When a remote site's numbers stop changing on the SCADA screen, an operator faces an unnerving question: is the process genuinely holding steady, or has the gateway quietly died and frozen its last values in place? A gateway heartbeat tag exists to answer exactly that. It is a value the gateway deliberately keeps changing on a fixed cadence so that the host has one tag it can watch that must always be moving. This page explains what a heartbeat tag is, how it distinguishes a live-but-idle site from a silently dead one, and how it differs from the hardware watchdog that runs inside the device.
Gateway Heartbeat Tag in one line: A gateway heartbeat tag is an application-level value that a field gateway continuously changes and publishes, typically an incrementing counter or a toggling bit that updates every cycle, so the host can confirm the device is alive and reporting. If the heartbeat stops advancing within an expected window, the host knows the gateway or its link has failed, even when every process value would otherwise look plausibly unchanged.
The trap the heartbeat tag guards against is a stale-value failure. If a gateway locks up, loses its link, or crashes in a way that leaves the last transmitted values sitting in the host's cache, those numbers keep displaying as if they were current. A tank level of forty percent looks entirely reasonable, so nothing on the screen betrays that the reading is now hours old. During normal operation many values legitimately sit still for long stretches, so the host cannot tell a healthy, idle site from a frozen one simply by watching whether numbers move.
A heartbeat tag removes the ambiguity by being the one value that is never allowed to be still. The gateway increments a counter or flips a bit on every cycle and publishes it alongside the real data, so the host has a tag that must advance on a known schedule. Real process values can hold steady all they like; the heartbeat cannot. If the counter keeps climbing, the device is alive and reporting, and its flat readings simply reflect a quiet process. If the counter stops, the device or the link is down regardless of how sensible the frozen values appear.
The host detects a failure by watching the heartbeat's freshness, not by comparing it to any particular number. It notes the expected update interval and raises a stale-data alarm when the tag has not changed within a tolerance of that interval, allowing for normal jitter and the occasional missed message. This turns silence into an explicit event: instead of an operator eventually noticing that a site feels frozen, the system asserts within seconds or minutes that the site has gone quiet, and can escalate that as its own alarm.
A heartbeat tag is easy to confuse with a watchdog timer, but they act at different levels and protect against different failures. A hardware watchdog lives inside the device: a timer that firmware must keep resetting, and if the firmware stalls and stops petting it, the watchdog reboots the device to recover it locally. Its whole job is to notice that the device itself has hung and to take corrective action on the spot, with no involvement from the host.
A heartbeat tag, by contrast, is an outward signal aimed at the host. It does not restart anything; it simply proves to a remote observer that the device is running and its telemetry path is intact end to end. Crucially, a watchdog cannot report a problem it cannot survive: if the gateway loses power or its modem dies, the watchdog is powerless to tell anyone, because the device is gone. The heartbeat tag catches exactly that class of failure, because from the host's side an absent heartbeat looks the same whether the device hung, lost power, or lost its link.
The two are complementary rather than competing. The watchdog handles local recovery, giving a hung device a chance to reboot itself back into service without a truck roll. The heartbeat tag handles remote awareness, ensuring that when a device cannot recover itself, someone at the host end learns about it quickly. A well-instrumented gateway uses both: the watchdog to self-heal where it can, and the heartbeat so that when it cannot, the failure surfaces as an alarm rather than as a screen full of quietly aging numbers.
For a cloud SCADA platform such as Merobix that supervises many unmanned sites over cellular and satellite links, the heartbeat tag is a foundational element of trust in the data. Because no one is standing at these wellpads and stations, the platform is the only thing that will notice a site has gone dark, and a per-gateway heartbeat gives it a definitive, low-cost way to do so. Every site publishes a value that must keep changing, and the platform continuously checks that it does.
This lets the platform distinguish, across a whole fleet, between sites that are genuinely quiet and sites that have failed, without an operator having to inspect each one. A stalled heartbeat becomes a communication-failure alarm that can be routed and escalated like any process alarm, so a dead gateway triggers a callout instead of hiding behind believable stale values. Over a large deployment this is what keeps a screen of green from being a false comfort: the platform has actively confirmed each site is still speaking, not merely displayed whatever it last heard.
It also protects every calculation and report the platform builds on top of that data. Allocations, totals, and trends are only meaningful if the underlying values are current, so a stalled heartbeat is a signal to treat a site's data as suspect until it recovers, rather than to keep accumulating totals from a frozen feed. Tying data quality to the heartbeat in this way means the platform can flag or hold affected figures automatically, which is far safer than silently rolling a stale reading into a day's production number.
Usually something deliberately simple that is guaranteed to change every cycle: an incrementing counter that rolls over when it maxes out, or a bit that toggles between zero and one on each update. The specific value does not matter; what matters is that it advances on a known schedule. The host watches whether the tag is changing within its expected interval, not what number it holds.
A watchdog timer lives inside the device and reboots it locally if the firmware stalls, taking corrective action on the spot. A heartbeat tag is an outward signal that proves to the host the device is alive and reporting; it fixes nothing but reveals failures, including total power or link loss, that a watchdog cannot survive to report. They are complementary: one self-heals, the other raises awareness.
Because process values legitimately hold steady for long stretches during normal operation, so a flat reading is not evidence of a fault. A frozen gateway and a genuinely idle site look identical if you only watch the real data. A heartbeat tag is defined never to be still, so its stalling is unambiguous evidence of a failure regardless of how plausible the frozen readings appear.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.