T-Mobile + SpaceX Direct Satellite-to-Handset Service: Lots of Hype and Little Reality

This will start as a technical post, but I will relate it to practical and commercial conclusions at the end.

T-Mobile and Space-X announced a partnership to provide direct satellite-to-handset service using T-Mobile PCS frequency band. While this is new to Space-X, there are other players in the industry working on the same objective, primarily AST SpaceMobile and Lynk. [Apple is rumored to launch this type of service with Globalstar.]

The service will enable users outside T-Mobile’s terrestrial network coverage to connect to a Starlink satellite (Gen 2) to send and receive text messages and voice calls (future phase).

Here, I dive deeper into the technical details of this type of system to highlight some practical and commercial conclusions that apply to all such direct satellite-to-handset networks.

The Technical Background

I will look at the case of the uplink from the mobile handset to the satellite (uplink). This is because the uplink is typically the weaker link in LTE and 5G systems.

Mobile handsets transmit at 0.2 W (23 dBm) and feature a low gain antenna (typically 0 dBi). This makes for an effective transmitted power (EiRP) of 23 dBm.

The receiving satellite is orbiting the earth at around 550 km. Using the Friis transmission equation we calculate the free space path loss to be 153 dB for 1900 MHz frequency (PCS band).

The mobile handset power that the satellite sees is then -130 dBm (23 – 153).

To decode an LTE signal at the lowest modulation (QPSK), the signal power needs to exceed -105 dBm. This value factors several assumptions such as a low user throughput of 32 kbps in a 5 MHz channel, and a single receive antenna. In practice, the receiver sensitivity is a matter of vendor implementation.

I would also add some margins to account for potential fading and lack of polarization alignment between the transmit and receive antennas. Let’s say this is about 4dB. This brings the signal level at the receiver to -101 dBm.

As a result, the satellite antenna gain needs to be -101 – (-130) = 29 dBi.

We can now calculate the effective aperture of the antenna which gives an idea about its size. This comes to about 1.4 m². The form factor of the antenna could take on different shapes, but its aperture, or area need to be 1.4 m². For comparison, base station antennas in this band are typically 17 dBi and 1.5 m tall (I estimate the aperture of such antennas in the neighbourhood of 0.1 m²). [A side note: Elon Musk mentioned the cellular antenna to be 5-6 m on a side, or about 25 m². Several antenna would be carried by a single satellite to make for as many coverage cells.]

Next, I estimate the antenna half-power beamwidth to get an idea of the potential coverage area. This is where things get very interesting. For a 29 dBi-gain antenna, assuming perfect efficiency, the half-power beamwidth is about 7.6 degrees. This results in a cell radius of about 36 km and coverage area of about 4,100 sq. km.

The relatively small coverage area has advantages and disadvantages. On the positive side, it allows targeting the antenna to serve specific locations which prevents interference to areas covered by the terrestrial network.

On the negative side, one needs many such antennas to cover a wide area. A satellite could carry multiple antennas. But this is not sufficient to provide permanent coverage over time. This is where some innovation is needed whereby the network could multiplex different areas maximize the utilization of the satellite network while maximizing the coverage spatially and over time.

The analysis above gives a feel for the trade-offs and establishes a framework to think about the commercial implications. It is not meant to be exact since different possibilities exist.

Practical Tradeoffs

We can draw a number of observations to help us understand the performance of the service and assess its commercial impact.

• Limited coverage area. The satellite needs to support a high gain antenna to close the link budget. This in turn leads to high antenna directivity and small coverage area. As a result, covering large areas requires many antennas and satellites (which is what Elon Musk said at launch). Scalability involves a trade-off between coverage and cost of the LEO network. How to solve this problem will differentiate this type of satellite networks. One hint is that the service, at least initially, will be non-real time as it needs to multiplex different areas. In all, whether this system really eliminates dead-zones is to be proven.
• Low-bit rate service. In my calculations, I used a relatively small channel bandwidth of 5 MHz with a user bit rate of 32 kbps. This bit rate is sufficient for low data transmissions including text messaging and voice (codecs run at 9.8 kbps and lower). A smaller channel bandwidth and/or lower bit rate reduces the antenna gain and size requirement, and results in a larger service area. Similarly, a wider channel, such as 20 MHz will increase the antenna size requirements and results in smaller coverage area. Taken all together, direct satellite-to-handset is not a broadband service – and could not be one.
• Low capacity network. Elon Musk expected a cell to support between 2-4 mbps. This is about 1250-2500 users per satellite if we assume 32 kbps/user and oversubscription factor of 20. Not a bad number, but we need to think of this in context of the overall market, user behavior and network capability.
• Spectrum separation and loss. Satellites will be assigned their own slice of terrestrial spectrum. The alternative option of using the same frequency band on both satellites and terrestrial network requires complex coordination and would still result in interference zones at the edge of coverage footprint. This raises the question on whether operators would perceive sufficient value to dedicate a slice of spectrum for this service.

These observations apply to all companies pursuing direct satellite-to-handset service. They are all governed by the same laws of physics.

Commercial Implications

Direct satellite-to-handset service is primarily a low-bitrate service for messaging and voice calling. It is not a solution for browsing the web or watching YouTube! It follows that one can ask the following questions:

1. The service is best described as that for emergencies in areas where terrestrial networks do not exist. Will the business case work out for the T-Mobile and SpaceX?
2. Will other operators join the project considering they need to set aside a [small] part of their frequency spectrum?
3. The service will impact a relatively small percentage of users – those who venture outside the terrestrial coverage footprint. Nevertheless, it’s a marketing coup for T-Mobile. In this respect, will T-Mobile succeed in getting subscribers from AT&T and Verizon to compensate for the net cash outflow for including this service into popular plans? And, what will AT&T and Verizon do in response?
4. Will such a service attract government, military and public safety users? And, to what extent could it compete with or be complementary to services of similar capabilities over LEO and GEO satellites?

This points particularly highlights how satellite communications impacts the competitive dynamics among network operators.

Final Comment

Direct satellite-to-handset was the topic of a workshop we at Xona Partners organized in collaboration with i2CAT. For a transcript of the event, download it here:

If you are interested in similar analysis linking technology and commercial aspects, contact me here.