It's so bloody silly, the silicon to do time of flight distance simulation would have added zero to the cost of manufacture. >10 year old cheap PICs can do it, so could they.
There hasn't been justification to do proximity detection based on signal strength in decades, yet we still get systems which do so.
It may be more complex.
Sure, EM travel time on a pulse is easily done. But a simple responce pulse could be spoofed by the relay itself. So the relay would know when to dumbly pass the data, and when to generate a pulse. To be secure, you would have to be measure response time between an encrypted challenge-response pair.
So a challenge token received (at least 64 bit, for a "minor" encryption, but I am sure people will yell about 256 or 512 bit) . Apply some public/private key encryption, in other word, processing time, and send the response back, again at least 64 bit.
When a 1uSec calculation and packet length is involved, it is a whole lot more difficult to determine if it arrived in the correct 5ns window.