This post was featured by New Space Enterprise - you can read the article on their site here.
They also happen to run one of my favorite Instagram accounts (@newspaceenterprise) 

_______________________________________________________________

The whole point of telecommunications is to rapidly send signals across large distances, and for many applications this is accomplished by terrestrial methods, be it underground cables or cell towers. But ground-based infrastructure can be expensive and unreliable, especially in remote areas; a well-placed satellite can provide scalable transmissions over vast areas

A ground station transmits the signal up to the satellite (uplink), which then amplifies it, alters the frequency, and sends it back down to a different station on Earth (downlink). But why does a satellite need to change the signal frequency? By the time the uplink signal reaches the satellite's receiving antenna, it has become very weak; however, the satellite's transmitting antenna is simultaneously blasting the downlink signal at a unique frequency that identifies it from other satellites in orbit. If the uplink and downlink frequencies were the same, the transmitted signal would wash out the received signal, and the satellite wouldn't be able to communicate with the ground station. So it's the job of the satellite transponder to perform this frequency shift between the receiving and transmitting antenna

So what's the right frequency to use? The answer, as is often the case in engineering, is it depends on some tradeoffs. The above diagram shows many frequency ranges that each have their specific uses, but the two most relevant to satellite internet are the Ku and Ka bands. Right now, there's a hot debate as to which is better, analogous to the AC/DC current wars of the late 1800s or Blu-Ray / HD-DVD in the 2000s. You'll find proponents on both ends of the spectrum (pun intended!) - here's an article from Gogo in-flight wifi touting the reliability of the more established Ku-band, while here's an article from Honeywell exalting the greater bandwidth of its higher-frequency Ka-band Jetwave!

Finally, the question remains of where to place the satellite. One common technique is to place satellites in geostationary orbit, a unique orbit 22,236 miles above the Earth where a satellite's orbital speed happens to match Earth's rotational speed, so the satellite remains permanently fixed over the same point on Earth. Just 3 satellites at that altitude can cover the whole globe! Alternatively, you could place a ton of smaller, cheaper satellites in low Earth orbit where they form an interconnected web as they quickly zip around the Earth. Given recent improvements in computer chips and electrical engineering that have dramatically reduced satellite size and cost, combined with the desire to cut latency, this appears to be the new trend

Here's the basic idea: we may call the 21st century the dawn of the digital age, but a staggering 4 billion people still don't have regular access to the Internet, particularly in remote and underdeveloped regions. So to capitalize on this massive business and humanitarian opportunity, four companies to date that have announced plans to construct consumer broadband satellite Internet constellations in low Earth orbit: SpaceX's Starlink, OneWeb, Telesat LEO, and Amazon's Project Kuiper, all currently at varying degrees of progress

Like any good investment banker, I make all the diagrams I need for Astronomical Returns in Excel and PowerPoint!

The potential revenue from satellite Internet is expected to be gargantuan; SpaceX alone projects \$30 billion from Starlink by 2025, which it crucially needs if it wants to send people on Mars (for reference, its current core business of launch services is projected to only reach $5 billion in revenue by 2025). But it's no secret that there are massive hurdles to monetizing satellite Internet:

1. Satellite Internet is inherently unsuitable for applications requiring low latency (like gaming or high frequency trading). Since the signal has to travel thousands of miles into space and back, it's outpaced by terrestrial connections. Granted, the incumbent satellite Internet providers, Hughes Network Systems and ViaSat, placed their satellites in geostationary orbit while the four new entrants above are building constellations in LEO reduce latency. But there's no topping the speeds of a direct fiber optic cable; not even the great Elon Musk can make light travel faster
2. Even once the satellites are in orbit, these companies will need to make massive investments in ground stations and convince customers to buy expensive antennae
3. Finally, how many constellations are needed before market saturation sets in? Given this will likely be a commoditized product, we may see a race to the bottom that will drive inefficient operators out of the market

Two things are going to be needed to win this battle - first mover advantage (SpaceX and OneWeb are ahead here) and a ton of cash (Amazon's war chest absolutely smokes the others). Whether this all comes to fruition is yet to be seen, but one thing's for sure: we've come a loooong way since Sputnik

To learn more, might I suggest this book, Satellite Basics for Everyone. Recommended to me by my good friend from middle school and high school, who just started working at SpaceX's Starlink division in Seattle. Tell Elon to hire me!