-- A transmitter has 50 mW output power.
50 mW is equivalent to +17 dBm.
(+17 is a magic number. It tells us that the transmitter could very well be based on a single modulated Gunn diode oscillator, which, after resonating and filtering to remove the unwanted puree, hash, and garbage, typically delivers right around +17 dBm at the output.
-- The power passes through a piece of lossy cable, where it loses 3 dB.
+17 dBm went into the cable. +14 dBm came out of the other end.
(The lost 3 dBm warmed the cable.)
-- The power was then coupled (losslessly) to an antenna with +16 dB "gain".
+14 dBm went into the antenna. It was shaped and focused so that coming out of the antenna in a certain direction, it sounded as loud as a source that's radiating (+14 + 16) = <em>+30 dBm = 1 watt</em> .
This is NOT 1 watt of real power output. The antenna has no batteries, it isn't plugged into a wall outlet, and it has no actual 'gain'.
That 1 watt is "eirp" . . . <em>"</em><em>E</em><em>ffective </em><em>I</em><em>sotropic </em><em>R</em><em>adiated </em><em>P</em><em>ower"</em>. The antenna focuses most of its power in one certain narrow direction, and then, <u><em>in that direction</em></u>, it sounds as loud as an antenna would that took 1 watt and spread it equally in <em>all</em> directions.