Alternative communications network design - Thoughts?

[RealKerbal3x]

I was browsing the interwebs a while ago and found this interesting communications network design that works around the annoying orbital drift bug (is it a bug? I'm not sure) by utilising four comsats in eccentric orbits instead of the normal synchronous orbits. I couldn't find where I saw the original diagram, so I knocked this up in five minutes on Paint - I hope it gets the point across.

(Note that this is looking at the planet from directly above the north or south poles).

There's a few things I was wondering about this setup:

1. What is the probability that you are in contact with the space centre?

2. How likely is it that you are in contact if the orbits are more eccentric or less eccentric?

3. What are the advantages and disadvantages of this setup?

[Rocket In My Pocket]

Considering the planets mostly reside on similar planes, you could prolly even just do one very eccentric satellite going straight "up and down" relative to the system's plane.

Edited December 19, 2018 by Rocket In My Pocket

[RealKerbal3x]

2 minutes ago, Rocket In My Pocket said:

Considering the planets mostly reside on similar planes, you could prolly just do one very eccentric satellite going straight "up and down" relative to the system's plane.

That would probably work, but this would still most likely be a useful alternative to a synchronous orbit constellation without having to muck about with precise orbital periods.

[Rocket In My Pocket]

3 minutes ago, RealKerbal3x said:

That would probably work, but this would still most likely be a useful alternative to a synchronous orbit constellation without having to muck about with precise orbital periods.

Absolutely, I think it's a pretty good idea!

I was just positing on the possibility of a super minimal version of it. (Since I'm kind of lazy when it comes to satellites lol.)

Like, the highest eccentric orbit you can get relative to the system plane so it spends most of it's time "above" the planets looking down at them, kind of the same way a high vantage point gives you clear LOS in a shooter.

Edited December 19, 2018 by Rocket In My Pocket

[RealKerbal3x]

1 minute ago, Rocket In My Pocket said:

Absolutely, I think it's a pretty good idea!

I was just positing on the possibility of a super minimal version of it. (Since I'm kind of lazy when it comes to satellites lol.)

Like, the highest eccentric orbit you can get relative to the system plane so it spends most of it's time "above" the planets looking down at them, kind of the same way a high vantage point gives you clear LOS in a shooter.

Yep, although if the satellite spent a lot of time high above the system it might need a very powerful relay dish to stay in contact.

When I get enough time to spare I'll set up a constellation like this in KSP and see how well it works.

[Rocket In My Pocket]

6 minutes ago, RealKerbal3x said:

Yep, although if the satellite spent a lot of time high above the system it might need a very powerful relay dish to stay in contact.

When I get enough time to spare I'll set up a constellation like this in KSP and see how well it works.

Yes, definitely, if you are only going to have one; I'm assuming it would be really powerful.

I'd stack dozens of relays on it if needed, if you factor in the cost savings of not building several smaller satellites I'd wager you could justify quite a budget for the one "mega sat"

Still I think you'd end up with more frequent coverage gaps/loss than with the original 4 sat concept.

Edited December 19, 2018 by Rocket In My Pocket

[RealKerbal3x]

1 minute ago, Rocket In My Pocket said:

Yes, definitely, if you are only going to have one; I'm assuming it would be really powerful.

I'd stack dozens of relays on it if needed, if you factor in the cost savings of not building several smaller satellites I'd wager you could justify quite a budget for the one "mega sat"

Yep. You could essentially use any number of satellites you wanted with this design, but my example just had four.

This would be able to be scaled up from moon system size to solar system size much more easily than a standard synchronous orbit relay network too.

However we need to work out if this can keep us in communication as reliably as the aforementioned network can.

[Lapis]

this seems interesting, I will launch as soon as I can.

[Snark]

23 hours ago, RealKerbal3x said:

I was browsing the interwebs a while ago and found this interesting communications network design that works around the annoying orbital drift bug (is it a bug? I'm not sure) by utilising four comsats in eccentric orbits instead of the normal synchronous orbits.

Yep, I've used a somewhat-simplified version of that myself for a long time.  I'll often set up a "network" that consists of only two satellites, in eccentric orbits that have their apoapses 180 degrees apart-- i.e. take your diagram and subtract two satellites, and it looks kinda like that.  However, with a few differences:

• Orbits generally a lot more eccentric than you've shown here-- specifically, the apogee is many planetary diameters high.
• The two orbits are inclined to each other 90 degrees-- that is, their semimajor axes are parallel (apoapsis 180 degrees apart), but they're rotated at different angles around that axis.  Like adjacent links in a chain.
• The plane containing the apoapses is generally tilted with respect to the planet's orbital plane around the sun (or the moon's orbital plane around the planet), rather than being in the equatorial plane.

Reasons I like this approach:

1. Parsimonious.  Only needs two satellites to set up.
2. Quick and easy to set up.  No ultra-precise fiddling to try to get anything synchronous.  I don't make any particular effort to give them the same period as each other, and it doesn't matter if they're not precisely aligned; nothing in the design depends on any number anywhere having a particular, exact value.
3. Very effective.  Each satellite spends the large majority of its time at very high altitude where it can see nearly 50% of the globe, and the two satellites' coverage areas are mostly non-overlapping.  So at any given moment, there's a 90%+ chance that any given spot on the planet's surface will have LOS on a satellite.  And if by any chance it doesn't, just wait a little bit and then it will.

In cases where I only have one satellite to work with, I'll put it into an eccentric orbit where the apoapsis is facing directly away from where Kerbin happens to be.  That way, the satellite spends the large majority of its time "loitering" on the blind side of the planet, which also gives pretty good coverage:  i.e. no matter where I land, chances are high that I'll have LOS either to the relay or directly to Kerbin.  (Of course, this one-satellite setup has the disadvantage that it doesn't last:  i.e. the apoapsis is facing the way I want at the moment I set it up, but as the planets orbit around the sun, the orientation of apoapsis relative to planet's orbit will "drift" and will be facing the wrong way half a solar orbit later.  But it works pretty well for a long period of weeks or months, and is "good enough" if I time it so that the relay arrives at the planet around the same time as my lander.)

The guiding principle, for me, is "bang for the buck", i.e. looking for the sweet spot of "maximum comms benefit" versus "minimum hassle required".  It's really really easy to set up a simple comms network that provides coverage 90% of the time.  Trying to make iron-clad 100% guaranteed connectivity at all times is a whole lot harder-- much more effort expended.  And for what?  For a "benefit" that, to me, is marginally useful at best.  Comms networks don't need to be 100%; occasional interruptions are fine.  As long as it's up most of the time (like 90%), and as long as the outages are fairly brief, it's plenty good enough and a lot less work. 

 

[RealKerbal3x]

9 minutes ago, Snark said:

-enormous wall of snarky text snipped-

 

Some great insights here!

So, to condense your detailed analysis down to a couple of bullet points:

--Satellite orbits are generally more eccentric than in my diagram.

--Four satellites is unnecessary for this - you can use two without many issues.

--One satellite can even suffice, but only for a limited period of time.

Edited December 20, 2018 by RealKerbal3x

[Snark]

45 minutes ago, RealKerbal3x said:

So, to condense your detailed analysis down to a couple of bullet points:

--Satellite orbits are generally more eccentric than in my diagram.

--Four satellites is unnecessary for this - you can use two without many issues.

--One satellite can even suffice, but only for a limited period of time.

Yep, that pretty much sums it up. 

Going for more-eccentric orbits-- specifically, much higher apoapses-- provides a couple of benefits.  First, the higher the satellite, the bigger the percentage of the globe it can cover (coverage approaches 50% as altitude approaches infinity).  Second, raising the apoapsis greatly increases the "loiter time" up there, so that the satellite spends a much higher percentage of the time up high where it's wanted rather than "down time" when it's rounding the planet at periapsis.  Or, to summarize, higher apoapsis = better coverage.

Another strategy with the one-satellite approach is to put its apoapsis not over the "blind side facing away from Kerbin", but rather over the daytime side, regardless of where Kerbin might happen to be.  Depending on mission design, that can be a useful strategy.  For one thing... I almost never do anything when I'm landed on a planet at night.  I don't want to land in the dark.  I don't want to EVA and go walking around in the dark.  Unless I'm wholly RTG-powered, I don't have any electricity for operations in the dark.  So I'll always do things on the daytime side, by preference... so it would be best to give relay coverage to that side of the planet.  (I don't mind losing comms in the dark, because I usually shut down and do nothing until sunrise anyway, so I don't need comms in the dark.)

Another option:  If I don't need to provide global coverage of the surface, but instead am primarily interested in providing coverage to a specific place (e.g. the site of a surface base), then that can open other options.  For example, suppose my surface base isn't right spang on the equator, but instead is at a middle latitude-- say, 45 degrees north.  In that case, I can put a lone relay satellite in a highly eccentric orbit that has its apoapsis over the north pole.  That satellite will therefore be in my surface base's LOS the large majority of the time... and that is a permanent solution that lasts for as long as the base is operational.  (It's similar in principle to using a tundra orbit, but without the orbit being synchronous.)

 

45 minutes ago, RealKerbal3x said:

One satellite can even suffice, but only for a limited period of time.

^ Note that if a satellite has some dV remaining, it can adjust the orbit to "refresh" it when it drifts too far from where it's needed.  Doesn't take much dV, either.  A satellite with a really high apoapsis only needs a tiny burn to circularize at apoapsis, and then only needs another tiny burn to lower the periapsis again.  So the "limited period of time" satellite is "reusable", for as long as its fuel holds out.  Doing a small "orbit reorientation" pair of burns every few months is something that's easy to keep up for a pretty long time.

[Cavscout74]

For Kerbin, I use this:

  1)  A trio of small relays at 750km around Kerbin to fill in coverage gaps for LKO.

  2) Another trio of medium relays just outside of Minmus orbit, around 60 Mm.  Lately I've been using the 50G Quetzel antenna from Probes Plus mod for this role since it becomes available sooner than the RA-15 and has much better signal.

  3) Finally a very large (usually JX2 antenna mod) relay in an elliptical polar orbit, usually going to 60 Mm or so

For planets/moons with a lot of activity (primarily Duna & Laythe) I do this::

  1) A pair of small (RA-2) relays in a semi-synchronous orbits 180 degrees apart

  2) For moons, a medium sized (RA-15 or 50G Quetzel antenna) relay in an elliptical polar orbit to link the small relays

  3) A single large relay (either around Duna or for Joolian moons, around Jool) in a highly elliptical polar orbit - typically going out to near the edge of the SoI.  If I'm feeling especially generous, I might add a second one, with the Ap over the opposite pole.  Usually either an RA-100 or a JX2 mod antenna

Planets without a lot of activity (Moho, Eve, Eeloo and that other one I forget about), usually get just a large relay in elliptical polar orbit.  Any surface probes, I aim for the northern hemisphere so they can usually see the relay.

I also use HG-5 antennas or other small relay antennas on my most of my satellites, which supplements the dedicated relays.  Since every body I'm putting a base on will have a polar resource scanner at a minimum, that helps fill in small gaps.   I also stick relays on my stations, again to supplement the dedicated network.

Finally, I typically have 3 asteroid spotter satellites near Kerbin orbit to cover all the spot asteroid contracts (one each in 10.0 - 10.5 Gm, 10.5Gm to 11.0 Gm & 11.0 to 11.5 Gm), and I stick medium sized relay antennas on these as well.  They can provide a relay around the sun if nothing else is available.

I used to put a lot more (more like the tutorial on KSP wiki), but I just don't see that's its needed, even on hard career. 

[PirateCat]

This is how I always would do my comsats after I installed RemoteTech. Plus I would have two additional ones going north and south (or above and below Kerbin) that also had the long distance dishes. Each of my equitorial sats would have an omni plus 4 dishes, two that pointed at the sats to either side, one pointed at either the north or south comsat, and one pointed at current vessel. There would be occasional communications blackouts, but they never lasted very long.

You can see the setup in this old screenshot I had lying around.

[eddiew]

Allow me to quote myself from back in the 0.25 days when I was running with RemoteTech...

Bonus with modern KSP, you don't have a single point of contact with the ground - it does however apply really well to probe rovers on other planets if you really really want constant comms.

The actual probability of reaching "any" relay is proportional to how high their APs are. The more eccentric their orbit, the longer they hang around at high altitude and the faster they sweep around their PE.

Edited January 10, 2019 by eddiew

[stibbons]

For a long time my communications network design has been "take a lot of satellite contracts and slap relays on the side of all of them". Works well, and I get paid for it.

[FleshJeb]

1 hour ago, eddiew said:

Allow me to quote myself from back in the 0.25 days when I was running with RemoteTech...

Coincidentally, that's the last time I played with a comms system.

This is a good time for me to point out that the majority of the moons are tidally locked, and Duna and Ike are tidally locked together. It's a good way to eliminate some drift from your network.

[RealKerbal3x]

So I was wondering, since several of you have set up a similar constellation in the past: once you've got your satellites into LKO, at what intervals do you deploy them? Since my kerbostationary relay network has recently and inexplicably fallen out of alignment, I've been thinking of setting this sort of network up.

[Xavven]

I do the same thing as the OP's picture except with 3 relays instead of 4, and make the orbits as eccentric as possible without crossing any moons' orbit lines. You get diminishing returns with each satellite you add, and so 3 is the magic middle-ground for cost/benefit for me, but everyone has their own preference.

I like having them in the equatorial plane of the planet or moon because my spacecraft usually arrive in that plane to begin with. One weakness of this setup is that you end up with two blind spots, one at the north pole and one at the south pole. Furthermore, at high latitudes you risk losing line of sight behind a hill or mountain because the satellite will be close to the horizon. To compensate, you can either add a satellite with a high polar apoapsis, or alternatively for very little dV cost you can change the inclination (not apoapsis) of each of your 3 relays to 90 degrees so that they alternate giving slightly better north pole and south pole coverage, albeit spotty coverage. This isn't usually a problem for me because polar visits are rare in my playthroughs.

I think Snark's 2-relay chainlink setup is more efficient from a cost/coverage perspective. The tradeoff is slightly more link downtime. You might find you prefer 2, 3, 4, 5, or 6 satellites depending on your appetite for setup time + launch costs vs. coverage + uptime. It's a balancing act.

As for intervals, do you mean synchronous orbits (i.e. orbital period vs. planet rotation)? I don't pay that any regard when setting up my relays.

Edited January 15, 2019 by Xavven

[septemberWaves]

In my current playthrough I am planning on positioning a satellite at the Mun's L5 point, and doing the same for Minmus (I am aware that these points do not actually exist in terms of gravity in KSP, but I will use RCS and fine control, along with precise data from MechJeb, to adjust the orbits to be very close to ideal). This should be sufficient to contact orbiters on the far side of both moons.

Later on I am unsure exactly what I will do for communication networks around other planets; since I am using Kerbalism there will need to be at least one interplanetary relay with a high-gain antenna, as well as several local relays. I will probably use satellites at the L4 and L5 points around local moons (in cases where moons exist), but for the main satellite network I am not sure; I feel like three satellites in elliptical equatorial orbits is probably a good plan for most planets (though for gas giants these orbits will likely have to be polar with the apses at the ascending/descending nodes, to avoid encounters with moons), with another in a Molniya-style orbit if necessary (for things like polar missions). Being able to see precise information with Mechjeb will definitely help with putting together satellite networks, and I will be able to do calculations to figure out timings and optimal apses (I might actually make an excel spreadsheet for the sake of being able to do calculations quickly).

[RealKerbal3x]

On 1/15/2019 at 6:05 AM, Xavven said:

As for intervals, do you mean synchronous orbits (i.e. orbital period vs. planet rotation)? I don't pay that any regard when setting up my relays.

Sorry, didn't notice this initially.

By interval I mean at what time after deploying satellite 1 into its orbit do you deploy satellite 2? When sat 1 is at apoapsis or some other time?

[Xavven]

12 hours ago, RealKerbal3x said:

Sorry, didn't notice this initially.

By interval I mean at what time after deploying satellite 1 into its orbit do you deploy satellite 2? When sat 1 is at apoapsis or some other time?

Oh, well in KSP it's very hard to get all of your satellites to have exactly the same orbital period. You might get it so close that they are synchronized for a while, maybe even a good while, but time warp a mission to Jool and back and you'll find your satellites are no longer in sync. For this reason I don't bother trying to get them to have the same orbital period, and therefore it doesn't matter when I launch the next satellite, because it's going to get messed up anyway.

 

[Dafni]

After finding the mod StationKeeping I went for circular orbits with evenly spaced sats again, it satisfies my OCD the best.

[Bill Phil]

Huh.

I never used the comm parts of the game before. Didn’t interest me.

Maybe I will after my current career (which is from 1.4.x but I am finishing the tree).