Well, generally speaking you have two typical ways of checking for service availability - either you periodicaly actively query a service (for example - ping a device) or have the device emit some form of keepalive signal and you just monitor if the last occurence you received is sufficiently "young". Your mechanism fits the second option.
You might - for example - have a script writing events to a file every minute or every five minutes and ingest it into splunk where you would check timestamp of latest event and alert if it was older than predefined limit. It's definitely doable and relatively easy.
Having said that...
It seems a huge overkill to do it with splunk. And it's not the tool meant for this type of monitoring - there are various monitoring suites out there or - if you want to just monitor this one thing - you could just write your own simple script to do so.
And of course - the more components you have, the more complicated it gets - you can be losing events not only because of power outage but also if UF crashed or windows decided it wanted to do an upgrade and didn't survive the reboot... So you wouldn't be just monitoring for power outages.
Of course it could be acceptable for you but it's important to understand how it works. If you really really need to monitor especially for power outages (for example for SLA purposes), not for other incidents, you need some device that is battery powered and which measures the mains voltage and an mains-independent network connectivity. So it's getting complicated. It all depends on what you really need.
But it's much less a splunk issue, more of a general infrastructure and monitoring topic.
Oh, and the Feedback forum doesn't seem like a proper place for this question.
