VIA E-MAIL to
22 September 2016
The Manager
Economic Advisory Section
Strategy and Research Branch
Australian Communications and Media Authority
PO Box 13112
Law Courts PO
Melbourne, VIC 8010
Re: Review of Taxation Arrangements for Satellite Services (“the Consultation”)
Dear Sir/Madam,
O3b Teleport Services (Australia) Pty Ltd[1], as an active satellite operator and user of the spectrum under consideration in this Consultation is grateful for the continued opportunity to work with the ACMA as it refines and updates its policies regarding the fast-evolving services being provided to Australia via satellite, such as broadband internet. O3b holds multiple apparatus licenses; is a licensed Australian telecommunications carrier[2]; and was recently fully acquired by SES S.A. (Euronext Paris and Luxembourg Stock Exchange: SESG).
As the ACMA know, O3b operates a multi-satellite, global, non-geostationary orbit (“NGSO”) Ka-band satellite constellation. O3b now has 12 satellites in orbit, and has secured US$460M in funding to launch another 8 satellites by early 2018. We are now the largest supplier of satellite capacity to the Pacific,[3] including Australia's own Norfolk and Christmas Islands. Our current constellation is scalable and we will continue to add satellites as market demand dictates. Two of O3b's nine gateway earth stations are in Australia (one outside of Perth, Western Australia, and another one in Dubbo, New South Wales). The others are in the USA, Peru, Brazil, Portugal, Greece, and Pakistan.
O3b would like to commend the Economic Advisory Section for addressing the disproportionately high taxes that Australia charges for Ka-band satellite spectrum. Australia’s spectrum fees are among the highest in the world and the current fee formula creates disincentives for the use of higher-bandwidth services such as satellite-enabled broadband. O3b has developed highly efficient broadband services at fiber-equivalent speeds using transponders that are sometimes as much as 240 MHz wide (typically 216 MHz) - over 6 times the bandwidth of the traditional GEO transponder. Thus the revamping of the fee formula is directly pertinent to us.
Below, we offer our answers to the questions posed in the Consultation.
Question 1. Do you have other evidence about demand and/or congestion in the Kaband that could improve the ACMA’s views about the Ka-band tax rates for satellite services?
O3b supports the conclusion that the Ka band is not congested, and thus economic incentives to “ease congestion” are not needed at this time.
Question 2. Do you have any evidence of more appropriate taxation frequency ranges (tax brackets) for the Ka band that would better support opportunity cost pricing principles?
O3b has no comment on this section.
Question 3. The proposed 50 per cent tax premium for NGSO services is based on spectrum denial analysis of a number of scenarios in the 18 GHz and 50 GHz bands using various elevation assumptions. Would spectrum denial analysis in other bands between 18 and 50 GHz produce results that are materially different to the analysis already undertaken?
Although O3b is pleased to see that the overall fees for satellite services at the Ka band will be reduced, we respectfully point out that the ACMA’s proposal for a 50% premium on NGSO services is based on faulty assumptions in the Plum Report, which threaten to render ACMA’s proposals arbitrary.
Not all NGSO satellite systems are the same, and thus a single inflexible rule is not appropriate. There are MEO, HEO and LEO systems; there are inclined, polar and equatorial orbits; there are elliptical and circular orbits; there are multi-planar networks comprising all of the above. (Smallsats, another recent satellite innovation, will likely merit another consultation entirely.) Therefore, many of the Plum Consulting “assumptions” over-reach, leading to potentially random or arbitrary conclusions.
Perhaps the most egregious example of faulty assumptions is that there is any need for a “denial area” premium at all. Plum explicitly acknowledges after extensive analysis that the Ka band is not congested (Plum Report, page 2). To then say they are interested in a “generic potential [denial] area particularly in a congested environment” (Plum Report, FN 76) exposes Plum’s bias toward imposing a solution even in the absence of any inefficiency. Plum could do an analysis of a potentially congested band, so they did. But both Plum and the ACMA seem to have overlooked Plum’s finding that the Ka band is not congested. There is a legitimate question of whether “denial areas” need to be compensated for at all.
Additionally, assuming that all NGSO earth stations will have low elevation operations occurring in all directions of azimuth (Plum Report page 82) is overly simplistic and frankly incorrect when assessing the denial areas, if any, of O3b.
Critical considerations come into play:
· Terrain around the earth station will impact what look angles the earth station could accommodate from any given location, meaning that terrain blockage may significantly reduce the area actually impacted, even at a 5-degree elevation angle.
· Power levels will change: Plum assumes “typical” FSS earth station levels (Plum Report Table 8-1), however, we know not all earth stations perform the same task, and hence may not all have the same HPA or gain pattern etc. For example, to determine the distance needed between a transmitting earth station and victim receiving terrestrial station, power output from the FSS earth station will be important to determine actual distance needed to protect the terrestrial station. The same principle will apply with respect to receiving earth stations interfered with by transmitting terrestrial stations. Not all terrestrial stations are the same (e.g. fixed vs mobile), and specifically, the power levels may be different based on application. The distance needed to protect the FSS earth station will correspondingly change.
Plum assumed the GSO and NGSO parameters in (Plum Report Table 8-1) were the same when in reality they may not be.
As an example, O3b is a circular, equatorial, MEO, NGSO satellite system operating with an earth station in Perth with 5-degree elevation angles from West to East (which is the worst case for O3b – it’s the maximum swing from one horizon to the other). This results in a modest 7.3% increase in the “denial area” above that of a collocated GSO earth station in Perth also having an elevation angle of 5 degrees operating with a hypothetical satellite at 42°E. See images below for this worst-case scenario where the GSO (right) has a single area denied but the O3b earth station (left) has the additional 7.3% area above that of the GSO area. This is the worst case premium for O3b, and the “denial area” will only decrease from there.
Figure 1: Hypothetical worst-case denial areas for O3b with earth station
compared with a GSO earth station both operating from Perth
GSO earth station
Notwithstanding the above, should the ACMA decide to use percent-area based premiums for NGSO systems above that of the GSO, the value should be derived from the actual operations and include location-specific details (e.g. local terrain). To support this assertion, using the same assumptions from the Plum Report, one NGSO earth station operating from Perth has the denial area shown in Figure 2 and is compared against a fixed-pointing GSO earth station. The NGSO earth station (left image) uses actual antenna pointing strategies for an existing earth station operating with O3b’s MEO NGSO system. The GSO earth station (right image) is at the same teleport (PITC) as O3b and communicating with a satellite at 62°E. For the NGSO earth station, the elevation angles vary from 24° at the azimuth angle of 136° to 20° at the azimuth angle of 38°. The maximum elevation angle is 38° at the azimuth angle of 90°. For the GSO earth station, the elevation angle is a constant 22°.
Figure 2: Representative denial areas for O3b compared
with a GSO earth station both operating at PITC
GSO earth station
Although there is a difference in area between these two, it is closer to a 2% increase in area for the NGSO earth station above the GSO earth station as opposed to the presumed 50% increase in denial area. Should the ACMA continue to feel that a penalty needs to be imposed on NGSO systems, the formula for that assessment must be able to take into account the actual system needs and actual denied area.
O3b notes that some existing GSO earth stations have lower elevation angles than O3b at the same teleport location. It seems arbitrary to assume than an NGSO will always have a larger denial area than a GSO.
Also, although some NGSO constellations may have full flexibility to point in all azimuth directions, and thus may demand that their earth stations have the same flexibility, O3b – as the only extant MEO NGSO operating in Australia - does not need this and so should not be penalized through this fee structure. Even Plum itself admits that its “denial area” calculations are based on a “generic potential area particularly in a congested environment,[4]” which Plum admits elsewhere is not the actual situation in the Ka band.[5] To impose a blanket “50% premium” is just as arbitrary as the current fee formula.
As examples of the limitations of Plum’s research and assumptions, we note the following:
1. O3b has been licensed in the 17.6-19.3 GHz band since 2013, so Table 2-5 on page 12 of the Plum Report is inaccurate.
2. O3b is not mentioned in Section 2.4.3, even though we have been providing Ka-band services (notably to Christmas and Norfolk Islands) in the Asia-Pacific region since 2014. If Plum meant to distinguish between GSO and NGSO, that was not clear.
3. The assumption in Section 3.3.1 that “trunked satellite communications services have shown slow growth in recent years partly because of competition from fibre for long-haul traffic” does not reflect O3b’s experience as a provider of high-throughput connectivity with fiber-equivalent performance, with up to 1.8 GHz of capacity per beam. In fact, O3b’s “trunk” capacity has seen such strong uptake (including multiple requests for upgrades from existing customers) that O3b has become the fastest growing satellite operator in the Pacific,[6] and is expanding its constellation to include eight additional satellites to meet customer demand.
4. Section 3.3.2 misstates that O3b’s licenses are “class licenses” (they are standard “earth” and “earth receive” apparatus licences), and implies that our system has yet to be proven “technically and commercially,” when in fact we have been operational in Australia since 2014.
5. Plum compares Australian earth station use (and fees) to 10 other counties (table 5-2), but these countries are mostly quite different in terms of geography and population density, which is directly relevant as Australia’s fee structure is based on population density. Comparing fees in Australia (where the population density is 3.1) to fees in Hong Kong (where the population density is 6,957.8) is untenable.[7] When Plum states in section 5.3.2 that there is a failing in fee systems that do not have an incentive to locate away from high density areas, Plum disregards that most of the countries in Table 5-2 barely have “medium-density” let alone “low-density” or rural areas (with the exceptions of Canada and the United States). Hong Kong would be hard-pressed to find a low-density area at all, let alone one in which it could encourage satellite earth stations to locate (section 5.3.2). It is inappropriate to consider those smaller-sized countries in terms of “regional” satellite services, since for most satellite systems a single beam covers that entire country. The use of data from Singapore and Hong Kong may have seemed appropriate from a shared-ITU-Region point of view, but these countries are not on par with Australia for the purposes of this particular assessment.
6. It is particularly troubling that Plum used auction prices as market benchmarks (section 6) for satellite services. It is by now a well-accepted economic principle that satellite services are not a good candidate for spectrum auctions[8] and so this entire section – based on auctions for terrestrial use of spectrum – is a distraction at best, but at worst could lead to reliance on an unsupported economic principle. (O3b notes as well that Plum omits from its list of comparative auctions (table 6-1) the very Australian auction that led to a wide swath of the Ka-band under discussion being left fallow for 15 years. Although the 1999 auction may have made money for the Australian government, it was at the expense of spectrum efficiency, and the poorly-utilized “spectrum licence” was a severe barrier to entry for innovative new users such as O3b until the ACMA returned to traditional licensing of that band in 2014.[9])
Question 4. Do you agree with the eligibility criteria for the proposed Earth station spectrum-sharing incentive of a reduction in taxes of 30 per cent?
O3b agrees with the idea of incentives to encourage spectrum-sharing, as opposed to penalties (“premiums”).
Question 5. Do you have any evidence that would improve the ACMA’s understanding of the application of opportunity cost pricing to CDMA satellite systems in the UHF band and the proposed tax reform?
O3b has no comment on this proposal.
Question 6. If the ACMA proceeds with the tax reform proposals presented in this paper, are there any implementation and/or transitional issues that the ACMA should be aware of?