FPGA-based processor redundancy system

The FPGA-based processor redundancy system addresses synchronization and fault detection issues by using an arbitration processing unit for centralized management, ensuring stable operation and high availability through time-sharing synchronization and real-time fault detection.

KR103043057B1Active Publication Date: 2026-11-06IMT INC
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
KR1020260018173
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-01-29
Publication Date
2026-11-06
Estimated Expiration
2046-01-29

AI Technical Summary

Technical Problem

Existing processor redundancy systems face issues such as incorrect data introduction due to malfunctioning processors, incomplete state synchronization, delayed master switching, lack of scalability and flexibility, high maintenance costs, and difficulty in real-time failure detection.

Method used

An FPGA-based processor redundancy system that uses an arbitration processing unit to manage communication and signal control between redundant processors, implementing time-sharing synchronization and centralized monitoring, with features like shared storage, heartbeat signals, and real-time fault detection to ensure stable operation.

Benefits of technology

Enables stable continuous operation without interruption by preventing data collisions and ensuring perfect state synchronization, improving system availability, especially in high-reliability environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026012604038-PAT00001_ABST
    Figure 112026012604038-PAT00001_ABST
Patent Text Reader

Abstract

The present invention discloses a processor redundancy system based on an FPGA, comprising a first processor and a second processor, wherein communication synchronization, fault detection, and failover between the two processors are centrally managed using an FPGA, thereby enabling stable continuous operation without interruption even if a failure occurs during system operation, ensuring perfect state synchronization without data collision through time-sharing control so that data can be processed stably by inheriting the current state in the event of a failure, significantly improving system availability even in environments requiring high reliability such as aircraft, and enabling complex control algorithms to be extended at the hardware level without firmware modification.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a processor redundancy system based on an FPGA, and more specifically, to a technology that uses an FPGA to redundantly redundantly redundantly two processors, monitor their status, and implement stable control transfer. In particular, the invention relates to a processor redundancy system based on an FPGA in which an FPGA acting as a hardware hub is responsible for the physical interface and signal transmission between the two processors, and the availability of the system can be maximized through hardware-based status monitoring and time-sharing synchronization. Background Technology

[0003] Generally, equipment equipped with processors is designed with redundancy to ensure stable and continuous operation. In other words, by providing two identical processors (redundant processors), normal processing can be continued by the other processor even if one fails.

[0004] In this case, the processor currently in operation is called the Master Processor, and another processor provided to prepare for the failure of the Master Processor is called the Slave Processor.

[0005] However, the processor redundancy structure according to the prior art had the following limitations because it relied mainly on direct communication between processors.

[0006] First, the lack of a function to immediately isolate malfunctioning processors can introduce incorrect data into the entire system.

[0007] In addition, there is incomplete state synchronization where state discrepancies occur between the two processors due to time differences or signal collisions during direct communication, and master switching may be delayed due to the absence of relay hardware in the event of a failure.

[0008] Furthermore, because it is dependent on specific hardware connections, it lacks scalability and flexibility due to the difficulty of changing protocols or adjusting communication speeds, and maintenance costs are high.

[0009] Finally, there was a problem in that it was difficult to detect in real-time which processor was the issue when a failure occurred, the speed of identifying the cause of the problem was slow, and it was difficult to centrally monitor the processor and communication status. Prior art literature

[0010] Published Patent No. 10-2006-0067076 (Published June 19, 2006) The problem to be solved

[0011] Accordingly, the present invention was devised to solve the problems of the prior art, and aims to provide a highly reliable processor redundancy system that uses an FPGA as a hardware hub for communication and signal control between redundant processors, thereby providing robust fault tolerance against complex problems such as interface errors and timing delays as well as processor failures.

[0013] The technical problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0015] To achieve the above objectives, the processor redundancy system based on an FPGA according to the present invention comprises a processor redundancy system including a first processor and a second processor, and includes an arbitration processing unit connected to the first processor and the second processor to manage the operating status of the processors and control the data processing time. The arbitration processing unit generates a synchronization signal having a certain period based on a reference clock, generates a first event trigger in the first processor in response to a first level change time of the synchronization signal, generates a second event trigger in the second processor in response to a second level change time of the synchronization signal, controls the first processor and the second processor to process system data mutually exclusively during each time-sharing interval distinguished by the first and second event triggers, and collects status monitoring information output from each of the first processor and the second processor and response information to the first and second event triggers, and controls the system to automatically switch the master authority by checking the normal status of other processors when a failure of the processor currently holding master authority is detected.

[0016] The above-mentioned mediation processing unit includes a shared storage unit internally, and the first processor and the second processor may be controlled to synchronize data by recording processing data and their own state information in the shared storage unit or reading information recorded by another processor from the shared storage unit during each of the time-sharing intervals.

[0017] The above status monitoring information may include at least one of a watchdog signal indicating an abnormality in the internal program of the processor and a parity signal indicating a data error.

[0018] The above arbitration processing unit detects in real time an external system failure, which is a physical or electrical abnormal state occurring in an external environment of the system, and outputs an immediate system interrupt signal to the first processor and the second processor upon detection of the external system failure, and the external system failure may include at least one of an external power input failure (AC Fail), a power management circuit failure (SYS Fail), and a temperature abnormality due to system overheating.

[0019] The above arbitration processing unit may output a hardware reset signal to a processor determined to be faulty during the automatic switching of the above master authority to initialize the processor and isolate it from the system.

[0020] The above arbitration processing unit checks the control status signal levels of each of the first and second processors upon system startup, and in the initial state where both processors are at normal levels, it prioritizes designating one of the first and second processors as the master, and if a failure is detected in one of the processors and the other of the first and second processors is at a normal level, it can control the switching of the master to the other processor.

[0021] The above arbitration processing unit may release the master designation and output a reset signal to both processors to switch the system to safe mode if it is determined that both the first processor and the second processor are in an abnormal state.

[0022] The above arbitration processing unit generates a heartbeat signal as the synchronization signal, and the processor designated as the master among the first processor and the second processor records processing data and its status information in the shared storage unit during its own time-sharing interval allocated by the heartbeat signal, and the processor designated as the slave among the first processor and the second processor can synchronize data by reading the information recorded by the master processor from the shared storage unit during another time-sharing interval allocated so as not to overlap with the recording operation of the master processor.

[0023] When the first processor is a master, the first processor writes data to the shared storage unit in response to the rising edge of the heartbeat signal, and the second processor reads data from the shared storage unit in response to the falling edge of the heartbeat signal; when the second processor is a master, the second processor writes data to the shared storage unit in response to the falling edge of the heartbeat signal, and the first processor reads data from the shared storage unit in response to the rising edge of the heartbeat signal. Effects of the invention

[0025] The FPGA-based processor redundancy system according to the present invention utilizes an FPGA to centrally manage communication synchronization between two processors, fault detection, and failover, thereby enabling stable continuous operation without interruption even if a failure occurs during system operation.

[0026] In addition, the present invention ensures perfect state synchronization without data collisions through time-sharing control, thereby enabling stable data processing by taking over the current state in the event of a failure.

[0027] In addition, the present invention has the effect of significantly improving system availability even in environments requiring high reliability, such as aircraft.

[0028] In addition, the present invention has the effect of enabling complex control algorithms to be extended at the hardware level without modifying the firmware. Brief explanation of the drawing

[0030] FIG. 1 is an overall configuration diagram of processor redundancy according to one embodiment of the present invention. Figure 2 is a flowchart of an automatic fault switching procedure using an In-Control signal. Figure 3 is a flowchart of an automatic failure switching procedure using heartbeat status. Figure 4 is a timing diagram illustrating heartbeat edge-based time-division synchronization and DPRAM access. Specific details for implementing the invention

[0031] A preferred embodiment of the present invention will be described in detail below with reference to the accompanying drawings. The following detailed description is merely illustrative and illustrates a preferred embodiment of the present invention.

[0033] FIG. 1 is an overall configuration diagram of processor redundancy according to one embodiment of the present invention.

[0034] Referring to FIG. 1, the processor redundancy system according to the present invention may be configured to include first and second processors (210, 220) and an arbitration processing unit (100) that manages the operating state of the processors between them, controls the timing of data processing, and mediates data and signal flow, and below, the arbitration processing unit (100) may be implemented using an FPGA.

[0035] Here, FPGA stands for Field-Programmable Gate Array, and it is a type of high-performance semiconductor device used in various fields such as AI, image processing, communications, and defense, as it allows users to program logic circuits after manufacturing and reconfigure them for specific purposes, designs hardware logic using HDL (VHDL / Verilog), and has excellent parallel processing capabilities and flexibility.

[0036] The mediation processing unit (100) may be configured to have a shared storage DPRAM (Dual Port RAM) and a control register internally, and to be connected to each processor via an interrupt (INT) line, a reset (RSTOUT) line, and an In-Control signal line for status monitoring.

[0037] The mediation processing unit (100) can directly monitor the status of each processor through the following signals.

[0038] Processor INT is a hardware-based output signal that allows a program running within the processor to be temporarily suspended to prioritize the handling of failure conditions that may occur within the processor, such as the Watchdog state and Parity error.

[0039] Processor INT Status is a processor status response signal to the above-mentioned Processor INT.

[0040] Heartbeat is a synchronization signal that is periodically output by the arbitration processing unit (100) to check whether the processor is operating normally.

[0041] Heartbeat Status is recorded in DPRAM as a signal indicating the processor's heartbeat reception status.

[0042] System INT is a hardware-based fault signal that detects physical and electrical abnormal conditions occurring externally, such as external power, external devices, and peripheral circuits, and is transmitted to an arbitration processing unit (100), and can be used to inspect the system environment status in real time.

[0043] In particular, it includes system-wide failures such as a fault caused by an abnormal power input (AC Fail), a system failure caused by an abnormality in a power management circuit (PMIC) that efficiently converts and distributes power to each component (SYS Fail), and a temperature abnormality caused by overheating, and is a signal that is immediately output to a processor when the arbitration processing unit (100) detects such a system-level fault.

[0044] System INT Status is the processor's status response signal to System INT.

[0045] In-Control is an output signal of a processor that the arbitration processing unit (100) refers to for determining the master processor. It can be used to determine whether the processor is normal.

[0046] RSTOUT is an output signal of the arbitration processing unit (100) for initializing the processor. It can be used to initialize a processor that has failed.

[0047] The arbitration processing unit (100) detects external system failures in real time, which are physical or electrical abnormal conditions occurring in an external environment outside the system, and outputs an immediate system interrupt signal (System INT) to the first processor (210) and the second processor (220) upon detection of an external system failure, and the external system failure may include at least one of an external power input failure (AC Fail), a power management circuit failure (SYS Fail), and a temperature abnormality due to system overheating.

[0049] Figure 2 is a flowchart of an automatic fault switching procedure using an In-Control signal.

[0050] In the process redundancy system according to the present invention, when the system starts up, the mediation processing unit (100) can determine the master processor by checking the level of the control status signal (In-Control) of the two processors.

[0051] For example, the arbitration processing unit (100) can check the level of the control status signal (In-Control) of each of the first and second processors (210, 220) when the system starts, and in the initial state where both processors are at a normal level, it can prioritize designating one of the first and second processors (210, 220) as the master, and if a failure is detected in one processor and the other processor is at a normal level, it can control the switching of the master to the other processor.

[0052] Additionally, the arbitration processing unit (100) may release the master designation to None if it is determined that both the first processor (210) and the second processor (220) are in an abnormal state.

[0053] Afterwards, the mediation processing unit (100) can switch the system to safe mode by outputting a reset signal (RSTOUT) to both abnormal processors.

[0055] For example, the arbitration processing unit (100) of the present invention may determine a master processor according to the rules of Table 1 below.

[0056] Case Processor #1 In-Control Processor #2 In-Control Master Processor note 1 Low Low #2 Initial state 2 Low High #1 3 High Low #2 4 High High None Abnormal

[0057] Here, assume that logic low indicates a normal state and logic high indicates a fault state.

[0059] According to FIG. 2, when the system starts, each of the first and second processors (210, 220) sets the level of the GPIO terminal corresponding to the control status signal (In-Control) according to its own state (S110).

[0060] Here, it is assumed that the setting is low in the normal case and high in the faulty case, and when both are normal, the second processor (220) is designated as the master processor first, and the opposite is also possible.

[0061] The mediation processing unit (100) can determine the state of the second processor (220) by checking the control state signal (In-Control) of the second processor (220), and determines that the second processor (220) is normal if the control state signal (In-Control) is low, and faulty if it is high, and the opposite case is also possible (S120).

[0062] If the second processor (220) is normal, the second processor (220) is maintained in a state designated as a master processor (S130), and if the second processor (220) is abnormal, the state of the first processor (210) can be determined by checking the control state signal (In-Control) of the first processor (210) (S140).

[0063] The mediation processing unit (100) can determine the state of the first processor (210) by checking the control state signal (In-Control) of the first processor (210), and determines the state as normal if the control state signal (In-Control) of the first processor (210) is low, and abnormal if it is high, and the opposite case is also possible (S140).

[0064] If the first processor (210) is normal, the first processor (210) is designated as the master processor and the master control is changed to the first processor (210) (S150).

[0065] However, if the first processor (210) is also abnormal, the arbitration processing unit (100) designates it as None (S160).

[0066] Afterwards, the mediation processing unit (100) can perform initialization by outputting a reset signal (RSTOUT) to either one or both abnormal processors.

[0068] Figure 3 is a flowchart of an automatic failure switching procedure using heartbeat status.

[0069] The arbitration processing unit (100) of the present invention is characterized by generating a synchronization signal (heartbeat signal) having a certain period based on a reference clock, generating a first event trigger in a first processor (210) in response to a first level change time of the synchronization signal, and generating a second event trigger in a second processor (220) in response to a second level change time of the synchronization signal, thereby controlling the first and second processors (210, 220) to perform time-division synchronization operations.

[0070] For example, during each time-sharing interval distinguished by the first and second event triggers, the first processor (210) and the second processor (220) are controlled to process system data mutually exclusively, and status monitoring information output from each of the first processor (210) and the second processor (220) and response information to the first and second event triggers are collected, so that when a failure of the processor currently holding master authority is detected, the normal status of the other processor is checked and the master authority is automatically switched.

[0071] Here, the status monitoring information may include at least one of a watchdog signal indicating an abnormality in the internal program of the processor and a parity signal indicating a data error.

[0072] Referring to FIG. 3, the arbitration processing unit (100) inputs a heartbeat signal, which is a periodic synchronization signal, to the first and second processors (210, 220), and each processor can be controlled to perform a synchronization operation in time division during a preset level interval of the heartbeat signal.

[0073] If the second processor (220) is designated as the current master processor, a heartbeat signal (Heartbeat) is input to the second processor (220) (S210), and the heartbeat status signal of the second processor (220) can be checked to see if the current master processor is abnormal (S220).

[0074] When checking the current status of the master processor, if it is confirmed to be normal, the second processor (220) designated as the current master processor is maintained as the master processor (S260).

[0075] If the state of the second processor (220) is determined to be abnormal, the control status signal (In-Control) of the first processor (210), which is the slave processor, is checked to see if it is normal (Low) (S230). If it is confirmed to be normal, the first processor (210), which is the current slave processor, is designated as the master processor (S250), and the second processor (220), which was the previous master processor, is designated as the slave processor.

[0076] The arbitration processing unit (100) can record the fact of master switching in the control register of the arbitration processing unit (100) and reassign control of the external bus or device.

[0077] A processor determined to be faulty can be initialized by a reset signal (RSTOUT) output from an arbitration processing unit (100).

[0078] However, if the control status signal (In-Control) of the first processor (210), which is a slave processor, is confirmed to be abnormal (High) and both processors are confirmed to be in an abnormal state, the arbitration processing unit (100) can set the master designation to None (S240).

[0079] In this case, the mediation processing unit (100) can switch the system to safe mode by outputting a reset signal (RSTOUT) to both abnormal processors to perform initialization.

[0080] Additionally, the mediation processing unit (100) may output a hardware reset signal (RSTOUT) to the processor determined to be faulty when the master authority is automatically switched due to a failure, thereby initializing the processor and isolating it from the system.

[0082] Figure 4 is a timing diagram illustrating heartbeat edge-based time-division synchronization and DPRAM access.

[0083] The present invention grants mutually exclusive time slots to the first and second processors (210, 220) using the rising edge and falling edge of a heartbeat signal generated in the mediation processing unit (100), and by alternately generating interrupts for each processor at a predetermined interval (e.g., 10ms), the two processors can be controlled to process input data at a constant interval (20ms) at different times.

[0084] Referring to FIG. 4, the arbitration processing unit (100) of the present invention can divide the system clock to generate a heartbeat signal of a predetermined period, for example, a period of 20ms. The present invention can utilize the level change point (Edge) of this signal as an event trigger signifying the start of a processor's operation.

[0085] The mediation processing unit (100) includes a shared storage unit internally, and the first processor (210) and the second processor (220) can be controlled to synchronize data by writing processing data and their own state information to the shared storage unit or reading information written by another processor from the shared storage unit during each time-sharing period.

[0086] Here, the shared storage unit may be a DPRAM or a control register provided inside the arbitration processing unit (100).

[0087] The processor designated as the master among the first processor (210) and the second processor (220) records processing data and its status information in the shared storage during its own time-sharing interval allocated by the heartbeat signal, and the processor designated as the slave among the first processor (210) and the second processor (220) can synchronize data with the master processor by reading the information recorded by the master processor from the shared storage during another time-sharing interval allocated so as not to overlap with the recording operation of the master processor.

[0088] For example, when the first processor (210) is a master, the first processor (210) writes data to the shared storage in response to the rising edge of the heartbeat signal, and the second processor (220) reads data from the shared storage in response to the falling edge of the heartbeat signal, and when the second processor (220) is a master, the second processor (220) writes data to the shared storage in response to the falling edge of the heartbeat signal, and the first processor (210) reads data from the shared storage in response to the rising edge of the heartbeat signal.

[0089] In FIG. 4, if the first processor (210) is a master processor, in the T1 interval (the first processor (210) active interval), when the heartbeat signal becomes a rising edge, the arbitration processing unit (100) generates an interrupt to the first processor (210), and at this time, the first processor (210) processes external data during the allocated time-sharing slot of 10ms, and updates the processed data and its own status information to the shared storage unit and shares it with the second processor (220).

[0090] In the T2 period (the active period of the second processor (220)), at the time when the heartbeat signal becomes a falling edge, the arbitration processing unit (100) generates an interrupt to the second processor (220), which is a slave processor, and the second processor (220) can synchronize data with the master processor by reading from the shared storage unit the information previously written by the first processor (210) to the shared storage unit during the allocated time-sharing slot of 10ms.

[0091] In this way, since the current state and processing results are shared and synchronized between processors, if a failure occurs in one processor, the other processor can be controlled to switch to the master role, take over the current state, and process data.

[0092] Through this mechanism, the two processors are physically blocked from accessing the bus or memory at the same time, thereby preventing data collisions at the source and allowing the newly designated master processor to continue data processing while minimizing latency.

[0093] The arbitration processing unit (100) may output a hardware reset signal (RSTOUT) to the processor determined to be faulty during the automatic switching of the master authority, thereby initializing the processor and isolating it from the system.

[0095] Through the above-described configuration, the FPGA-based processor redundancy system according to the present invention centrally manages communication synchronization between two processors, fault detection, and failover, thereby enabling stable continuous operation without interruption even if a failure occurs during system operation. Furthermore, by ensuring perfect state synchronization without data collisions through time-sharing control, it can stably process data by taking over the current state in the event of a failure, and has the effect of significantly improving system availability even in environments requiring high reliability, such as aircraft.

[0096] Although the present invention has been described and illustrated above based on preferred embodiments to exemplify the principles of the invention, the invention is not limited to the configuration and operation as described and illustrated. The embodiments described above should be understood as illustrative in all respects and not restrictive. The scope of the invention is defined by the claims set forth below, and all modifications or variations derived from the meaning and scope of the claims and equivalents thereof should be interpreted as being included within the scope of the invention. Explanation of the symbols

[0098] 100: Arbitration Processing Unit 210, 220: 1st and 2nd processors Heartbeat: Heartbeat signal In-Control: Control status signal System INT: System interrupt signal RSTOUT: Reset signal

Claims

Claim 1 A processor redundancy system comprising a first processor and a second processor includes an arbitration processing unit connected to the first processor and the second processor, which manages the operating state of the processors and controls the data processing time; wherein the arbitration processing unit generates a synchronization signal having a fixed period based on a reference clock, generates a first event trigger in the first processor in response to a first level change time of the synchronization signal, generates a second event trigger in the second processor in response to a second level change time of the synchronization signal, and controls the first processor and the second processor to process system data mutually exclusively during each time-sharing interval distinguished by the first and second event triggers, and includes a shared storage unit internally, wherein the processor designated as the Master among the first processor and the second processor records processing data and its status information in the shared storage unit during the time-sharing interval allocated to it, and the processor designated as the Slave reads the information recorded by the processor designated as the Master from the shared storage unit during another time-sharing interval allocated so as not to overlap with the recording operation of the processor designated as the Master, and the data A processor redundancy system characterized by controlling to synchronize, collecting status monitoring information output from each of the first processor and the second processor and response information for the first and second event triggers, and, when a failure of the processor currently holding master authority is detected, primarily determining the failure using the response information of the processor holding master authority, secondarily confirming whether the other processor is normal by checking the control status signal level of the other processor, and, if the other processor is at a normal level, controlling to automatically switch the master authority to the other processor. Claim 2 delete Claim 3 A processor redundancy system according to claim 1, characterized in that the state monitoring information includes at least one of a watchdog signal indicating an abnormality in the internal program of the processor and a parity signal indicating a data error. Claim 4 A processor redundancy system according to claim 1, wherein the arbitration processing unit detects in real time an external system failure, which is a physical or electrical abnormal state occurring in an external environment outside the system, and outputs an immediate system interrupt signal to the first processor and the second processor upon detection of the external system failure, and wherein the external system failure includes at least one of an external power input failure (AC Fail), a power management circuit abnormality (SYS Fail), and a temperature abnormality due to system overheating. Claim 5 A processor redundancy system according to claim 1, wherein the arbitration processing unit outputs a hardware reset signal to a processor determined to be faulty during automatic switching of the master authority to initialize the processor and isolate it from the system. Claim 6 A processor redundancy system according to claim 1, wherein the arbitration processing unit checks the control status signal levels of each of the first and second processors upon system startup, prioritizes designating one of the first and second processors as the master in an initial state where both processors are at a normal level, and controls the switching of the master to the other processor when a failure is detected in one of the processors and the other of the first and second processors is at a normal level. Claim 7 A processor redundancy system according to claim 1, wherein the arbitration processing unit releases the master designation and outputs a reset signal to both processors to switch the system to a safe mode when it is determined that both the first processor and the second processor are in an abnormal state. Claim 8 A processor redundancy system according to claim 1, wherein the arbitration processing unit generates a heartbeat signal as the synchronization signal. Claim 9 A processor redundancy system according to claim 8, wherein when the first processor is a master, the first processor writes data to the shared storage unit in response to the rising edge of the heartbeat signal and the second processor reads data from the shared storage unit in response to the falling edge of the heartbeat signal, and when the second processor is a master, the second processor writes data to the shared storage unit in response to the falling edge of the heartbeat signal and the first processor reads data from the shared storage unit in response to the rising edge of the heartbeat signal.

Citation Information

Patent Citations

  • Method for controlling reset of lockstep replicated processor cores and lockstep system using the same

    KR101560497B1

  • A Master Authority Set Appartus and Method usingWatchdog Interrupt in a Duplicated Board

    KR1020010087547A

  • Method for synchronization and desynchronization in a multiprocessor system and the multiprocessor system

    KR1020070083772A