A CPU watchdog reboots a frozen processor. A communications watchdog does the analogous job for a link - it decides when a device has stopped talking and forces a safe response.
Communications Watchdog in one line: A communications watchdog is a timer that monitors the link between two systems - typically SCADA and an RTU, or a controller and a remote device - and takes action if valid messages stop arriving within a set interval. When the timer expires, it declares a comm failure and drives predefined responses: raising an alarm, invalidating stale data, and, critically, moving outputs to a safe state so equipment is not left running on a lost command.
A hardware watchdog timer inside a controller watches the CPU: if firmware stops petting it, the watchdog resets the processor. A communications watchdog instead watches a link. It does not reset a chip; it reacts to the absence of messages from a communication partner. The two are related in spirit - both detect a hang and force a safe response - but they operate at different layers.
The distinction matters because a device can be perfectly healthy while its link is dead, or vice versa. A CPU watchdog will not notice a cut cable, and a communications watchdog will not notice a locked-up task that is still sending heartbeats. Robust systems use both.
The watchdog is fed by evidence of communication - most often an incrementing heartbeat or simply the arrival of valid poll responses. Each valid message resets the timer. If no valid message arrives before the timeout, the watchdog fires. The timeout is set to allow for a few missed messages so momentary noise does not trip it, while still catching a genuine outage promptly.
The most important action on a timeout is failsafe output handling. Consider an RTU that receives a run command from SCADA over a link that then fails: without a watchdog, the RTU would happily keep the equipment running on that last command indefinitely. A communications watchdog in the RTU detects the lost link and drives the output to its configured safe state - stopping a pump, closing a valve, or holding the last value - depending on what is safe for that process. It also flags the affected data as bad quality so upstream logic and operators do not trust it.
Remote oil and gas sites lose their links regularly, so comms watchdogs are standard in RTUs and controllers there. An injection pump or chemical dosing skid that takes commands from a central system must have a defined behavior when that system goes silent - typically shutting down or holding safe rather than continuing blind. The watchdog is what enforces that behavior automatically.
On the SCADA side, a communications watchdog per device is what produces the comm-fail alarms operators rely on to know which sites have gone dark. When it trips, the SCADA marks that device's tags bad-quality so trends, calculations, and alarms are not driven by frozen values, and it notifies operators or an on-call escalation.
Tuning the timeout is a real engineering decision. Too short and normal jitter on a cellular or radio link produces nuisance comm-fail alarms; too long and a genuine outage goes unnoticed for minutes while equipment runs on stale commands. The right value reflects the link's real latency and the safety consequence of acting late.
The heartbeat is the periodic signal of life that the sender emits. The communications watchdog is the timer on the receiving side that waits for it - and takes action, such as failsafe outputs and alarms, if it does not arrive in time. They work as a pair.
The device drives its outputs to a configured safe state rather than holding the last command indefinitely - stopping a pump, closing a valve, or holding a safe value, depending on the process. It also flags the data as bad quality so upstream systems do not act on frozen values.
Set the timeout to tolerate a few missed messages, matched to the real latency and jitter of the link. A cellular or radio link needs a longer timeout than a wired one. Too short a value trips on normal variation; too long a value delays detection of a real outage.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.