Speady said:
The inertial frame is that of Earth / Mars (they don't move in each other's direction).
Move “in each other’s direction” or “move relative to one another”? I‘m assuming that you meant the latter, which makes the problem much easier to analyze. And since then I’ve seen your clarification that when you said
the final flash on Earth will also be observed neatly 4 minutes later (and 4 light minutes further than at the start of the measurement) than the initial flash.
the bit in parentheses was intended to say that the length of the pulse was four light-minutes.
The last ambiguity is whether the “distance between the start and the end flash” for each satellite is the distance using the frame in which that particular satellite is at rest, or the distance using the frame in which Mars and the Earth are at rest; presumably you mean the former because it’s trivially 4 light-minutes using the latter.
OK, with all that specification out of the way so we agree about the problem...
This is actually pretty easy. Using the frame in which the Earth and Mars are at rest, write down the t and x coordinates of six events:
1) the leading edge of the pulse reaches the first satellite
2) the trailing edge of the pulse reaches the first satellite
3) the leading edge of the pulse reaches the second satellite
4) the trailing edge of the pulse reaches the second satellite
5) the leading edge of the pulse leaves Mars
6) the trailing edge of the pulse leaves Mars
Next use the Lorentz transformations to calculate the coordinates of events 1,2,5,6 using a frame in which the first satellite is at rest; and again to calculate the coordinates of events 3,4,5,6 using a frame in which the second satellite is at rest.
Now you can read off from these coordinate values the distance traveled by both edges of the pulse, the width of the pulse, the time between the arrival of the edges of the pulse, as measured by each satellite. I have no idea if it will turn out to be the numbers you’re suggesting - 3000 km shorter (shorter than what? Than as measured using the frame in which Earth and Mars are at rest?) for one satellite, 3000 km longer for the other - but whatever it is, that’s how we calculate it.