Direct-to Handset Satellites: Are They Doing it Wrong?

Direct-to-handset (DTH) satellite service is better off using millimeter wave spectrum than mobile spectrum. The three models prevailing today – spectrum infill, spectrum slice, and over-the-top – have inherent challenges that limit market access (see my previous post). In this post, I share a perspective on how to improve the commercial viability of DTH constellations using mmWave spectrum.

Shortcomings of current models

For DTH constellations to be commercially successful, they need to solve difficult technical challenges while removing barriers to market access that limit the revenue potential. Some of the inherent shortcomings in the current approaches are:

• Spectrum infill: The satellite operator is at the whims of MNOs who, in part, see satellite operators as competitors. Moreover, this approach leads to a complex and costly technical solution. Therefore, this approach is very risky.
• Spectrum slice: Dedicating a slice of spectrum for DTH services implies an opportunity cost: spectrum could be used for more profitable services. This steers the MNOs to allocate small bandwidth leading to low throughput and/or capacity service. The satellite operator is still at the whim of MNOs. Also, both the spectrum infill and slice models are national or regional models: they are limited to agreements with MNOs for specific areas; therefore, they are not global in nature.
• Over-the-top spectrum: This model is proprietary to a single handset manufacturer – Apple. It’s a sound approach for Apple to increase the appeal of new iPhones, but that does not help the wider market. (See here for Apple & Globalstar’s approach.)

mmWave for DTH services

mmWave spectrum for DTH satellite services avoids many of the challenges inherent in the existing models. Here, I am using the term ‘mmWave’ in a liberal form – spectrum in 10-28 GHz; especially in the high-X / low-Ku band (e.g. 10-13 GHz). I rationalize here different reasons for this choice. Of course, there are challenges to using mmWave, but I think these challenges will be easier to overcome in the medium and long-term. Both from technical and commercial aspects, mmWave spectrum is a better choice for DTH on the whole.

Mobile spectrum is inferior for DTH performance

The reason for using mobile spectrum – primarily sub 2 GHz spectrum – is that it provides instant access to billions of mobile subscribers. This is very attractive and logical argument for DTH service providers: What point is there in using spectrum that no handset supports? I will come back to this point later!

The challenge in using mobile spectrum for DTH service is two folds: 1. Limited bandwidth; and 2. Challenging technical requirements, especially for the design of the satellite antenna system.

Mobile spectrum largely comes in 5 MHz or 10 MHz allocations that are simply too narrow to provide meaningful ‘broadband’ throughput and/or user capacity from space.

The planned DTH constellations offer peak throughput of up to 30 Mbps in ideal situations (this is what AST advertises; SpaceX advertises peak downlink of 18.3 Mbps in 5 MHz channel according to their latest FCC filings). However, users would experience effective throughput in the range of tens of kbps. This is sufficient for emergency text messaging and voice calling, but that in itself constrains the addressable market. The limited performance raises commercial risk: Is there sufficient revenue to compensate for the billions of Dollars needed to build DTH constellations?

Another challenge in using mobile spectrum is the onerous requirements placed on the design of the system. Mobile handsets are designed to communicate with cell towers at a range not exceeding a few kilometres. Handset have low-gain omnidirectional antennas for this purpose (gain between -6 – 0 dBi). The handset power is also restricted to 23 dBm (effective radiated power, class 3 device). The satellite antennas need to compensate for the weakness of the mobile handset to overcome over 150 dB in path loss to low-earth orbit. Additionally, the satellite antenna needs to mitigate against co-channel interference and protect licensees in adjacent spectrum license areas.

This pushed DTH satellite operators to design large and sophisticated antennas. The largest to date is AST’s 64.4 m² phased array antenna (8 metres a side) which is currently under test (download our Insight Note on AST). Elon Musk mentioned SpaceX Gen 2 satellites would be equipped with a 25 m² (see here). The size of DTH antennas raises questions on satellite flight and structural dynamics among other issues.

mmWave brings many advantages to DTH services

mmWave offers 100 MHz channel bandwidth with the possibility to combine multiple such channels. Therefore, it is possible to achieve much higher data rate and capacity than in mobile spectrum. Moreover, the short wavelength leads to more manageable satellite antenna dimensions than mobile spectrum.

To illustrate, consider the table below comparing 800 MHz, 1900 MHz and 28,000 MHz frequencies. Even with higher path loss, the power at the satellite is only ~3 dB worse than using a PCS frequency. This is because mobile devices are allowed to beam-form in the direction of the base station and transmit at 20 dB higher EiRP than in sub 6 GHz frequencies.

Frequency (MHz)	800	1,900	28,000
λ/2 (cm)	18.75	7.9	0.54
Satellite altitude (km)	2,000	2,000	2,000
Mobile handset EiRP1 (dBm)	23	23	43
Free space path loss (dB)	157	164	187
Power at satellite (pre-antenna) (dBm)	-134	-141	-144
1 EiRP as defined for Power Class 3 device in FR1 and FR2

If we add the gain of the satellite antenna, the system gain for mmWave will be far superior. It is quite feasible to get A mmWave antenna of the same surface area (e.g. 8×8 m) as that of sub 2 GHz spectrum will offer over an additional 20 dB of gain. Alternatively, one could reduce the size of the mmWave antenna to avoid the challenges experienced with having a large antenna on the satellite. In practice, one would opt for some compromise between gain and size: this is more readily possible with mmWave than <2 GHz spectrum.

Another advantage is that mmWave spectrum is already available for satellite applications as primary use basis. While it still requires approvals from national regulators to operate, it would be an over-the-top model in many markets but without a proprietary constraint to a specific handset vendor. This has the potential to make the service available to the largest proportion of the total addressable market, thus reducing commercial risk.

Challenges in mmWave DTH

I return to the point I raised earlier that mmWave is not available on most phones in use today. This would negatively impact DTH service adoption, especially in markets where 5G deployments are lagging. However, DTH is not a service to connect the unconnected. Simply, the economics of the service cannot scale to meet the constraints of that business case. DTH service is a premium or a differentiating service, at least for the foreseeable future, just like terrestrial 5G mmWave service is today.

From this perspective, I think it would be less risky for the overall business case to get people to buy a new phone if they need to use DTH services than going with one of the current models. As time progresses more and more phones are integrating mmWave modules. The user device will not be a major barrier to adoption in a relatively short time.

As a side note, I believe the high X and low Ku bands such as 10-12 GHz would be ideal for this use case! Atmospheric absorption losses are lower than in the Ka band. It is interesting that 3GPP specifies FR1 to end at 7125 MHz and FR2 to start at 24.25 GHz. I think there’s a good case to include the X and Ku bands into the list of 3GPP LTE and 5G official bands. This will help silicon vendors develop modules to incorporate these bands into future handsets.

Concluding thoughts

First, the propagation characteristics for mmWave in terrestrial mobile communications are abysmal in comparison to any sub 6 GHz band. The cost/benefit equilibrium is still on the side of high cost for the offered benefit. It is no wonder that most service providers have no appetite to deploy mmWave. Korea is a case in point: the regulator stripped the mmWave licenses from the MNOs. Verizon is an exception, but I believe that is due to their market positioning as a premium service provider.

Second, the mobile industry went for a spectrum land-grab at the WRC15, but that mostly failed with the ITU reaffirming the important nature of mmWave frequencies for satellite communications. That was the right call and could develop the basis for the two largely antagonistic industries – satellites and mobile – to collaborate. I am of the opinion that for DTH to be commercially viable, the DTH operator should not be a terrestrial MNO.