Think about the data, THEN think about how you want to visualise it.
You have a variable number of devices that might be turned on at given times of the day.
This gives you already a fixed axis to work with: the 24hr in any given day.
On that, depending on what intended device you want your data to be visualised, you might distribute the 24hr timeline in the horizontal axis or the vertical. Mind, though, this will inform the scrolling behaviour in case the number of devices tracked exceeds the screen size.
Now, I can start thinking of how to display time, and the agenda view of a calendar app comes to my mind.
You are basically trying to show events on a timeline. (My assumption)
Something like this below, where instead of showing the days on the horizontal axis you will have the different devices, and the users can scroll horizontally to navigate through different ones.

Now, the questions arise (and this is why it is good to start from the data you have and how you want to use it instead of jumping to how is it better to visualise it):
- Do you need to see the data in real-time? Do you need a "NOW" marker on the timeline?
- When the users look at the data, are they interested in one particular device? Do they want to "compare" different devices with one another? Would they want to move them closer to each other?
- Will the data show trends? Do you need to pick different dates, or perhaps different timeframes (like week-on-eek, instead of a one-day view)?
- Is it important to show when the device went off?
- Is it important to show the count of times a given device was turned on in a given day? What are you trying to compare by placing the data in a graph?
Asking these questions will inevitably guide you in the design decision of what would be the best way to visualise the data at hand.
I could keep going with making assumptions on your behalf, but I would rather get you going on your own for a bit before further narrowing down the design decision for you.