Can SpaceX’s Starlink provide high bandwidth internet to Mars?
One essential need for any possible long term and scaled mission on Mars is keeping an steady, reliable and high bandwidth communication between the two planets.
It has been just over a century since the first transatlantic radio message was transmitted from Cornwall and received at St. John’s, Newfoundland [1]. Today for interplanetary communication we do not have much in place.
What NASA has
From NASA website “ The NASA Deep Space Network (DSN) is an international network of antennas that provide the communication links between the scientists and engineers on Earth to the missions in space and on Mars.” [2]
There is only one DSN network for all space communications. The data rate to the Mars Reconnaissance Orbiter can be as high as 2 Mbps and for the Odyssey orbiter it is 128 or 256 Kbps if no other mission needs the channel [2]. This is the highest communication speed between the two planets; and if we are planning to colonize Mars this would simply not be enough.
The DNS consists of three big transmission antennas located on Earth so that at each angle there is one antenna facing Mars. Each of these antennas are about 30 meters with high transmit power and high receive gains. Even though it might be easier to build more antennas on Earth to enhance the link, using many, many smaller antennas is more beneficial as we see below.
What SpaceX could do
SpaceX is planning to deploy lots of tiny satellites around Earth to provide high speed internet around the globe. The service is called Starlink.
Starlink satellites are designed so that they can provide high speed internet to any location on earth. The nearest satellites will be approximately 340 Km above the surface [3], therefore transmitted signal will travel 340 Km from the satellite before reaching the receiver antenna on Earth.
Let’s look at what will happen to the same signal (transmitted using the same power level) if we had a receiver on Mars, a location much much further away.
In the worst case scenario Mars and Earth are far away from one another and at a 2.5 astronomical units distance (about 370000000 Km).This means if the receiver on Earth gets a power of P, on Mars it will get:

While this means a loss of 120 dB, the good news is that probably there will be lots and lots of these satellites in orbit and each could chip in one of these P powers. Estimates are that up to 42000 satellites could be launched into orbit [3]. If at any given time half of them are on the side of Earth facing Mars, 21000 such satellites could transmit the same signal, simultaneously. The power gain can actually be much more than 21000, up to 21000 * 21000!
Keyword: Constructive interference
When two waves are combined together in a constructive manner (meaning they are sent so that at the receiver the peaks arrive at the same time) the amplitude of the signal is doubled. For 21000 waves constructively combined that would make the amplitude 21000 more. But the power of a cosine wave corresponds with the amplitude to power of 2. So if A cos (wt) gives us power P; 2 * A cos (wt) will give us power 4P. This means the 21000 constructive combinations gain the imaginary Mars receiver a power increase of:

Combining the two equations above, the imaginary receiver on Mars could get a power that is:

The above figure would mean a loss of power equal to -33 dB.
2 coverage bars on Mars instead of 4!
Just to get this into perspective, when your phone’s receive signal drops from -90 dBm to -110 dBm (a change of -30 dB), you will see 2 bars instead of 4 bars [4]. Same way, a typical starlink receiver getting excellent receive coverage on Earth will get a fair receive coverage if put on Mars when we have 21000 constructive interferences. The price is paid by 210000 satellites that are cooperating together to send the signal instead of only one.
But, it is not that easy..
We made one little assumption that all 21000 satellites can transmit their signal so that for the specific receiver on Mars all of the signals are constructively combined. Eq. 2 is only achievable under this assumption; otherwise some signals will be deconstructing and some constructing and we will end up with 21000 time P, instead of 21000 * 21000 time P.
This is a bold engineering challenge! For a typical 3 GHz signal, the wavelength is 10 cm. Therefore all those satellites should know their exact distance to the target receiver up to a precision of 1 cm! Since each satellite is rotating around the Earth itself, this can be a real challenge to achieve. Using some mechanism for reference signals that all of these satellites could tune into could solve the issue.
Also, for an average user on Mars this means high speed download, for upload we would still need to use NASA’s DSN, until starlink satellites are put around Mars orbit too
Some references
[1]: https://en.wikipedia.org/wiki/List_of_Marconi_wireless_stations
[2]: https://mars.nasa.gov/msl/mission/communications
[3]: https://en.wikipedia.org/wiki/Starlink_(satellite_constellation)
[4]: https://www.signalbooster.com/blogs/news/how-to-measure-signal-strength-in-decibels-on-your-cell-phone
