The satellite industry needs new ways to assess the risk of interference, identify which mitigations and regulations would be most effective, and work together to avoid interference. I’ll describe two of them: risk-assessment and multi-stakeholder coordination.
The satellite game is changing in at least three ways.
First, the planned non-GSO constellations will increase the number of objects in LEO and MEO (low- and medium-Earth orbit, respectively) by at least a factor of ten. For example, if just SpaceX completes its planned deployment, it will add more than 11,000 satellites to the 1,070 LEO objects recorded in the UCS database as of August 2017. OneWeb’s planned 1,280 MEO constellation is more than ten times the 97 MEO objects counted by the UCS.
Second, there’s increasing diversity of competitors in the satellite business. Billions of dollars are being invested in NewSpace start-ups. There are new launch companies like SpaceX, Blue Origin and Rocket Lab; new entrants like OneWeb, SpaceX, Boeing and Telesat planning to compete against existing non-GSO services from Iridium and SES-O3b; a burgeoning number of companies offering earth observation (Planet, Spire, Theia, etc.) as well as ancillary in-orbit communications services (LeoSat, Audacy); and last but not least, GSO operators are deploying a new generation of high-throughput satellites.
Third, the competitive landscape is getting more complicated. There are worries about over-capacity, as a new generation of GSO satellites increase per-satellite throughput dramatically. GSO players are fighting to protect their investments against commercial and interference encroachment from non-GSO systems. For their part, the non-GSO systems compete against each other not just for customers, but also in mitigating interference between systems with very different orbits. And that doesn’t even address the non-satellite competitors: terrestrial cellular coverage is spreading ever deeper into rural areas, threatening one of satellite’s key value propositions, and new technologies like stratospheric drones and balloons may compete in earth observation services as well as delivering rural broadband.
There is a lot of uncertainty. Nobody’s sure which non-GSO systems will make it into orbit; the only certainty seems to be that not all the currently proposed systems will fly. Opinions about the profitability of non-GSO constellations range from optimism to profound skepticism. We don’t know how much inter-operator coordination will be needed to avoid harmful interference. It might be none, or it might be a lot – and it’ll depend on who makes it to orbit. Even if no coordination is required, there’s uncertainty about which interference risk mitigations – such as narrow antenna beams or transmit power limits – will be needed.
Risk-Informed Interference Assessment
Avoiding harmful interference between satellite constellations – some with thousands of satellites, at lots of altitudes, some moving rapidly across a user’s field of view – is complex. Different constellations, e.g. low vs medium earth orbit, present different risks to each other; the risk depends on operational parameters like orbit, uplink transmit power and antenna beamwidth; and the risk will vary from place to place depending on local conditions (e.g. rain fade) and orbit alignment (e.g. polar vs. equatorial). The high uplink transmit power needed to reach MEO satellites may interfere with LEO; and large LEO constellations may interfere with smaller MEO and highly elliptical constellations. There’s also the underlying uncertainty about which systems will make it into orbit, multiplying the number of interactions to model.
Nobody’s sure what measures will be needed (if any) to ensure that these non-GSO systems can coexist without degrading each other’s throughput. The risk of interference – in the absence of mitigation – could be negligible, or it could be substantial, at least for some configurations. Mitigation adds cost and complexity, but there is little published information on the quantitative risk of inter-service degradation, let alone on the cost-benefit trade-offs between different mitigation strategies.
We need new ways to think about managing inter-system interference given all these uncertainties. Quantitative risk assessment, a mature technique used by just about every regulated industry except wireless, shows promise as a spectrum management tool.
Engineering risk is the likelihood and consequence of potential sources of harm (aka hazards). Risk combines probability with severity. For example, I live in the Seattle area, and the hazard is getting wet. The risk is not just the probability of rain (likelihood), but also whether it’s a drip or a downpour (consequence). Weather forecasts often provide both the probability of rain and number of inches expected – that’s a risk assessment.
Risk assessment can be compared to worst-case analysis, which is frequently used in spectrum studies. Worst-case focuses on the single hazard with the highest perceived severity, regardless of likelihood. It’s just one scenario – a corner case – that doesn’t help one to make complex trade-offs between permutations of operating variable values.
Risks and mitigations
The dominant inter-constellation interference hazard is co-channel operation on overlapping paths between earth stations and satellites, aka in-line events. These events can occur when a ground transmitter’s beam illuminates a space receiver from another constellation that’s in the line between the ground and the intended receiver; or when two ground receivers close to each other are looking at respective space transmitters that are in the same direction. Other risks include adjacent channel interference, and interference between earth stations, or between space stations. There are also baseline hazards even without interference, including degradation of the desired signal (e.g. due to atmospheric attenuation), and non-interference faults and failures like misconfiguration, hardware faults, and operator error.
There are many possible mitigations. For example, look-aside: if a constellation has multiple satellites visible simultaneously, its ground transmitter could point to a space receiver that’s not in-line with a receiver from another constellation, so avoiding interference. This would also work on the downlink, where ground receivers close to each other look in different directions and reject interfering signal due to high antenna gain. If this isn’t possible (say because satellite paths that aren’t in-line aren’t available, or operators of two constellations haven’t agreed on how to point in different directions) then systems can split the band – one constellation uses the lower half, say, and the other the upper half.
Risk-Informed Interference Assessment can improve system design by showing which elements are most important, integrating information from many sources, including historical data, expert intuition, understanding of system behavior, and modeling. It allows one to challenge assumptions and reveal misconceptions, correcting misplaced optimism or pessimism. One could explore the key vulnerabilities of different kind of satellite architectures, i.e. LEO vs. MEO, and large vs. small constellations. Operators argue before the FCC about who’s the bigger victim –it could be both, depending on the circumstance, e.g. downlink vs. uplink.
It can also help to compare the relative risk of degradation under various assumptions for everything from the number of satellites to the out-of-band emission mask. For example, which system parameters have the biggest impact on interference risk: the number of satellites or ground stations, altitude of a constellation, antenna gain, uplink transmit power, out-of-band emission mask, receiver selectivity, and/or something else? Similarly, one could explore which mitigation techniques offer the most leverage: band splitting, look-aside, align channels, adaptive links, or something else? Mitigation adds complexity, which generates operating error hazards; it’s also costly. Is mitigation in fact needed, or is it money spent to avoid interference that doesn’t occur? RIIA can thus be used to assess what mitigation technologies or operating rules would be most effective. Is an incremental dollar (or kilogram, or watt) better spent on antenna gain or more sophisticated in-line avoidance? If the modeled risks are sufficiently low, the community might decide to trigger inter-operator coordination by observed performance degradation, rather than trying to prevent putative low-risk hazards in advance.
As more non-GSO satellite constellations become operational, the industry’s customary approach of avoiding interference through bi-lateral coordination becomes more challenging. With three players, there are three bi-lateral relationships; but with six players, there are fifteen bi-laterals. A dozen companies have applied to the FCC to operate a non-GSO constellation in the Ku/Ka-band.
Some sort of multi-stakeholder process seems to be necessary. However, the satellite industry hasn’t had much success with multi-stakeholder arrangements beyond ITU-R study groups, and satellite operators prefer proprietary solutions over industry standards. This is in stark contrast to the terrestrial wireless industry, where global standards bodies like the 3GPP and IEEE, and trade associations like the CTIA, GSMA and Wi-Fi Alliance are a fixture.
The loner approach makes sense when interference problems are relatively simple, and low equipment volumes don’t justify the cost of global standardization. But both conditions are changing for satellites: avoiding interference between many independent constellations appears to be quite complex, and satellite constellation business models only make sense if device volumes are large.
Experience in other industries suggests that sharing data is essential for risk reduction, and that industry forums are a good way to do this. For example, after the Piper Alpha oil-rig disaster in the North Sea, Norway required probabilistic risk assessments. This necessitated compiling data on component failure rates, but operators were unwilling to share company-confidential information with each other. This led to the creation of a 3rd party database; as an incentive to participate, data contributors got free access, while non-contributors had to pay. This database not only reduced risk, but also reduced operating costs since data on part failure increased reliability.
Sharing satellite operating data may be necessary, and will be helpful, in reducing the risk of orbital collisions and harmful interference. Several satellite-related multi-stakeholder groups already exist, though one can debate their suitability to host a non-GSO coordination multi-stakeholder group. One candidate is the SDA (Space Data Association), a non-profit international association of satellite operators that works towards sharing safety-critical data about the space environment and the RF spectrum. ITU groups such as the SFCG (Space Frequency Coordination Group) and professional bodies like the AIAA (American Institute of Aeronautics and Astronautics) may also have a role to play.
While risk-informed interference assessment and multi-stakeholder processes are helpful on their own, they’re better together. A multi-stakeholder organization provides a forum for engineers to hammer out rough consensus on the base assumptions for risk assessments. The computer modeling that follows can then inform community decisions about which mitigation techniques to focus on, and when.
This synergy is like the virtuous circle of "rough consensus and running code" pioneered at the IETF by internet engineers interested in practical, working systems that can be quickly implemented. The satellite business is speeding up and spreading out. The rich decision support and agility provided by risk assessment and multi-stakeholder collaboration will soon be a competitive advantage and may become essential to success.