On the sender side, there would need to be a "smell microphone" to inhale and encode smells based on chemical properties. The data can be transmitted over HTTP or another transfer protocol.
On the receiver side, there would be a "smell speaker" (and maybe amplifier) that decodes and exhales the smell.
I can imagine various technical challenges on both ends, specifically chemical analysis and synthesis on a small enough scale to be practical.
EDIT: This page has a history of such attempts, most recently Aromajoin with "the first wearable scent device" in 2016.
> The iSmell failed to get the interest of the public. When looking at what went wrong for the iSmell, it is revealed that the missing link was a market survey. According to Startupover’s Andrea Dusi, the iSmell “was definitely a nice idea, but not a useful one”. DigiScents has shut down due to a lack of funding, although it still “continues to license its technology and is looking for funding for a relaunch. [1]
I wonder if with virtual reality, something like this will pick up.