Understanding real-time processing, from automation to mobile telephony
A real-time system responds to an external event within a defined and predictable time. It is still necessary to distinguish hard, soft and fixed real time: the requirements have nothing in common.
Real-time capability is often talked about: real-time processing is indispensable in many areas, from computer technology to automation, fieldbus systems and mobile telephony. But what exactly does the notion cover, and why is it so important?
Real-time capability relates to systems that respond to external events or data within a defined and predictable time. The reaction time is so short that the system responds virtually “immediately”. A distinction is made between three types of real time, to be implemented depending on the application requirements.
Hard real time
Strict time conditions must be adhered to: any exceedance of the time limit can lead to a critical malfunction. The typical example is the ABS in a car. When the driver brakes, the ABS must respond immediately to prevent the wheels from locking up. The slightest delay, even in the millisecond range, could have serious consequences. In such systems, any delay beyond the defined time limit is considered a malfunction.
Soft real time
These are systems where occasional delays are acceptable, as long as the performance remains good on average. Video streaming is a good example: if the video usually runs smoothly and only buffers occasionally, the user experience is not significantly impaired. Soft real-time systems are therefore more tolerant, within the limits of acceptable overall performance.
Fixed real time
Less strict than hard real time: time specifications matter, but exceeding them does not necessarily lead to serious consequences. In such systems, the result or the message simply becomes irrelevant or useless once the set time has expired.
From the second to the millisecond range
Requirements vary greatly depending on the application, from seconds to milliseconds. In logistics, in an automated warehouse for example, robots and conveyor systems can be timed based on seconds: moving goods from one point to another matters, but delays of a few seconds can usually be tolerated.
In the event of a collision, on the other hand, the acceleration sensor data must be used to deploy the airbags as quickly as possible, taking the force and position of the impact into consideration. This decision is made within 20 to 30 milliseconds to ensure effective protection; any delay would put the safety of the occupants at great risk.
In the industrial networking of devices, Industrial Ethernet networks are often used, and their real-time capability is also frequently praised. EtherCAT is one example: its development focused on short cycle times (≤ 100 µs) and low jitter. When designing a complete real-time capable system, all subcomponents matter — from the networking of sensors and actuators to data transmission and processing.
In summary
Real-time processing is a key aspect of many technical areas. It refers to the capability of systems to respond quickly to external events. Depending on the requirement, a distinction is made between hard real time, where strict conditions must be observed (ABS), soft real time, which tolerates occasional delays (streaming), and fixed real time, with less strict specifications. The required times range from seconds to milliseconds, or beyond.
