4 ms·

I wonder if it's possible (and sensible) to have a bunch of cyclers after each other, spaced eg a month apart. The distance between them would be much shorter t

I wonder if it's possible (and sensible) to have a bunch of cyclers after each other, spaced eg a month apart. The distance between them would be much shorter than Earth -Mars and thus allow much faster data transfer. With these cyclers as relays, there's always one close to Mars, another close to Earth. Sort of an interplanetary Starlink.

jvanderbot 3y ago
We did look at that. In fact it's quite possible to have a "train" like this, but: * Continuous relay is power-hungry, and the distances are still huge unless you have a _lot_ of relays on _very_ different heliocentric orbits. The scale of the inner solar system is _enormous_ compared to LEO. High power on lots of satellites means huge cost on a per-unit basis. LEO is 100km away vs 1/5 distance to Mars is still 80 million km or so. Transmission power for a given rate is d^2 (or d^3 in some cases for lasers). Enormous power requirements. * On earth, you can "space out" an orbit so that you basically always have either another satellite on the same orbit, or a "nearby" orbit that has one coming up. Look at GPS tracks, for example. In inner solar system, everything orbits the _sun_, meaning nearby orbits might be nowhere near earth, and even a train on the same orbit is going to miss Earth 99.9% of the time by a million km or more. * An interplanetary relay has been studied. Instead of cyclers which pass close, you basically put a "train" of spacecraft on a heliocentric orbit between earth and mars. Again, this was a lot of spacecraft required to get relays such that the cost paid off relative to just building bigger transmitters / listeners on earth or having more orbiters around Mars. The sweet spot of cyclers is low thrust requirements, comparatively low system requirements (big burst of data once in a while with lots of charging between), which combined to make sats small enough to justify a dozen or so on a single launch of this purpose. I _do_ think we'll have solar-system-starlink. Probably at Mars, definitely at Moon, and likely some lagrange point relays with a cycler or two funded by an eccentric billionare in 50 years. EDIT: I misunderstood, below is old comment / reply That is precisely the subject of the paper I linked. Except you can't confuse visit frequency with visit _latency_. We can have many visits per year but all the data coming down is usually over a year old. For Grand Science on a planetary scale this is ok. And for lower latency data requirements (of which there are not really many cases outside human impatience) you can use regular downlink but only .01% of the throughput. But it gets even more interesting if the data "ferry" has enough compute to work while moving. You could run models on the ferry to answer questions, then every so often (when close to earth) download all the data to verify answers. What we derive and optimize is the orbits to achieve such a "high frequency" cadence by formulating a set cover instance. The sets are the visits for each cycler and we want to "cover" a given time span with "enough" visits. And the cyclers themselves are a large parameter sweep conducted somewhat open loop.
- nwiswell 3y ago
  > That is precisely the subject of the paper I linked. GP was actually proposing inter-cycler links to create a kind of cycler network with Earth on one end and Mars on the other. That is, your latency comes from the speed of light, rather than the inter-cycler latency.
  - ragebol 3y ago
    Exactly. Have some sort of chain of cyclers that each cycle independently, launched/spaced (hehe) eg. a month apart. As they cycle, only one of them will be closest to Mars, that gets the uplink from Mars, then relay the data to the cycler behind it, that relays to the one behind etc etc until we reach the cycler that is closest to Earth, in mere minutes. Can have laser links between them even maybe. Lets call it PlanetLink :-)
  - jvanderbot 3y ago
    Good catch, thanks. I updated my reply. it's a good idea that has some backing in literature and I do think we'll see it soon enough.

jvanderbot 3y ago

We did look at that. In fact it's quite possible to have a "train" like this, but: * Continuous relay is power-hungry, and the distances are still huge unless you have a _lot_ of relays on _very_ different heliocentric orbits. The scale of the inner solar system is _enormous_ compared to LEO. High power on lots of satellites means huge cost on a per-unit basis. LEO is 100km away vs 1/5 distance to Mars is still 80 million km or so. Transmission power for a given rate is d^2 (or d^3 in some cases for lasers). Enormous power requirements. * On earth, you can "space out" an orbit so that you basically always have either another satellite on the same orbit, or a "nearby" orbit that has one coming up. Look at GPS tracks, for example. In inner solar system, everything orbits the _sun_, meaning nearby orbits might be nowhere near earth, and even a train on the same orbit is going to miss Earth 99.9% of the time by a million km or more. * An interplanetary relay has been studied. Instead of cyclers which pass close, you basically put a "train" of spacecraft on a heliocentric orbit between earth and mars. Again, this was a lot of spacecraft required to get relays such that the cost paid off relative to just building bigger transmitters / listeners on earth or having more orbiters around Mars. The sweet spot of cyclers is low thrust requirements, comparatively low system requirements (big burst of data once in a while with lots of charging between), which combined to make sats small enough to justify a dozen or so on a single launch of this purpose. I _do_ think we'll have solar-system-starlink. Probably at Mars, definitely at Moon, and likely some lagrange point relays with a cycler or two funded by an eccentric billionare in 50 years. EDIT: I misunderstood, below is old comment / reply That is precisely the subject of the paper I linked. Except you can't confuse visit frequency with visit _latency_. We can have many visits per year but all the data coming down is usually over a year old. For Grand Science on a planetary scale this is ok. And for lower latency data requirements (of which there are not really many cases outside human impatience) you can use regular downlink but only .01% of the throughput. But it gets even more interesting if the data "ferry" has enough compute to work while moving. You could run models on the ferry to answer questions, then every so often (when close to earth) download all the data to verify answers. What we derive and optimize is the orbits to achieve such a "high frequency" cadence by formulating a set cover instance. The sets are the visits for each cycler and we want to "cover" a given time span with "enough" visits. And the cyclers themselves are a large parameter sweep conducted somewhat open loop.

nwiswell 3y ago
> That is precisely the subject of the paper I linked. GP was actually proposing inter-cycler links to create a kind of cycler network with Earth on one end and Mars on the other. That is, your latency comes from the speed of light, rather than the inter-cycler latency.
- ragebol 3y ago
  Exactly. Have some sort of chain of cyclers that each cycle independently, launched/spaced (hehe) eg. a month apart. As they cycle, only one of them will be closest to Mars, that gets the uplink from Mars, then relay the data to the cycler behind it, that relays to the one behind etc etc until we reach the cycler that is closest to Earth, in mere minutes. Can have laser links between them even maybe. Lets call it PlanetLink :-)
- jvanderbot 3y ago
  Good catch, thanks. I updated my reply. it's a good idea that has some backing in literature and I do think we'll see it soon enough.