The emergence of fieldbus has greatly promoted the realization of device-oriented automation systems. However, fieldbus, as a dedicated real-time communication network, suffers from drawbacks such as high cost, low speed, and limited application support. Furthermore, the diversity of bus communication protocols prevents different bus products from interconnecting, interoperating, and sharing, thus severely limiting the further development of fieldbus industrial networks. With the development of Ethernet technology, especially the emergence of high-speed Ethernet, Ethernet has overcome its inherent limitations and entered the industrial field as Industrial Ethernet. This allows Ethernet devices to replace expensive industrial network equipment.
1. The main drawbacks of Ethernet
Before discussing the main drawbacks of Ethernet, it's necessary to understand its communication mechanism. Ethernet refers to a network that follows the IEEE 802.3 standard and can transmit data over fiber optic cables and twisted-pair cables. It first appeared in 1972, created by Xerox PARC. Current Ethernet uses star and bus topologies, with transmission rates of 10Mb/s, 100Mb/s, 1000Mb/s, or higher. The main cause of latency in Ethernet is collisions, which arise because it utilizes CSMA/CD technology. In traditional shared networks, since all stations in an Ethernet network use the same physical medium for connection, this means that when two devices send signals simultaneously, signal collisions will occur. To solve this problem, Ethernet stipulates that before a station accesses the medium, it must first listen to see if any other stations are using the medium simultaneously. If so, it must wait, and this is when a collision occurs. To reduce the probability of collisions, Ethernet often uses algorithms such as 1-persistent CSMA, non-persistent CSMA, and P-persistent CSMA. Because Ethernet was designed for office automation, it does not fully meet the requirements of industrial environments and standards, and using traditional Ethernet in industrial fields still has significant shortcomings. However, its cost is lower than that of industrial networks, and its technology is more transparent. In particular, its compliance with the IEEE 802.3 protocol has opened up new possibilities for various fieldbus manufacturers. However, to make Ethernet meet process requirements, the following drawbacks must be overcome:
Certainty
Because Ethernet's MAC layer protocol is CSMA/CD, collisions are possible on the network, especially under heavy network load. In an industrial network, a large number of collisions necessitates multiple data retransmissions, significantly increasing the uncertainty of inter-network communication. This uncertainty, transmitted from one point to another, inevitably leads to a decrease in system control performance in industrial control networks.
Real-time
In industrial control systems, real-time can be defined as the measurability of a system's response time to an event. That is, after an event occurs, the system must respond within a predictable timeframe. However, industrial applications have extremely stringent real-time requirements for data transmission, often updating data within tens of milliseconds. Due to Ethernet's CSMA/CD mechanism, when a conflict occurs, data must be retransmitted, potentially attempting up to 16 times. Clearly, this conflict resolution mechanism comes at the cost of time. Furthermore, even a brief disconnection, lasting only a few seconds, can potentially halt entire production or even lead to equipment damage and personal injury.
reliability
Ethernet was not designed for industrial applications. When applied in industrial environments, harsh conditions and severe inter-line interference inevitably reduce its reliability. Industrial networks in production environments must possess high reliability, recoverability, and maintainability. This means ensuring that the failure of any component in a network system does not lead to the collapse or paralysis of applications, operating systems, or even the entire network system.
2. Ethernet Industrial Application Solutions
To address the three major shortcomings of Ethernet and the specific requirements of industrial networks, various methods have been employed to improve Ethernet performance and quality to meet the demands of industrial applications. Several solutions are described below:
Switching technology
To alleviate network congestion under heavy Ethernet loads, Ethernet switches can be used. These switches employ effective collision domain partitioning techniques within a shared local area network (LAN). Switches connect the various collision domains to reduce collisions and erroneous transmissions caused by the CSMA/CD mechanism. This minimizes collisions and improves system determinism, but it is costly and introduces latency during allocation and buffering.
High-speed Ethernet
We know that the higher the network load, the greater the probability of collisions. Data shows that when a network load is below 36%, collisions are virtually nonexistent. At loads below 10%, the collision probability for 10M Ethernet is once every five years, and for 100M Ethernet, it's once every 15 years. However, above 36%, the probability of collisions increases exponentially with increasing load. Clearly, increasing Ethernet communication speed can effectively reduce network load. Fortunately, high-speed Ethernet with communication rates of 100M/s and 1G/s is now available. With meticulous and comprehensive design and control over the number of network nodes and communication traffic in the system, Ethernet can be fully adopted as an industrial network.
IEEE 1588 time synchronization mechanism
IEEE 1588 defines a Precision Synchronization Protocol (PTP) for precise clock synchronization in measurement and control networks, relating to network communication, local computation, and distribution. This protocol is not exclusive, but it is particularly well-suited for Ethernet-based technologies, achieving microsecond-level accuracy. It uses a time stamping mechanism to synchronize local time. Even when synchronization control signals fluctuate during network communication, its accuracy remains satisfactory. This makes it especially suitable for Ethernet-based systems. By adopting this technology, the Ethernet TCP/IP protocol can operate in high-precision network control systems without major modifications. Its accuracy in area buses far exceeds that of existing systems. Furthermore, using Ethernet TCP/IP-based network technology at all levels of an enterprise offers significant advantages.
A simple system incorporating the IEEE 1588 time synchronization mechanism includes at least one master clock and multiple slave clocks. If multiple potential master clocks exist simultaneously, the active master clock is determined according to an optimized master clock algorithm. All clocks continuously compare their clock attributes with the master clock. If a new clock joins the system or an existing master clock disconnects from the network, the other clocks will re-determine the master clock. If multiple PTP subsystems need to be interconnected, this must be achieved using boundary clocks. One port of a boundary clock is connected to a subsystem as a slave port and provides the clock standard for the entire system. Therefore, the master clock of this subsystem is the original master clock of the entire system. The other ports of the boundary clock act as master ports, through which synchronization information is transmitted to the subsystems. The ports of the boundary clock are ordinary clocks to the subsystem.
The precise network synchronization protocol defined by IEEE 1588 achieves a high degree of synchronization in the network, eliminating the need for dedicated synchronization communication when allocating control tasks. This separates communication timing from application execution timing. Due to its high-precision synchronization, the data transmission time fluctuations inherent in Ethernet technology are reduced to an acceptable level, without affecting control accuracy. A major advantage of IEEE 1588 is its representativeness and openness. Because of its openness, many control system suppliers now use this standard in their products. Furthermore, different equipment manufacturers adhere to the same standard, ensuring good synchronization between their products.
3. Prospects and Outlook for Industrial Ethernet
Industrial Ethernet, with its unique advantages of low cost, high efficiency, high scalability, and high intelligence, is attracting more and more manufacturers. On the one hand, with so many manufacturers researching and developing Industrial Ethernet technology, without unified standards, similar to the situation with fieldbus, numerous standards lead to poor compatibility, thus hindering the development of Industrial Ethernet. Therefore, the international community has begun to develop an Industrial Ethernet standard. At the third meeting of the International Industrial Ethernet Standards Drafting Working Group (IEC/SC65C/WGs) held in Beijing in May 2004, a preliminary international Industrial Ethernet standard was already visible. This series of standards was finalized in August 2005, and after two rounds of consultation in February and December 2006, it was officially released in the second half of 2007. This made the series of standards IEC Standard 6, evolving from an IS standard. On the other hand, the rapid advancements in Ethernet and communication technologies have also spurred the further development of Industrial Ethernet technology. Industrial Ethernet technology is now moving towards real-time Industrial Ethernet and wireless Industrial Ethernet. In particular, B&R of Austria has developed Ethernet Powerlink, a truly real-time Ethernet technology, and in the near future, a new generation of industrial Ethernet components for future industrial networks will also emerge. Because Ethernet is known for its "endless connectivity," meaning it can extend all the way to the enterprise's field device control layer, the development of industrial Ethernet technology will replace current fieldbus-based industrial networks and become the mainstream technology in industrial networks.
