The SKA interferometer will be able to directly see a planet's atmosphere at a range of 100 light-years. If two or more gasses are present where they react in each other's presence AND the ratio of those gasses is stable over time, you have concrete proof of life. This cannot be achieved by known (or unknown) natural processes, a dynamically maintained equilibrium that would cease to exist through any process other than direct action requires a biological process.
Actually, it requires at least two. Any organism that tries to make things favourable for itself must necessarily alter some second dynamic to be unfavourable to itself. You cannot do more work without producing more byproducts (conservation of matter) that are in a lower energy state (conservation of energy, since energy has been taken out) where some of these are toxic to the organism (if it wasn't, it would be processed for energy and matter until it was toxic).
So, one organism always produces an instability. Two is the minimum. The more you have, the more stable the dynamic becomes as there are increasingly better solutions to the set of equations. If an organism develops that tries to exploit the equilibrium (which is inevitable), the equilibrium is lost and the new organism is put at a deficit. A new equilibrium will emerge as a result.
This, by the way, falsifies Nash's argument against his equilibrium. The equilibrium is an emergent phenomenon, so if the dynamic changes, the equilibrium changes. Nash made an error by assuming a dynamic equilibrium has to itself be around a static point. No. The dynamic equilibrium has one Strange Attractor per class of actor in the system. That really should have been obvious and I'm honestly shocked Professor Nash did not see this in his original work or his later appraisal.
Now we get onto communication. Could, in principle, a SKA-class array or the half kilometre single dish in China, be used to communicate at a distance of 100 LY to a civilization of like ability?
Much more difficult. The so-called waterhole is the obvious line to use, as there is virtually nothing natural emitting there. Incredibly quiet. Long baseline interferometry can be used to cancel out much of the random noise from individual telescopes, terrestrial sources, etc, as can long timebase interferometry. So you're essentially taking a lot of radio-frequency photos that are, themselves, taken with a very long exposure time. Stuff in common accumulates, stuff that's different cancels out.
A sufficiently slow, pulse-modulated, message at that frequency will be extremely obvious above the noise, even if it's well below noise level any given instant. You're relying on the fact that noise is random, so that the average can be set to zero. The objective is to guarantee that the signal, after sensitivity, loss of strength and less-than-ideal capture time, strictly exceeds zero at the desired distance.
Once the law of big numbers kicks in, noise is not an issue. The average of any number of zeroes is zero. What matters is signal. If the pulse, transmitted for a second, would be 3,600 times too weak, transmitting for an hour would mean that someone capturing for an hour would detect the pulse.
Interferometry means you can also use constructive interference. Even Linux supports nanosecond accuracy and data from nanosecond-accurate PPS sources, and there are atomic clocks now that are millions of times more accurate than the official definition of the second. With that kind of gear, getting the phase such that the waves constructively interfere wherever we want is not going to be difficult. We know the phase difference already, because powerful natural radio sources must be visible from all telescopes and that same accuracy tells us how out of phase they are relative to said source.
Is that enough to go 100 LY, though? Even if both planets were ringed with telescopes, you're limited to less than the shortest year of the two per pulse and one pulse is not enough to say hello. To be unambiguous, you need a prime number of prime numbers signalled by pulses. Preferably pulses short enough that someone will notice there are some to notice.
Probably not 100. 50 would quadruple the chances of detection by any life but would butcher the chances of there being life to detect it. I don't think you can go below 25, just not enough candidate worlds, and the probability of detection only quadruples again.
A pulse of an hour duration is probably acceptable, short enough for someone to detect something strange but long enough to have enough power to stand a chance of, again, someone detecting something strange. After that, it's just a case of proper summation.
Signal power, itself, is the least important part as it falls off with the square of the distance. The challenge is to make it irrelevant, just as you make each emitter very low power in a gamma knife but very powerful at the point of interest.
Even so, you need enough bits for the sum to matter. SKA might not quite be up to the task.
Ok, it's probably not possible to transmit yet. Receive, yes, but it might take another 50 years for transmission to a reasonable number of stars to be possible.