Application of remote time synchronization in distributed measurement and control and real-time simulation system Application of
remote time synchronization in distributed measurement and control and real-time simulation system
1. Introduction to
Distributed Measurement and Control System and Real-time Simulation System Distributed measurement and control system usually consists of multiple subsystems, which coordinate their work to complete measurement and control tasks together. Distributed measurement and control system can relieve the burden of single-machine measurement and control system. With the increasing development and maturity of measurement and control technology, modern engineering experiments, especially large-scale military experiments, require more and more types of items to be tested and controlled. The real-time, synchronization, and measurement accuracy of various measurement and control projects are proposed. Higher requirements.
The stepwise real-time simulation system adopts the same structure, standard and algorithm, and connects different types of simulation applications scattered in different geographical locations with the real world through the network, and supports real, virtual and heterogeneous platform-level simulations distributed in different places. Data exchange and interoperability between applications establish a comprehensive simulation environment in which people can participate in interaction. Due to the limited area of the test site, the large-scale simulation non-standard equipment is often distributed in the test rooms of different buildings. The completion of the simulation test requires multi-building and multi-laboratory cross-domain joint conduct. This puts forward an urgent demand for remote interconnection and parallel testing of multiple laboratories, and the real-time data interaction technology in remote collaborative simulation technology is the key to solving the above problems.
2. Distributed real-time simulation system needs remote time synchronization.
Distributed simulation technology combines computer network technology and virtual reality technology, adopts a coordinated structure, standard, protocol and database, and integrates the software of each simulation node through a local area network/wide area network. The hardware and simulation environment are integrated to complete the simulation task together. With the continuous development and improvement of virtual reality technology, distributed simulation technology puts forward higher requirements on the real-time performance, transmission rate, and diversity of transmission data for the communication network connecting various simulation nodes.
1. System time synchronization
In a distributed real-time simulation system, remote time synchronization between various parts of the system is very important. For example, the time synchronization between graphics workstations, drive control systems, simulators, master computers, turntable control computers and other equipment, and how to determine the simulation start time, how to determine the start time of graphics generation, how to determine the data transmission time, and how to read in time Data and so on are all issues that need to be faced.
2. Simulation time synchronization
The starting point of a unified time period is the basis to ensure the normal progress of the simulation and the authenticity of the data. Simulation synchronization requires that each node of the simulation be at the starting point of a unified time period, and advance forward according to the specified simulation step. The unified time period starting point means that each sub-node of the system running simulation needs to use a unified time as the simulation start time stamp, and the simulation step length of each node afterwards is accumulated on the basis of this simulation start time stamp.
Third, the solution of distributed real-time simulation system
In modern test systems, how to ensure the synchronization between various modules and the real-time measurement is a very important topic. Fiber reflective memory network (RMN, Reflective Memory Network) is because of its inherent The characteristics of delay determination, short delay time, and support for cross-platform have been widely used in real-time systems. A fiber optic hub (fiber HUB) can be used to build a real-time fiber reflection memory network hardware platform to realize multi-laboratory collaborative experiment simulation.
Optical fiber reflective memory network is a high-speed real-time computer network sharing memory data. Optical fiber reflective memory network actually installs a reflective memory card in each connected computer, and each reflective memory card is connected by optical fiber.
Based on the optical fiber reflective memory network for remote remote collaborative simulation, it solves the problems of laboratory equipment interconnection and high-speed real-time data sharing. It greatly improves the flexibility of fiber optic networking and wiring in the hardware-in-the-loop simulation laboratory, and can achieve 20km level The cross-regional low-latency real-time data sharing enables the laboratory to have the ability of multiple laboratories to remotely coordinate and carry out multiple experiments in parallel, and it also has the ability of a single laboratory to serve multiple experiments at the same time.
1. The working principle
of optical fiber reflective memory network The optical fiber reflective memory network based on memory sharing technology is a real-time network composed of computer nodes interconnected. To form a fiber reflective memory network, it is necessary to insert a reflective memory card in each computer, so that the computer and the reflective memory card constitute each node of the fiber reflective memory network. Each node is connected by optical fiber and other transmission media. In order to save costs, devices that are closer (within 300m) are connected to the local optical fiber HUB using multimode optical fibers, and devices that are farther apart (300m-20Km) are connected using single-mode optical fibers.
The memory of the reflective memory card of each node has a copy of the shared data of the reflective memory card of other nodes. Each node includes a reflective memory board, and each reflective memory board is equipped with a large-capacity dual-port memory. One end of the dual-port memory is connected to the local bus of the computer, and the other end is connected to the optical fiber after processing such as FIFO, encoding/decoding, and parallel/serial conversion. Each reflective memory board on the network occupies a memory address. When any computer on the Internet writes data to the local reflective memory board at any time, the data and the corresponding memory address will be broadcast to all other reflective memory boards on the network in a very short time. Stored in the same location. Moreover, the update time of all nodes has nothing to do with the number of nodes actually connected on the network. The onboard memory of the fiber interface board of each node can be shared by other nodes, so logically all nodes in the entire network share the same memory. Data is written at one point and updated at multiple points at the same time. Through this broadcast method, high-speed data transmission and sharing are realized. The reflective memory network has a definite very small data delay. The data transmission delay between two nodes is only 400ns. Therefore, on a ring network with a fixed number of nodes, the data transmission delay between any two nodes is determined and can be calculated.
The core component of the optical fiber reflective memory network is the optical fiber reflective memory network interface board, which can use VME, PCI, CPCI, PMC and other bus architectures to connect the computer and the optical fiber network. The entire optical fiber network is connected by optical fiber and optical fiber reflective memory network interface board. The computer completes the writing and reading of the shared memory by operating the interface board. Take GE's PCI5565 interface board as an example. It can support a shared memory of 256MB at most, and the network can accommodate 256 nodes. The theoretical maximum transmission rate can reach 170MB/s. The actual point-to-point test result is as high as 80MB/s.
In the optical fiber reflective memory network, the reflective memory is physically distributed in each computer (in the reflective memory card), and logically shares the same memory address. Any computer can access shared reflective memory as easily as ordinary memory. Because the reflective memory network adopts a simplified network protocol, it has a very high transmission speed and fully meets the requirements of a simulation system with high real-time requirements such as an infrared image real-time generation system.
2. Topological structure of optical fiber reflective memory network
There are two main physical topological structures of the optical fiber reflective memory network: one is a ring topology; the other is a star topology. If the ring topology is selected, all reflective memory boards on the network will be connected in series via optical fibers. If the star topology is used for connection, all reflective memory boards on the network will be connected to the automatic optical fiber bypass board on the optical fiber HUB. This kind of star connection is only a physical star connection, and logically it is still a ring connection.
The advantage of the ring type reflective memory network is that the entire network does not require additional hub resources, and the number of optical fibers used is about half less than that of the star type, and the cost is very low. The disadvantage is that when one node in the ring network fails, the entire reflective memory network will be paralyzed. The star connection uses fiber optic hubs (also called fiber switches) as data relay and forwarding equipment. Each node in the network first transmits data to the hub, and the hub processes the data accordingly and then forwards it to other nodes at the same time. The use of fiber optic hubs can monitor the data flow in real time, bypass the wrong node, and greatly shorten the data update time. Generally speaking, the ring topology is more suitable for reflective memory networks than the star topology.
Fourth, the remote time synchronization solution
In order to solve the remote time synchronization problem of the distributed measurement and control system and the real-time simulation system, the company proposes to adopt the TFT series of high-precision optical fiber time and frequency transmission equipment and time signal reception based on the optical fiber reflective memory network The way of daughter board (PCI or PCIe) realizes the time synchronization between distributed measurement and control system and real-time simulation system.
The entire time system equipment is composed of the master station clock, the slave station clock, the optical fiber hub, the time system signal receiving sub-board (PCI or PCIe) and the supporting connection optical cable, as shown in Figure 3.
The time system equipment is transmitted through high-performance atomic clock and optical fiber exchange, giving precise time or definite time reference to meet the high-precision time synchronization requirements between users in the system. In addition, it can be achieved through Beidou/GPS satellite timing. Long-distance time synchronization and collaboration between other departments and units of the system.
The master station clock outputs a high-precision time-frequency signal reference source, which is transmitted to the slave station clock via optical fiber, and then transmitted to the time-system signal receiving board via optical fiber. The time-system signal receiving sub-board obtains and analyzes the time-frequency information issued by the master station clock, and through calculation and processing, provides users with standard time information and user-set clock synchronization interrupt signals for users to synchronize real-time simulation system data. The optical fiber hub is mainly used for networking with multiple optical fiber reflective memory network interface boards in the real-time simulation system. In addition, the company also provides special portable testing and debugging equipment for reproducing the master clock signal for normal testing and debugging.
Figure 3 Schematic diagram of timing equipment networking
Using Jingzhun's remote time signal synchronization solution, the realized measurement and control system and simulation system can achieve the following technical indicators in terms of time system performance:
- Clock synchronization accuracy is better than ±10ns;
- Frame clock (0.2ms) transmission delay is less than 10μs;
- The stability of the clock source of the master station of the timing equipment is better than 1×10-11/day (satellite signal lock);
- On the time stamp synchronization network, the error between the local synchronization interrupts or synchronization pulses generated by each time synchronization signal receiving board is not more than 40ns;
- The minimum output interval period of the synchronous clock signal is not more than 10μs;
- The real-time optical fiber network communication rate supported by the time system signal receiving board is not less than 2.125Gbps;
- The maximum transmission distance of the clock signal is not less than 5km (the transmission delay of the clock signal is not considered).