Abstract: This paper introduces several object models for the real-time problem of industrial networks and provides an in-depth analysis of these models. It elaborates on various factors that meet the real-time requirements of networks, which is helpful for analyzing the real-time problems of different industrial control systems.
Keywords: real-time performance, industrial networks
Abstract: This article introduces several models about industrial network's real-time problem, also it has a deeply analysis of the models and expatiates various elements to satisfy the demands of industrial network's real-time characteristic, so it can benefit us to analyze the network's real-time characteristic better in different industrial control systems.
Key words: real-time, industrial network
1. Introduction
In industrial communication networks, the most important aspect is to achieve dynamic data exchange between on-site production equipment and monitoring equipment, thereby realizing remote real-time monitoring and real-time data exchange. Furthermore, the system is required to transmit only data relevant to user needs, and transmission delays must be controlled within a certain timeframe; that is, real-time performance is a requirement. Any industrial communication network without real-time performance becomes meaningless. Real-time performance is of paramount importance in the field of industrial monitoring. Therefore, this paper primarily focuses on further research into the dynamic exchange and real-time performance of data in industrial communication networks.
2. The concept of real-time performance
Real-time refers to the completion of signal input, processing, and output within a certain time frame, and timely handling based on changes in production process conditions and on-site circumstances. A real-time system, on the other hand, refers to a processing system that can respond within a specified timeframe as soon as an event or data is generated, processing it quickly enough and promptly sending the results to the destination. Real-time and fast are not the same thing. Regardless of network transmission speed, as long as a response action is generated within the specified response time, the system is considered to have real-time performance. A real-time network, however, refers to a network where data transmission time is deterministic and predictable; in other words, data transmission in a real-time network has a time limit.
In essence, real-time performance constrains the time it takes for a system to respond to input information. The correctness of a system depends not only on the correctness of its information processing results but also on the time it takes to obtain those results. Only when the system processes information correctly and obtains results within a specified timeframe is it considered a real-time system. When information is input into the system, it must respond within a certain time. If the response is correct but exceeds the time limit, the system is considered to have failed. Meeting the real-time requirements of a task means that its response time is less than the specified time limit. Real-time applications generally have two prominent characteristics: first, they are highly time-sensitive, requiring the collection of information from the external environment at specific moments or within a certain timeframe, the retrieval and processing of acquired information according to their interrelationships, and a timely response; second, they process "transient" data, which is only valid within a certain time frame and becomes meaningless after that time.
The real-time performance of industrial communication networks has two meanings: First, it refers to the real-time performance of the basic controllers. Typically, each controller needs to possess a certain level of real-time capability. Generally, each controller undertakes more than one task, but each task has specific and potentially different real-time requirements. These requirements are handled by the "real-time multitasking management program" configured in the basic controller. Second, it refers to the real-time performance of the communication network itself. Basic units with communication capabilities in an industrial communication network are interconnected through the network; these units are called "stations." When a station requests communication from the network, it has specific "response time" requirements. Different stations may have different real-time requirements, and even different communication tasks within the same station may have different real-time requirements. Real-time performance is the most significant characteristic that distinguishes industrial communication networks from ordinary LANs. Generally, the information response time requirement in industrial control networks is 0.01–0.5 seconds, while the information response time in ordinary LANs is 2–6 seconds. This means that the real-time requirements of industrial control networks are much higher than those of ordinary LANs; some industrial control networks have real-time requirements hundreds of times higher than ordinary LANs. This can only be achieved by sacrificing some channel utilization. Clearly, if a regular LAN is subject to time constraints to meet the real-time requirements of industrial communication networks, it can be applied to industrial control systems as their communication network. Generally, the real-time performance of industrial communication networks is mainly related to the following aspects:
① Network hardware performance: This includes network topology, communication media, and network interface transmission rates. The higher the transmission rate of the communication media and the faster the transmission rate of the network interface, the higher the real-time performance of the network.
② Network communication protocols: These include media access control methods, the hierarchical structure of network communication protocols, transmission reliability, connection-free control, etc. The simpler the hierarchical structure, the higher the system's real-time performance. However, reliability and real-time performance are contradictory; connectionless, non-responsive communication methods offer higher real-time performance but lower reliability than connection-based, responsive methods. For example, the PROFIBUS-DP fieldbus protocol uses a master-slave approach combined with low-level token ring passing for channel allocation. The entire network can be segmented into bus systems and established step-by-step, with repeaters connecting segments. Each segment can have 32 network stations, and the entire network can reach 126 stations. Because the maximum transmission rate can reach 12Mbps, and its Layer 2 uses SRD (Send and Request Echo) functionality, input and output data can be completed within one cycle, significantly improving transmission speed and minimizing bus cycles.
③ Network information volume: also known as network load, refers to the amount of information that the network needs to transmit within a certain period of time. The less information the network transmits, the higher its real-time performance.
④ Real-time performance and channel utilization of the communication subnet are contradictory: In industrial control networks, in order to improve the real-time performance of the system, it is necessary to sacrifice some channel utilization.
In addition, industrial communication networks have the following characteristics in the time domain, which distinguish them from other systems:
① Time limit: Tasks performed in industrial communication networks generally have time limit requirements, stipulating that a specific function must be completed within a specific time and cannot exceed this time.
② Real-time control: Industrial communication networks often include real-time control, which receives input data and makes control decisions.
③ "Reactive" system: Industrial communication networks are generally "reactive" systems, meaning they are event-driven and must respond to external events.
④ Concurrent processing: A key feature of most industrial communication networks is concurrent processing, where the order in which events occur is usually unpredictable.
⑤ Interaction with the external environment: Industrial communication networks typically need to interact with the external environment.
3. Real-time object model
When studying the real-time performance of industrial communication networks, it is necessary to consider issues in the time domain, hence the definition of real-time objects is given.
Definition 1: A real-time object can be represented by the following quadruple.
(Equation 1-1)
Where realtime stands for Real-time Object, representing an object with real-time requirements; II stands for Input Interface; OI stands for Output Interface; P stands for Inner Process, representing the object's internal processing; and t stands for Time, representing the object's time requirement. With the real-time object model, the real-time performance of industrial communication networks is mainly reflected in the interaction between real-time objects, that is, satisfying the time requirement t of the real-time objects. The real-time performance of industrial communication networks discussed here aims to find feasible solutions such that tr ≤ tΔ, where tΔ represents the time requirement of a certain real-time object.
In industrial communication networks, synchronization and data exchange generally involve message passing. Real-time communication plays a crucial role in ensuring the timely completion of real-time tasks. Its most important property is the need for a defined, bounded message passing delay. Unpredictable message delays can cause tasks participating in real-time communication to violate time constraints. Message passing delay refers to the time interval between the sending node beginning to send a message and the receiving node fully receiving the message. It mainly includes the following components:
① Message waiting delay within each node communication object: The amount of time a message is blocked on each communication object is caused by messages from multiple channels arriving at a communication object simultaneously.
② Message sending delay: The time required for a node object to send a message. It depends on the message size and sending rate, and is generally a constant.
③ Propagation delay on the link: the time it takes for a data bit to propagate on the link.
The transmission delay and the propagation delay on the link are determined by the network bandwidth and the signal propagation speed, respectively, while the message queuing delay of the node communication object is determined by the system's software structure.
4. Classification of communications in industrial communication networks
Industrial communication networks are complex and integrated systems. The data that needs to be transmitted in network control systems includes both real-time and non-real-time data.
① Real-time data: such as real-time data from the I/O ports of various detectors and controllers, signals, interlock signals between controllers, and some system status monitoring data. Real-time data has stringent time requirements, generally allowing no delays on the order of seconds, and in some special cases, even delays on the order of milliseconds. On the other hand, for most real-time data, only the latest data is meaningful. If, within a certain time period, a piece of data fails to take effect for some reason, and the next piece of data has already been generated, then that data will be discarded, and the latest data will be used. Therefore, real-time data generally does not require retransmission.
② Non-real-time data: such as user programming data, configuration data, and some system status monitoring data. Non-real-time data has less stringent time requirements and allows for relatively long delays, but the data volume is relatively large, resulting in higher bandwidth consumption. For most non-real-time data, the transmitted data is meaningful and generally cannot be lost. Error control and retransmission mechanisms are needed to ensure the integrity and accuracy of the data.
From a time-domain perspective, the transmission of real-time and non-real-time data in industrial communication networks can be summarized into three types of communication: periodic communication, random communication, and burst communication.
① Periodic communication: such as sensors periodically transmitting sampling data, and controllers periodically transmitting control signals.
② Random communication: such as a client requesting services from a server.
③ Sudden communication: such as report information, etc.
5. Periodic Real-Time Message Model
Since most communications in industrial communication networks are periodic, we will focus on periodic real-time communications.
Definition: The message flow in periodic communication can be represented by the following periodic real-time message model:
CM = (L, C, S, D); [Equation 1-2]
In the formula, CM (Cyclic Message) is a periodic message; L (Length) is the length of the periodic message stream, representing the transmission time of the message stream, including all the contents of the message frame, such as the information field, check field, and preamble as specified by the network protocol; C (Cycle) is the communication cycle, that is, the message generation cycle; S (Start Time) is the time when the communication request is sent, that is, the time when the message is generated; and D (Deadline) is the message time limit, that is, the maximum allowable delay time from the generation of the message to its arrival at the destination node.
The periodic message stream k is represented as:
CMK = (Lk, Cx, Sk, Dk); [Equation 1-3]
From the definition of periodic real-time messages, it can be concluded that the real-time performance of periodic communication is determined by the following conditions:
① tp-S≤D, where tp is a certain moment in the message transmission process. This condition means that the time from the generation of the message to its final arrival at the destination node will not exceed its time limit.
② Tmax≤C, where Tmax is the maximum time interval during which a node performing periodic communication gains control of the bus.
③ tR≤L, where tR is the bus control time. This condition means that after a node gains control of the bus, it should have sufficient time to send all periodic messages.
For periodic communication, it is necessary to ensure the real-time performance of all message streams, rather than just one message stream. Therefore, the worst-case scenario in communication must be considered. On the other hand, while ensuring real-time performance, the overall efficiency of the system must also be considered, that is, to make reasonable use of the communication channel and ensure the stability of the channel utilization rate.
6. Conclusion
The above analysis provides a more comprehensive and in-depth understanding of the real-time performance of networks, which will help us better meet real-time requirements and improve the response speed of industrial control systems when designing industrial communication networks.
7. References
1. Industrial Communication Networks for Automatic Control Systems, Tang Jiyang, 2000(1)
2. *Information Network System Integration Technology*, Hu Daoyuan, Tsinghua University Press, 1996.
3. "Design of Real-Time Network Communication for Distributed Control Systems" by Qin Xiaozhen, Computer Engineering and Applications, 1997.5
About the author: Dong Jie (1979-) is a male with a master's degree from Shandong University and a professional title of lecturer. He is currently engaged in teaching and research on computer control systems at Shandong Provincial Youth Management Cadre College and has rich practical experience in projects.
