Chapter 3 SCADA communication
3.6 Data communication
3.6.5 Media access control (MAC) techniques
Communicating devices in a network access the network and send or receive data as per a set plan or are governed by a set of rules. These rules constitute the media access control (MAC). The MAC layer in a communi- cation protocol defines these rules and is a sublayer of the data link layer in a protocol, as explained in Section 3.7.1. However, MAC is discussed separately, as it is essential for a power automation engineer to under- stand these types of media access. The following types of techniques are used for media access:
Selection Techniques: The device to transmit the data is selected in a con- venient manner or sequence, which includes schemes like polling and token pass.
Reservation Techniques: Here the same channel is split into subchannels using physical mechanisms like frequency division multiplexing or time division multiplexing, and a virtual channel is reserved for each node for data transmission.
Contention Techniques: In the contention schemes, the channel is used by the node which is alert and gets hold of the channel for transmis- sion, and schemes include Aloha and CSMA/ CD.
The following short descriptions present some of the schemes used in power system SCADA protocols.
3.6.5.1 Polling
In a master- slave configuration, the master requests data from a slave in a specific preset sequence and receives the data in a sequence. Priorities can be set for some slaves, which can be polled faster if the slave node has critical data. If the slave does not respond, the request is re- sent, and after a number of requests, as set by the program, the slave is marked as non- responding. Polling can be initiated on any topology and physical media. However, it is an inefficient way of media usage for obvious reasons. The major advantage is that collision seldom happens and failure of the chan- nel is detected instantly. The programming is easy and the system is reli- able. However, a major handicap is that the slaves cannot communicate with each other, as data must go to the master and then be redirected, which makes it slow, and also interrupts from a slave cannot be handled.
3.6.5.2 Polling by exception
In this case, the event that has changed after the last polling is requested by the master, and hence with no event change happening, no data are returned. Switch positions change once in a while during a fault or some other disturbance, and most of the time, the positions are unchanged.
This is a more efficient way of communication. In this case, there can be a provision for time tagging of the event, indicating the exact time of the change, which is very important for power system performance evalua- tions. For analog points, only when the value of the point changes beyond its significant dead band between scans is the point reported.
3.6.5.3 Token passing
In the token passing technique, the nodes are numbered in a sequence order so that a logical ring is formed and a token, usually a message, passes around the ring. If a node or station has something to report, it will access the token, and the specific station has complete access to the channel for transmission of data. Once the transmission by the station is complete, the token is released and any other station that has to report can access the token, and the process goes on. Token pass can be used on both ring and bus topologies, where in a physical ring, the token passes around and any break in the channel causes disruption. In a bus topology, the token passes in a logical ring among the stations, and the loss of one node will not affect the communication between other nodes.
This scheme is much more reliable and faster than the polling scheme, as peer- to- peer communication is possible and there will be no collisions. This scheme is usually used by systems where the data transfer require- ments may vary from node to node. The disadvantage is that sometimes a station with an urgent message will have to wait until the token is avail- able to it, and communication failure detection is slow. To detect com- munication failure, techniques such as background polling or integrity polling are used by the traffic administrator, where the nodes are polled continuously at a slower rate to check the health and data status.
3.6.5.4 Time division multiplex media access
In this MAC, each station will get a fixed time slot for data transfer (time- division multiplexing [TDM]) and deterministic time responses are pos- sible without collisions. This is used in bus or ring topology and a traffic controller controls the traffic in the system. The major disadvantage is that interrupts cannot be handled by this method, and it is difficult to detect a communication failure where background polling will have to be done.
3.6.5.5 Carrier sense multiple access with collision detection (CSMA/ CD)
In this scheme, each node checks the channel (multiple access) whether it is free and cannot start the transmission until the channel is free (carrier sense). However, due to the propagation delay in the physical medium, more than one node may transmit messages and there will be collisions. The transmission is stopped by the node when a collision is detected (collision detection), and the message is retransmitted after a random
period. In this technique the channel is used most efficiently; however, each node should have the ability to detect collisions and avoid collisions and message retransmission and recovery schemes. CSMA/ CD is used by Ethernet communications (IEEE 802.3 standard). The major advantage is that peer- to- peer communication is possible and variation in data transfer requirements can be handled well by this scheme. A centralized traffic controller is not necessary and priority access for urgent messages can be built into the protocols. The major drawback is the data collision and detection, avoidance, and recovery schemes have to be well developed at each node. Communication link failure is difficult to detect, and nonde- terministic times of data transfer are further drawbacks.