantly disturbed by external devices, making it more efficient to process applications. It can be a complex front-end mainframe computer interface or simple devices such as multiplexers, bridges, and routers. These devices convert the computer's parallel data into serial data transmitted on the communication line, and perform all necessary control functions, error detection and synchronization. Modern devices also perform data compression, routing, security functions, and collect management information. Front-end processors are generally small or

Research At present, multiple integrated monitoring systems have been launched in cities across the country. From the perspective of the operation of FEP, faults that easily cause communication interruption levels are mainly concentrated on hardware, such as loose link connectors and damaged interface modules. In addition, the instantaneous large amount of data interaction also has a certain impact on FEP performance, which can easily cause FEP's high CPU and memory utilization or temporary data blockage, etc., resulting in system malfunction. In response to these problems, the FEP design scheme should be optimized to ensure that subsequent lines will not repeat the same mistakes ^[1] .

Research on hardware reliability analysis

At present, the FEP interface module has only two types: Ethernet port and serial port, and the transmission medium is RJ45 twisted pair and serial port line; no matter what kind of port, the number of docking points is directly proportional to the level of failure. For devices that use network ports and FEP to connect, increasingly mature network distribution frames have modularized the interconnection between the two, and the reliability of the interconnection has reached a very high level. In contrast, the serial port is based on its characteristics and requires manual wiring at the connector. The stability is naturally lower than that of the network port; but the cost of the serial port is lower, and it is still widely used in systems with low data interaction frequency. application. However, due to the level difference during the operation of the serial port module, when the number of data interaction instants is high, the heat will be very serious, causing the entire FEP temperature to rise, affecting the system performance and the service life of the device.

In short, the network port and serial port modules have their own advantages. As for the subway project, because the functions of the subsystem are simple and complicated, the Ethernet interface is the best choice for systems with frequent data volume interactions; but for systems with moderate or little system data volume interactions From the perspective of reducing costs, you can choose to use a serial port module. Therefore, in the initial stage of interface design, the type of interface module can be determined by estimating the frequency of data interaction, the amount of data, and the daily maintenance requirements of the user on the link.

FEP hardware can support multiple communication interface modules at the same time, that is, support many network ports and serial ports to run at the same time. Therefore, in order to prevent communication failure caused by address confusion, the corresponding address should be planned for each port when the operating system is installed to achieve network isolation. In addition, it is necessary to ensure that each functional module has the functions of self-diagnosis and self-recovery, and to ensure that the communication can recover by itself after the link is restored.

Software reliability analysis

Research In the early stage of the establishment of the connection between the FEP and the subsystem, FEP will detect the first communication module according to the agreement between the two parties. After the two parties normally communicate, follow this step to detect the remaining communication modules one by one until the initialization process is completed. Each module of the FEP only starts polling to collect subsystem device information and caches it in a local register to implement the function of obtaining device information through protocol conversion.

Because the maintenance of the communication connection between the two parties is achieved by continuously sending and receiving messages, and the message is encapsulated by the program according to the protocol format, the protocol itself is a key to the stability of the message. According to the protocol developed according to the system function, the stability of the interactive message is much lower than that of the standard protocol. Although the message is allowed to be erroneous during data interaction, and the receiver can discard the message based on the error check, it wastes bandwidth resources. Therefore, from the perspective of generality and ease of maintenance, the agreement of the agreement should try to choose the type of standard to reduce unnecessary errors caused by FEP custom development. At the same time, when the link between the FEP and the device encounters a problem of communication interruption, different systems should have different processing methods; for systems that require high real-time information, FEP must meet the conditions of connection interruption or blockage. With the function of discarding messages, such as broadcast system (public address (PA), passenger information system (PIS), access control system (ACS), etc .; systems with higher requirements for information integrity, The FEP must cache these instructions locally and continue to interact after the link is restored, such as the sequence of event (SOE) of a power supervisory control and data acquisition system (PSCADA).

Data communication performance analysis and research

When the amount of instant data interaction between the FEP and the subsystem is large, the performance of the FEP will definitely be affected to a certain extent, ranging from short-term data blockage to serious system failure. Take the train monitoring system (automatictrain supervision, ATS) connected to the integrated monitoring system as an example. There are about 2000 information points on the entire line. Based on 2 times / min data exchange calculations, up to 5 million information points can be exchanged throughout the day. This amount of data is for any server. It's all a challenge. Therefore, for such a large data volume interface device, from the perspective of data transmission, the efficiency of using block transmission is definitely higher than the efficiency of using point transmission, but the disadvantage is that the readability of the data will decrease accordingly. In this regard, a compromise method can be adopted. FEP uses block transmission to transmit data, and data analysis is achieved through an external maintenance workstation. Debuggers can observe the data interaction between the two parties on the maintenance workstation, so the load on the FEP itself will be small. Many, data transmission efficiency will naturally increase with it.

From the perspective of data processing, for the data that does not need to be processed by the server, it is best to convert and forward it by the FEP direct protocol, so as to reduce the intermediate link and improve the working efficiency of the entire integrated monitoring system. In the picture

(a) It is a scheme for ATS information forwarding PIS of a subway integrated monitoring system: After receiving the data uploaded by ATS, FEP transmits the data to the server for processing, and then returns the data to FEP, and the FEP transmits to the PIS system. The whole process A total of 4 data transmissions and receptions are required. (B) in the figure is the optimization solution after the FEP has the protocol conversion function: After the ATS data is uploaded to the FEP, it is directly forwarded by the FEP to the PIS system, reducing the data transmission and reception twice, which not only reduces the system's failure points, but also Improved system reliability. Therefore, the future development of FEP must be more and more adapted to such large data volume interface devices.

Research on Redundant Mechanism

The integrated monitoring system has a large number of interfaces and a variety of interface methods. Therefore, the stability of the interface performance is the key to maintaining normal communication between the integrated monitoring system and the subsystem, but the interface equipment cannot reach a 100% failure-free rate. When a certain part of the system fails, normal data can be sent and received through the spare parts to ensure the reliability of the system. This is the most common way to improve system reliability at present-redundant switching. The redundancy mechanisms used in each integrated monitoring system are different, but they have a common purpose, which is to make the system more simple and more reliable.

Redundant mode 1

Most current FEPs use device-level redundancy mechanisms, as shown in the figure.

As shown. FEP sets the redundant flag bit by counting the number of subsystems that it communicates with normally. Most of them are set to "master" and less to "standby". When the communication between FEP1 and the A subsystem is interrupted, FEP1 is connected to the switch card 2 of FEP2 through its internal switch card 1 to complete the communication establishment with the A subsystem and implement redundant switching. The FEP structure of this mechanism is relatively complicated, and it is necessary to configure a switch card internally. The implementation difficulty is mainly concentrated on the routing conversion between the two FEPs, so it is called "redundancy based on the FEP itself".

Redundant method 2

From the perspective of the construction of a comprehensive monitoring system, under the premise of the same redundancy function, the overall plan should be as simple as possible. Therefore, by transferring the redundancy decision right to the server-side FEP driver, replacing the tasks performed by the FEP internal switch card to achieve the redundancy function is called "redundancy based on higher-level drivers", as shown in the figure.

As shown. FEP1 and FEP2 are relatively independent. There is no need to configure a switch card internally, but FEP needs to transmit the connection status with each subsystem to a higher-level FEP driver. The FEP driver determines the link according to the port communication. As shown in the figure, when the connection between FEP1 and A system is interrupted, and the connection between FEP2 and B system is interrupted, the higher-level driver selects FEP1's B system and FEP2's A system as the communication link to achieve redundant functions. The FEP structure of this mechanism is relatively simple, and its implementation difficulty is mainly concentrated on the development of the higher-level FEP driver. Compared with method 1, the advantage of this mechanism is that it has a simple structure, and because it has fewer internal links, it is also easier to maintain.

Combination of two redundancy methods

After all, FEP is an independent device. The server cannot directly detect the true status of the port, so the two methods can be combined, as shown in the figure.

As shown. A heartbeat connection is established between the two FEPs. Through the "heartbeat cable", the FEPs can understand the status of each other's subsystem ports, and then write the configuration file to specify which FEP driver interacts with which FEP subsystem. Data; only after the "heartbeat" is interrupted, the decision is handed over to the upper FEP driver for selection, so that the original redundant function of FEP will not be lost, and the reliability of data in the event of multiple points of failure is improved.