LEO Broadband Constellations - Technical and Economic Truths

Sep 22, 2015 | Publisher: incog | Category: Technology |  

LEO: Roar or Whimper Low Earth Orbit Broadband Constellations:: Technical and Economic Truths ICG 22 September 2015 LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 2 22 September 2015 LEO: Roar or Whimper? Low Earth Orbit Broadband Constellations: Technical and Economic Truths 1. INTRODUCTION There is currently much activity related to the development of new Low Earth Orbit (LEO) satellite systems designed to deliver high throughput broadband connections, and aiming to address those regions of the World which remain underserved by existing data connectivity. The emerging markets exhibit huge addressable demand for these high throughput satellite (HTS) systems. If LEO systems can overcome their economic challenges and raise the many billions of dollars needed to deploy, services could theoretically be launched around 2020. This would be in addition to the many existing geostationary orbit (GEO) satellite services which address similar requirement. Several organisations promoted similar LEO systems in the 1990s and in the process consumed over $10 billion of Wall Street Capital. The risks to the current batch of systems are similar, but potentially even more extreme:  System Complexity: The design, manufacture and launch of a large number of LEO satellites, ten times the scale of any past constellation, is complex and costly. Those satellites then have to be put into orbit and distributed across a number of orbital planes. Once that is complete and the constellation is operational they then have to be managed and maintained for the lifetime of the overall system. None of the stages of this process are trivial and collectively present a huge programmatic challenge.  Cost of Manufacture: The new LEO constellation proposals are founded on the notion that the changes in technology, with miniaturisation of components, high processing power and solid state sensors have enabled a new paradigm in low cost satellite design and manufacture. This is a false notion: The need for high reliability, radiation-hardened electronics is as demanding now as it was 25 years ago. The costs have not changed and in many areas where testing is now more complex to meet higher performance requirements, and in some areas, such as RF emissions, those costs are now higher.  Schedule: The need to manufacture a large number of satellites of the same type and launch those in a relatively short period of time is as challenging today as it was in the 1990s. Satellites have never been commoditised. They have a tendency, even within a fleet, to have individual design goals that require changes to the spacecraft in each manufacturing cycle. The low cost LEO constellation manufacturing model requires large volumes of spacecraft and their associated subsystems to be produced in a production line manner. This has to be achieved whilst maintaining LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 3 22 September 2015 quality, reliability and thorough testing regimes. There is no manufacturing model on which to base this. As has been detailed in this paper no satellite facility available today that can support this operation. It would need a new approach with a new manufacturing method and test regime in a large facility. That requires an extensive investment that falls to the bottom line cost of the LEO satellite system it’s supporting.  System Operation: The operation of a LEO constellation is more demanding today than it was 25 years ago. The need to manage the satellites in orbit for collision avoidance comes with added vigilance given the growth in the number of satellites in similar orbits during this period. 25 years ago the possible collision of satellites was speculated about but today it’s a reality and has been witnessed. The consequences have been demonstrated. This increases the constraints on satellite operations and maintenance with the additional associated cost. In addition the need to deorbit satellites has been implemented in the past 25 years and adds to the design complexity of the satellite and the chosen orbit.  Regulation: Regulatory rules drive radio frequency interference requirements. These now have tighter controls that are more exacting than any time in the past 25 years. They are more demanding now than they have ever been. Analysis and measurement techniques have driven the ability to understand and control RF parameters at the level of refinement not seen in past LEO programs. The growth in GEO satellites and the use of Ku and Ka-Band has driven the need for ever tighter coordination between spectrum users. The new LEO constellations are required to operate on a secondary basis to the established GEO operators in their spectrum allocation. The burden of interference avoidance falls heavily on the LEO operators. This comes with very complex and expensive satellite switching and frequency coordination requirements that drive constellation design. GEO operators are not expected to be tolerant about the risks of interference and will demand certainty, not theoretical promises.  User Terminals: There has been a huge change in electronic design and manufacturing in the past 25 years. Terrestrial communications technology has matured to the point that commercial mass production with high reliability is now common place. However, the core challenge for LEO User terminals is the same today as it was 25 years ago. They need to track the fast moving satellites in view without interruption in service. This requires the terminal to have some mechanical tracking system, with associated reliability issues, or an electronically steered antenna with extremely high component and manufacturing cost which at present is a theoretical concept subject to some memorandums of understanding (MOU)s and R&D projects. Either of these solutions brings huge challenges to the user terminal design and this was the main downfall of the high frequency LEO constellations of the 1990s.  Service Cost to the End User. The extensive design and manufacturing requirements drives additional cost into the final end user service pricing. Given the many regulatory, manufacturing, launch and on-orbit operational costs as well as the subscriber terminal pricing, the cost of service delivery will fall outside of a competitive range when compared to GEO delivery costs. But also most of the LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 4 22 September 2015 capacity of a LEO system will fall on unpopulated oceanic, polar or desert regions, meaning the true cost per usable Mb must be re-calculated. When compared with one megabyte of data delivered via a GEO system the comparative LEO cost will be much greater – for a user in Africa for example, the capital investment will be 50 cents per MByte for a LEO system compared with 7 cents per MByte for a GEO solution. So LEO system are 7 times more costly than GEO, which begs the question – why would a financial investor fund it?  Constant High Capital Expenditure: Given the need to manufacture a large volume of satellites at a pace that will meet the strict launch cadence, there will be significant schedule pressure. Even if we optimistically assume that all the schedule pressure is successfully absorbed and launches are completed, the entire cycle will have to be restarted. This is needed to replace satellites reaching the end of their life before failure and to ensure the user service experience is maintained at a given quality level. This presents a need for constant investment of capital and the need to meet tight schedules over many years to first deploy the service and then undertake ongoing service maintenance. This paper establishes that not unlike the LEO systems of the 1990s, project promoters, component suppliers and aerospace manufacturers stand to make profit from building the proposed systems and thus their support is rational (albeit it perhaps cynical) but we find it highly improbable to foresee a successful outcome for financial investors. The most likely outcome will be that some systems reach design testing and initial satellite deployment to establish spectrum rights and run constellation tests, but as the cost to implement the full system becomes evident due to the complexities of the user terminals, the pace of implementation will slow. As further regulatory and operational challenges become evident and costs climb, the LEO systems will be in an incomplete state or close down after initial launch and testing. In conclusion, there are technologies to be invented, factories to be built and new manufacturing concepts to be deployed for the first time and regulatory changes necessary in conflict with the interests of many other companies. Even if these challenges are overcome, the system would not be cost competitive and given its short duration it is difficult to see how a financial investor could get a return. As in the 1990s, LEOs are concepts which could enrich manufacturers at the cost of Wall Street. LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 5 22 September 2015 2. BACKGROUND This paper examines the technical and business case for proposed Low Earth Orbit (LEO) satellite systems. These systems have seen a recent increase in interest and a number of competing but similar initiatives have been publicly announced. LEO systems require complex technology and prolific constellations of satellites arranged to provide constant communications access to users. There are many aspects of implementing these systems that have to be considered in order to provide comprehensive coverage and meet cost constraints. This paper examines the case for the various proposals both from a business perspective and the technical complexity of implementing the system. The design choices made when creating these systems drives the satellite, ground segment and launch costs as well as the ongoing user experience and costs. There has been a surge of interest in satellite constellations to provide global, high bandwidth communications services with the announcement of a number of projects over recent months. LEO satellites are located between 500 and 2000km above the Earth’s surface in inclined orbits, necessary to cover all or most of the Earth’s surface. The satellites in a constellation are launched into a number of orbital planes, with the satellites moving in phase with others in the same plane. The use of many satellites, in many planes, can provide coverage of the Earth all the time, even at the North and South Poles. Satellite constellation design is extremely complex, involving many issues and trade-offs that simply aren’t required for geostationary satellites. Iridium, for example, spent nine years designing and developing its technology before the decision to start satellite manufacture was taken, then another seven years to get landing rights (and never got them in some countries). However, the dream of placing very large numbers of low cost satellites into LEO has once again captured the imagination of many just as it did back in the late 1990s when a raft of constellations were proposed, all of which were either scrapped on the drawing board or fell into bankruptcy once in orbit and after many billions of dollars had been spent. Table 1 shows some of the many proposed LEO programs during this period. Constellation Name Partners Number of Satellites Altitude (km) History/Status Globalstar Loral, Alcatel, Qualcomm 48 1410 Started 1991, Chapter 11 1999 ($3.3B debt), Cited: Constellation issues 1 Iridium Motorola 66 780 Started 1998, Chapter 11 1999, Cited: Lack of satellite in orbit and handset issues 2 1 http://www.skyhelp.net/acrobat/Globalstar-chap11.pdf LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 6 22 September 2015 Constellation Name Partners Number of Satellites Altitude (km) History/Status Skybridge Alcatel, Loral 64 (later 80) 1469 Halted development in 2002 as the cost of ground terminals was higher than expected 3 Teledesic Motorola, Boeing, Matra Marconi, Bill Gates, Craig McCaw 288 1375 Never got beyond a paper design 4 Table 1: LEO Satellite Constellations Proposed and Built in the 1990s In the years between the 1990s and today much has changed in electronic technology. Advances in very compact complex electronics has been rapid and the move towards digital signal processing has changed how voice, video and data is communicated. These changes have encouraged a new wave of optimism and as a result, interest has grown in revisiting LEO constellations5. The main argument which still drives this interest in LEO solutions is that it can offer information access to everyone, everywhere. Also in the intervening years, the roll out of cellular technology with advances through 2G, 3G and 4G LTE, as well as the increasing coverage of High Throughput Satellites (HTS) in the US, Europe, Africa and Latin America, has made ubiquitous access to information an expectation. The ease with which connectivity is provided to a huge part of the world’s population today has driven some to again question if LEO is the ultimate answer. They look at how terrestrial-based ‘anytime, anywhere’ technology, which makes access so easy, could be used in space to make seamless access across the whole globe a reality. Riding on this new wave of optimism a number of new LEO satellite constellation have been proposed as shown in Table 2. System Name Company Number of Satellites Orbital Altitude OneWeb WorldVu 648 1200km SpaceX SpaceX ~4000 TBD LEOsat LeoSat 80-120 1400km Table 2: Recently Proposed LEO Satellite Constellations The underlying hardware and software complexity used in commercial terrestrial services is constantly pushing at the edge of technological capabilities. Constant change 2 http://www.washingtonpost.com/archive/business/1999/08/14/iridium-files-for-chapter-11/29828512-3dce- 4a0a-8e8e-b5725c72cad5/ 3 https://www.flightglobal.com/news/articles/alcatel-set-to-scrap-skybridge-project-140940/ 4 de Selding, Peter B. "Teledesic Plays Its Last Card, Leaves the Game". Space News, July 14, 2003. Accessed March 15, 2010, via the Internet Archive 5 http://www.nsr.com/upload/presentations/NSR_Webinar-_LEO_HTS_Constellations-_May_2015.pdf LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 7 22 September 2015 is a hallmark of the terrestrial communications market. However, when this technology is moved into LEO applications the on-orbit challenges of space raise the quality, test and qualification requirements and in the meantime the laws of physics have not changed with engineering advances. Commercial technology is not easily applied in space and the costs of making the latest technology usable in space escalate quickly. The need for reliability and redundancy in satellite systems, even in LEO orbits, is demanding. The added challenges of launching and establishing constellations requires a level of oversight that is not required in terrestrial systems. Today’s Geostationary Orbit (GEO) satellites provide communications that is competitive in cost and global in coverage. The LEO constellations being proposed today have to exceed both the existing GEO level of service and at a lower cost to be competitive. This paper examines each challenge, compares and contrasts the options and sets out the expected future for the proposed commercial LEO satellite systems. LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 8 22 September 2015 3. THE PROMISE LEO communication satellite constellations offer the chance to place large amounts of capacity into orbit to provide ubiquitous coverage of the Earth’s surface. As these satellites are orbiting only a few hundred kilometres above the Earth’s surface they should, in theory, provide improved link budgets and lower latency when compared with their GEO counterparts. Strong potential link budgets combined with large numbers of satellites means that a large volume of capacity can be placed in orbit. For example, a 600 satellite constellation with 6 Gbps of capacity per satellite can deliver 3.6 Tbps, a huge amount of capacity equivalent to about 18 GEO HTS systems. If this can be delivered at a cost of $4B or less it seems to compare well at first sight, with the GEO HTS current equivalent cost of about $7B. The LEO constellation is capable of providing coverage over the whole globe including at the North and South Poles. The number of satellites in the constellation is seen as offering ‘graceful degradation’ if a single satellite becomes unusable and is easily replaced from in-orbit spares. However, even given these perceived advantages LEO communications constellations are not without challenges and a more detailed analysis reveals that they may not deliver what they promise. LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 9 22 September 2015 4. THE CHALLENGES When designing a LEO constellation there are a number of operational and technical requirements that need to be met. These present significant challenges to the programme design team. They must navigate a wide range of compliance needs whilst meeting complex design constraints, within tightly controlled cost boundaries and to a demanding schedule. 4.1 Constellation Design When designing a constellation, all the design criteria applied to a single satellite mission still apply. Thus, it needs to be established that each satellite meets the standards for launch, has the capability to survive and provide extended on-orbit operations, will be able to seamlessly communicate with ground stations and user terminals and can be de-orbited at the end of its design lifetime. In addition, for constellations, the number of satellites, their relative positions, and how these positions change with time, both in the course of an orbit and over the lifetime of the constellation also need to be taken into consideration. All in all, a very complex trade-off analysis and an iterative process has to be performed from the start of the definition stage of any satellite constellation. This long and expensive research and development period can be a difficult challenge to overcome. As a result, many constellation proposals were never actually manufactured or deployed. Even the Iridium constellation was in the design phase for 9 years prior to any manufacturing taking place, having evolved through several iterations regarding the number of satellites, orbit selection, technology choices, etc. O3b provides a similar story, where even a far simpler constellation with only eight operational satellites (the four others being the subject of a total loss insurance claim) still took 8 years from inception to full operations. A trade-off analysis is always necessary when dealing with constellation design. Table 3 below summarises the most common trade-off parameters to consider when designing a constellation. Principal Design Variables Secondary Design Variable  Number of Satellites  Inclination  Constellation Pattern  Plane Phasing  Minimum Elevation Angle  Eccentricity  Altitude  Size of Station-Keeping Box  Number of Orbital Planes  End-of-Life Strategy  Collision Avoidance Table 3: Trade-off Parameters for Constellation Design LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 10 22 September 2015 Most of the parameters introduced in Table 3 relateto coverage, with the number of satellites, constellation pattern, altitude, number of orbital planes and inclination the most important factors in this regard. The fact that coverage is defined by at least five parameters shows that its definition is a very complex task. Furthermore, the elevation angle required for visibility from a user terminal also influences constellation coverage as it determines single satellite coverage. Increasing the number of satellites in a constellation is one way to achieve global coverage however, the complexity and the cost of the constellation also increases. Nevertheless, the inclination of the constellation orbital planes is an important factor. If, for example, the inclination is 30°, then the coverage will be restricted to ±30° latitude regardless of the total number of satellites in the constellation, whilst if high inclinations are adopted, a disproportionate amount of coverage is focussed on the polar regions. Achieving global coverage is the main reason constellations such as Iridium, Teledesic or OneWeb decided to choose an inclination close to 90°. Polar orbits are the only kind of orbit that allows a constellation to have complete latitude coverage over the Earth. This type of constellation is called a Polar or Star constellation. Star constellations are characterised by the formation of a “street” of coverage at every orbital plane. In general, there is little or no satellite coverage overlapping in low to mid- latitudes and, to continuously cover the Earth, a high number of planes are needed. Coverage overlapping only occurs over Polar Regions where numerous satellites can be seen at any given time. The number of satellites per plane and the number of orbital planes depend on the altitude and the minimum elevation angle required to operate the user terminals. This relationship between street of coverage, elevation angle and altitude produces discrete jumps in coverage. Therefore, coverage obtained by this kind of constellation does not vary continuously and smoothly with altitude. Figure 1: Star Constellation Layout, Teledesic design (288 Satellites) On the other hand, Delta or Rosette constellations are characterised by their latitude coverage constraints and the overlapping of single satellite coverage. Delta constellations intend to provide continuous multiple, overlapping coverage of the Earth’s surface with the smallest number of satellites. However, they do not provide LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 11 22 September 2015 good coverage at latitudes higher than the orbital inclination. Instead, they provide coverage with visibility of multiple satellites from a single ground terminal at the mid- latitudes where most human population lies. The Globalstar constellation is an example of Delta constellation. Figure 2: Delta Constellation Layout, Globalstar Another important characteristic of any satellite constellation is the number of orbital planes in which the satellites reside. Constellations tend to look for a symmetric structure since it simplifies operations. Constellation symmetry requires an equal number of satellites in each orbital plane, but because moving satellites between planes uses much more propellant than moving them within a plane, it is highly advantageous to place more satellites in a smaller number of planes. If a satellite fails or a new satellite is added to a given orbit, the remaining satellites can be re-positioned so that they are uniformly spaced. Star constellations are not very efficient in terms of the number of orbital planes needed for continuous global coverage. Constellations such as Teledesic or OneWeb would need about 20 orbital planes in order to service the whole Earth. The high number of orbital planes increases the complexity of the whole system and launch programme as satellites are launched into one plane at a time. Moreover, the more planes a constellation has, the more in-orbit spares are needed (at least one per orbital plane). The altitude of the satellites in a constellation is a significant factor in determining the number of satellites that is required to cover the Earth and the characteristics of the constellation. A lower altitude decreases free space loss and propagation delay, but means that the single-satellite coverage decreases, hence, to fully cover the globe more satellites are needed. Increasing the number of satellites in a constellation increases frequency reuse and overall system capacity, but also increases the overall system construction and maintenance costs. Besides, satellites at lower altitudes move faster relative to the ground which increases the frequency of handover and causes larger Doppler Effect on signals between ground terminals and satellites. Figure 3 shows the total Field Of View (FOV) depending of the altitude and elevation. As it can be seen, both parameters have a huge influence in the final coverage obtained. LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 12 22 September 2015 Figure 3: Coverage area depending of altitude and minimum elevation angle to the user terminal Figure 4 below shows another example of how elevation affects single-satellite coverage. Figure 4: Effect of elevation in Iridium Coverage (Left, 8.2°; Right, 20°) 0 5,000,000 10,000,000 15,000,000 20,000,000 25,000,000 30,000,000 35,000,000 40,000,000 45,000,000 50,000,000 55,000,000 60,000,000 65,000,000 70,000,000 75,000,000 80,000,000 85,000,000 90,000,000 95,000,000 100,000,000 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 C o ve ra ge a re a [k m 2 ] Altitude [km] Total coverage (FOV) 10 deg elevation 5 deg elevation 15 deg elevation LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 13 22 September 2015 4.2 Interference and Regulation Whilst the top level potential advantages of LEO constellations are apparent, there are in fact many complex technical, regulatory and commercial hurdles to be overcome both during development and during operations of a LEO constellation. These can be summarised into discrete challenges that, when considered at programme level, present considerable technical, cost and schedule impacts. LEO constellations are required by the ITU, as part of their spectrum allocation conditions, to avoid interfering with previously established satellite networks operating in GEO and MEO6. Unlike the LEO systems of the 90s, the new generation plan to operate in the same frequency bands as the world’s TV and satellite broadband services. This means that the LEO satellites must avoid transmitting co-frequency signals into existing satellite dishes when moving in the line of sight of, say, a TV broadcast satellite and its users receiving TV signals. Similarly, the LEO’s ground terminals cannot transmit towards any GEO satellites. The only way to achieve this is to include many additional satellites in the LEO constellation which the user terminal can point to whenever a satellite it is currently communicating with moves into line of sight with a protected GEO satellite. As a pre- requisite the user terminals must be capable of this complex tracking and re-pointing activity, without losing connectivity whilst it is taking place. Unlike the LEOs of the 90s, this new generation of constellations plan to operate in Ku or Ka bands, which are frequencies already extensively used by satellites in the heavily populated GEO arc for TV broadcasting, satellite broadband and a variety of government and scientific applications. Interference with terrestrial and GEO systems will be a key issue to be addressed seriously by national and international regulators if any of these projects are to be funded, unless financial investors are prepared to accept an open ended project termination regulatory risk, which has not typically been the case in the history of telecoms investment. At minimum, sophisticated and very expensive ground equipment that has not yet been commercialised (or possibly even invented) would be required to manage this process. But it is also reasonable to expect that the operators of GEO satellites are highly unlikely to take a relaxed view of spectrum interference, and those in the ITU Master Register always have the ability to say “Stop!” if interference is experienced. Indeed, the first skirmish between the new LEO constellations and GEO operators has already begun, with INTELSAT requesting the US FCC to block the launch of SpaceX test satellites7. Even if an interference mitigation plan could be demonstrated to be commercially practical, highly damaging interference could occur from faulty satellites or ground terminals in the LEO systems. The only way for LEOs to operate without the need for permission from GEO operators is to operate at such a low power level that the signals are not strong enough to interfere. But a weak radio signal is not 6 http://www.itu.int/ITU-R/go/space/en 7 http://spacenews.com/intelsat-asks-fcc-to-block-spacex-experimental-satellite-launch/ LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 14 22 September 2015 of much use to customers. It is therefore highly improbable that GEO operators will give their agreement through a Co-ordination Agreement prior to launch, and so a different approach would be necessary by the LEO systems. As well as LEO-GEO interference, the cases of interference between individual LEO satellites must also be considered. Constellations are, by nature, always in movement as their main goal is to cover most of the globe, thus intra-system geographical overlap, where two or more spacecraft from the same constellation cover the same area, is a further major challenge for LEO communication constellations. Whilst frequency coordination between two spacecraft in the same LEO constellation is challenging, coordinating spacecraft from different LEO networks is even more so. As the number of usable LEO satellite systems in space increases, the need for frequency co-existence between the new LEO satellite systems and the already existing LEO and MEO satellite systems increases rapidly. This coexistence can be in the space and time domains or any other possible domains such as polarisation or radiation pattern. The interference environment generated by LEO satellite systems is not completely known yet and studies have been conducted with the purpose of examining the feasibility of frequency sharing between other services and LEO satellite systems. Regarding intra-system interference, beam overlapping for LEO multi-beam systems reduces considerably the available capacity of the whole system. Things get even worse in the case of a LEO constellation that experiences additional inter-satellite interference coming from overlapping beams of other satellites in simultaneous visibility. The trend is evident toward multi-spot systems with increasing values for the frequency reuse factor, therefore an analysis of co-channel interference is mandatory during the design of these systems. On some occasions (especially when satellites are near inter-orbital plane crossings, i.e. poles for Star constellations), co-channel spot-beam footprints intersect on the ground. In these cases, the interference level is unacceptable, unless some countermeasures are taken. The most common techniques that can be used to avoid or minimize the occurrence of these situations are spot beam turn-off, intra-orbital plane frequency division, and inter-orbital plane frequency division: A) Spot Beam Turn-off: whenever the spots overlap by a determined amount, one of the two spot beams has to be turned off. Normally, the measure for overlapping is based on the distance on the Earth surface between the intersections of the boresights of the spots. B) Intra-orbital plane frequency division: satellites on the same orbital plane are assigned different frequency subsets up to specified modulo R (where R is preferably chosen to be a divisor of the total number of satellites on an orbital plane). LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 15 22 September 2015 C) Inter-orbital plane frequency division: the available frequency spectrum is subdivided into as many subsets as the number of orbital planes so that satellites on different orbital planes are non-interfering. All these techniques bring significant reductions in the reuse of the available frequency spectrum. Technique ‘C’ uses the spectrum least efficiently, while technique ‘B’ is generally less efficient than technique ‘A’. Technique ‘A’ is also the one that ensures the maximum spectrum efficiency. Therefore, for Star constellations, where several spacecraft are covering the geographical area at the same time, solution ‘A’, of switching spot beams off, is usually the preferred interference mitigation strategy. However, implementing any of these mitigation strategies brings additional complexity to operation of LEO constellations with increased operational planning and hence cost. A further regulatory issue relates to “landing rights”, i.e. the permissions required from national governments to allow signals to be transmitted in to their territory. LEOs require landing rights to serve virtually every country. GEOs have a well-established landing rights system in most countries in EMEA, the America and parts of Asia as a result of many decades of operation, although China and India in particular (containing 37% of the World’s population) continue to present difficulties for international satellite operators in this regard. LEOs will have to establish brand new landing rights regimes everywhere. It is highly improbable that China or India would grant landing rights (even Inmarsat took decades to achieve landing rights in India). This means that a LEO is going to be severely restricted in its addressable market. There will be little or no demand in mature economies. India and China should be considered to be off the table, placing most of the emphasis for revenue generation on South America, Africa and the Pacific. 4.3 Cost of Satellites Although LEO constellation satellites may be quite small, they are still costly. A typical 800 Kg LEO satellite currently will cost about $25M. It may be possible to bring this down if new manufacturing paradigms are implemented but preparations for this are likely to take several years. Cheaper satellites can be built using commercial off the shelf components (to a certain extent). However, lower cost means lower reliability. Space is subject to radiation levels not seen on Earth and the components of a spacecraft must therefore be robustly designed to withstand high radiation, as well as the harsh thermal, vibration and sonic conditions of launch and operations. This makes it very hard to use commercial off the shelf components designed for terrestrial use and therefore difficult to achieve a low cost. Current Cubesat manufacturers do use COTS components, but Cubesats themselves typically last a year or less in orbit. Some manufacturers are looking at radiation tolerant commercial components and screening them as a way to reduce costs. This would involve careful selection through testing of component parts - an expensive and time consuming option. In addition, further mitigations must also be put in place in the form of operational workarounds such as power cycling of circuit parts that are classed as radiation tolerant. This would LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 16 22 September 2015 have to be conducted on a regular basis through autonomous resets that regularly cycle power or whenever an anomaly is detected. In terms of components, some arguments are put forward that building 1,000 satellites drives cost down. This is not necessarily accurate; in electronics manufacturing, one must see many millions of sets manufactured before a meaningful scale advantage is forthcoming. Traditional component manufacturers change their designs routinely and are unlikely to be helpful to low batch customers – thus the use of COTS components may necessitate constant re-design of the replacement satellites. The LEO satellite systems pose a huge manufacturing challenge to enable output of the high volume of satellites required. The bus and payload could be made in the same way that aircraft manufacturers make fighter jets in production lines. This is a high reliability production environment and whilst that is clean and debris free it is not in the class of environment that satellites require to ensure a maintenance-free 5-to-7 year life. There are also many subsystems that would be sourced from smaller suppliers who use high- touch labour production lines and that will not work in this volume production model. This would need considerable investment. By far the major cost driver in satellite manufacturing is touch labour. Reductions in this type of labour can be found with test automation but that comes with the need to invest in test facilities, training and test-script generation. This works well for testing digital and some linear components and the full satellite. However, this method is not easily adopted for Ku and Ka band RF components requiring repeated tuning and testing. Here the traditional test and alignment methods for high reliability space subsystems is highly labour intensive. To move to an automated test process would require changes in the approach to product design for manufacture which would be an expensive exercise requiring redesign of existing, high reliability space components. This would be very time consuming and challenging. Alternatively, components could simply be launched untested. If this is the approach taken in order to reduce costs, the GEO operators and world regulators concerned about RF interference, de-orbiting and collision risk may have more reason for concern. Whatever the final choice of production method at subsystem and spacecraft level, in order to meet the volume demand of the proposed LEO constellations, volume production of traditional space components would be required. 4.4 Capacity and Coverage Whilst constellations can put large amounts of capacity into orbit, most of this is unused while the satellites orbit over the world’s oceans and polar region’s (two thirds of the Earth’s surface is water). LEO constellations typically have highly inclined orbits in order to provide global coverage. The consequence of this is that the orbital planes of the satellites meet near the two poles. This causes “bunching” of the satellites at the poles enabling massive amounts of capacity to be delivered into the Arctic and Antarctic. However, proximity of the satellites in this global region, due to the constellation arrangement, would cause their transmissions to interfere with each other, requiring LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 17 22 September 2015 many of them to be switched off to avoid this. The figure below illustrates the position of the satellites from the now defunct Teledesic constellation, seen from the Earth’s North Pole at a given instant in time. Figure 5: Distribution of Teledesic Satellites as seen from the North Pole To understand the practical implications of this, consider Africa which occupies 6% of the Earth’s surface. It extends approximately 60° in both longitude and latitude. A system with 650 satellites delivering 3,500 Gbps will have roughly 15 satellites covering Africa at any one time (the 21 orbital planes are spaced by about 17° in longitude and each satellite in a plane is separated by about 12° in latitude). These satellites deliver about 2% of the constellations overall capacity or about 70 Gbps. This is only about one third of the capacity of a single large geostationary HTS. To have a better understanding of how the coverage of a Star constellation is distributed and the effect said distribution has on usable capacity, two examples are presented: Iridium and Teledesic constellations. Iridium consists of 66 satellites equally distributed in 6 orbital planes with a minimum elevation angle to close the link of 8.2°. Figure 6 shows the number of satellites that can be viewed by a user over the equator and the North Pole. It can be observed that over LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 18 22 September 2015 the equator there is only one satellite within the user coverage, but there are always between 6 and 9 satellites within the coverage range of the North Pole. The same behaviour can be observed for Teledesic. The Teledesic constellation pattern consists of 288 satellites equally distributed over 21 orbital planes with a minimum elevation angle of 40°. In this case, at the equator there is only one satellite that can be viewed by a user, but over the North Pole there are 16 satellites within the user coverage area (Figure 7). Figure 6: Number of satellites in-sight, Iridium constellation (Left: over Equator, Right: Over North Pole) Figure 7: Number of satellites in-sight, Teledesic (288 s/c) constellation (Left: over Equator, Right: Over North Pole) If the same analysis is now repeated over the whole African surface, it can be determined that at any given time Iridium has only 4 satellites covering the African continent which represents the 6% of the constellation total capacity. Meanwhile, the Teledesic concept would cover Africa with a maximum of 12 satellites at any given time. Thus, the African capacity is only 4% of the constellation total capacity. Ironically, the LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 19 22 September 2015 more satellites that are employed, the smaller the percentage of the system’s overall capacity is delivered around the equatorial countries. The analysis presented identifies the physical coverage without taking into consideration limitations in FOV from the antennas on the satellite (i.e. beamwidth). Antenna beamwidth coverage must also be evaluated in order to obtain the real coverage capabilities of any constellation. 4.5 Use of Inter-satellite Links One issue with LEO constellations is that the relatively low orbital altitude of the spacecraft means that only a small portion of the Earth’s surface can be seen at any one time. This means that a large number of gateway earth stations are required to connect users via the satellites into the terrestrial communications network. Both user terminals and gateway must be in view of the satellite at the same time, necessitating many gateways spread across the Earth’s surface and precluding communications whilst the satellites are crossing the oceans, for example. Some constellations use inter-satellite links to simplify the ground segment. The aim of including inter-satellite links is to create a space-based network. Satellites in such constellations must support on-board routing as well as on-board switching as the signals are passed from satellite to satellite before being terminated at a gateway. This allows a user terminal on the ground, below the satellite, to exchange traffic with gateways out of view of the satellite and onwards into the terrestrial network. However, constellation pattern or topology influences the complexity of an inter- satellite link system. There are two types of inter-satellite links: inter-plane and intra- plane. Intra-plane inter-satellite links lie within the same orbital plane between satellites following each other. They are generally permanent if the orbits are circular, as the satellites’ positions remain fixed relative to each other. Inter-plane links are between satellites in different orbital planes. These may not be permanent. Neighbouring orbits cross each other near higher latitudes, where each satellite’s neighbours will swap sides. This requires inter-satellite link antennas, that provide the communications between them, to either physically slew through 180° to follow the neighbour and maintain the link, or that the links be broken and remade. The increased relative velocities as orbits approach also results in increased Doppler shift, which must be compensated for. Adjacent orbits with inclinations near 90° have minimal Doppler shift and rate of inter-satellite link antenna slewing, and are therefore best-suited to inter- plane inter-satellite links over a wide range of latitudes. It is therefore unsurprising that inter-satellite links are usually integrated in Star constellation rather than Delta constellations. Unfortunately there are downsides to the use of inter-satellite links. These include their high cost and hence their impact on the overall cost of the space segment. More importantly, it also increases the risk of catastrophic failure since network failures due to hardware on the LEO spacecraft on orbit can only be repaired using redundancy which will drive up spacecraft cost. With 4-6 links required on each spacecraft plus LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 20 22 September 2015 redundancy, they add significant mass to each satellite in the constellation, thus increasing launch costs. They must also have extremely high capacity. If a single plane of 30 satellites is served by two gateways and each satellite delivers 8 Gbps of capacity, the inter-satellite links must support a maximum of 112 Gbps (i.e. 14 satellites with 8 Gbps). In addition, the gateways must support these throughputs as well in order to provide the ground to satellite links. Whilst inter-satellite links provide efficiencies from a technical perspective they can act as a block when one considers commercial aspects. For example, customers in developing economies are increasingly seeking locally launched traffic due to the proliferation of locally built data centres. Since inter-satellite links land traffic at a small number of gateways there is unlikely to be a gateway in the region concerned, potentially posing competitiveness issues. 4.6 User Terminals GEO systems benefit from the fact that their user terminals need have no moving parts. The most sophisticated GEO HTS systems have user equipment costing $325 per user. Millions of these terminals have been deployed and so most of the possible economies of scale have been realised for this type of equipment. It is unlikely that any satellite system (apart from GPS) will realise the benefits of scale of the consumer electronics industry, where additional benefits are reached at the hundreds of millions level. Unlike GEOs, LEO constellations require terminals that can track the movement of a LEO satellite as it crosses the sky. And not only that, they must also be able to seamlessly switch between two satellites when the first moves too close to the horizon or in line with the GEO arc. They would then have to make further switches to avoid interference with GEO satellites. One manufacturer has estimated that a switch would be required every 45 seconds. This inevitably gives rise to the kind of service interruption that occurs on a speeding train as one moves between cellular base stations. Having sets of two or more steerable antennas, which are delicate instruments located at high cost in harsh environments is not a sustainable business model for mass market adoption. Some companies have speculated that solid state phase array antennas could provide a solution. The physics is theoretically possible but not yet a practical engineering option, and unlikely to reach a consumer price point in the timescales of current projects. This was an issue for proposed LEO projects in the 1990s and became the downfall of the Ku-band Skybridge project. The technology in complex mechanisms is still constrained by implementation cost, serviceability and reliability issues today. Phased array antennas are currently under experimental research for use in the aircraft market with a 2-5 year view, but industry sources suggest that if the design and manufacturing problems could be overcome the cost per antenna is likely to be as high as $1m. This may be an acceptable cost for airlines, but not for rural villages seeking satellite broadband. LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 21 22 September 2015 The table below shows typical pricing for a GEO user terminal suitable for standalone operation in a remote region. It provides WiFi and 3G/LTE connectivity to the end user, as the OneWeb marketing collateral also proposes. Component Cost 75 cm fixed antenna $75 Satellite modem $725 Solar cells $120 Modular Power Supply & Distribution $250 WiFi PoP $250 3G / LTE small cell $250 Consumables $100 Installation $1,350 Total $3,120 Table 4: Cost of LEO user terminal The costs for a LEO user terminal may be similar to a high-end GEO terminal. Such terminals cost up to $25,000, which makes OneWeb’s purported price point of $2508 seem wholly unrealistic. The reality is that the terminal will cost somewhere in between, especially if phased array antennas are not utilised. In this case, the LEO user terminal will be required to have two steerable reflectors which operate in tandem, providing make-before-break communications. Such antennas may be similar to the LEO tracking antenna produced by AVL Technologies, illustrated below in Figure 8. Thus, a GEO system can do what OneWeb proposes to do for around $3,000 per site, with the benefit of decades of heritage and volume manufacture. We find it unlikely in the extreme that a LEO system can do the same thing for $250 even if the blue sky physics of phased array antennas can be made to work. Figure 8: AVL Technologies LEO/MEO Tracking Antenna (85cm) 8 “OneWeb's Plan to Beat SpaceX to Provide Satellite Internet to Everyone on Earth”, J. Koebler: http://motherboard.vice.com/read/onewebs-plan-to-beat-spacex-to-provide-satellite-internet-to-everyone-on-earth LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 22 22 September 2015 4.7 Launch Launchers such as the Soyuz can deploy about 4.5 tonnes to Low Earth Orbit, or about 30 small 150kg satellites. With a price tag of about $70M, this implies launch costs of approximately $2.3M per satellite. Arianespace currently launches about 6 Soyuz per year. To launch 650 satellites would require 22 launches at a cost in excess of $1.5B. This equates to the entire Arianespace Soyuz 5 manifest for 4 years. A more likely scenario is the use of multiple launch sites, but even so, it will still take approximately 2.5 years to launch this many satellites. This is unwelcome news for other operators hoping to launch in that period. One problem that always gains attention from satellite manufacturers is the mass budget for the spacecraft. Whilst most GEO satellites can manage errors in the initial estimation of the satellite mass of 100kg or more, LEO constellations face much more significant impacts if the initial mass estimation is not accurate. For example, if a new satellite system is designed to have a mass of 150kg per satellite, but this increases by 50kg, the overall impact is an increase in launch costs by 33% - or another $0.5 billion in the case above. Of particular note is the fact that for Star constellations it is difficult to launch service without a full constellation in orbit to avoid service outages. When utilising many different planes for the satellite orbits, all planes must be populated to give global coverage. So service commencement is not only dependent on the speed at which satellites can be built but also on the speed at which they can be launched. As noted above, a LEO Star constellation needs to be complete to operate commercially. With LEO satellites designed to last 5 years it leaves 2.5 years for exploitation before satellites begin de-orbiting and must be replaced. This results in an almost permanent CAPEX cycle with inadequate time to commercialise. Given that telcos, especially those in emerging markets, typically take several years to adopt new technologies, test them and then build business cases, it is hard to see how any investor could see a rational case for return on investment, but perhaps there are non-commercial investors willing to consider funding such projects. 4.8 Operational Deployment and Start of Service The constellation launch and deployment procedures are driving factors for the constellation design and start of service. It is vital to understand how a constellation’s performance evolves during the deployment phase. The need to start a revenue stream flowing as early as possible is a driving aspect of any telecommunications system. Unlike GEO systems, LEO constellations do not allow revenue results after the launch of their first satellite, thus partial deployment options are needed. The number of orbital planes relates strongly to a coverage issue often overlooked in constellation design: the need to provide the constellation both performance plateaus and graceful degradation. Constellations with a small number of orbital planes have a distinct advantage over many-plane ones. A single-plane constellation produces LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 23 22 September 2015 performance plateaus with each added satellite, whereas one with two planes would have plateaus at one, two, four, six, eight, (and so on) satellites. Thus, more complex constellations will require more satellites for each performance plateau. Performance plateaus are important because individual satellites are extremely expensive. Verifying if the constellation will meet its design goals without the necessity of launching all the satellites can be an important strategic design decision. A major difference between Delta and Star constellations emerges when dealing with performance plateaus. Unlike Delta constellations for which the performance plateaus concept can be applied, Star constellations must have all satellite or almost all satellites arranged in orbit before continuous service can be provided to customers in mid- latitudes. Star constellation operators are forced to design a rapid deployment campaign in order to have an operational constellation as soon as possible. For a constellation which utilises 20 orbital planes and a total of 650 satellites, if it was a Delta constellation, then 20 satellites would be required to produce each performance plateau, one in each plane. However, in the case of a Star constellation such as OneWeb all of the satellites would need to be deployed to provide continuous coverage over the mid-latitudes. This would make early role out of service to users somewhat sporadic. This is the same case with Iridium in the 1990s where the lack of coverage early in their deployment impacted both their subscriber number growth and resulting revenue. There are no short cuts to deploying an operational LEO constellation. It needs a significant investment to enable the start of service and once deployed it has to be maintained to ensure that the subscriber quality of service is maintained. In sharp contrast the deployment of a GEO satellite enables the immediate start of service. A second satellite ensures continuity of service should an issues arise with the primary spacecraft but service can start as soon as the first satellite is on station shortly after launch. 4.9 Control and De-Orbiting With many hundreds or even thousands of satellites in orbit, controlling them throws up enormous challenges. The collision between an Iridium satellite (which has 66 satellites in its constellation) and a Russian satellite in 2009 demonstrates how hard it is to control the movement of just 66 satellites to avoid in -orbit incidents. The problem is compounded by the fact that the satellites have to fly in a geometry synchronised with nearby satellites in the constellation to ensure uniform coverage of the ground. Moving one satellite may require management of the orbits of many other satellites to keep this geometry intact. Satellites and other objects placed in low-Earth orbit will remain in orbit for many years, depending upon their orbital altitude. Satellites in orbits above about 700 km will stay in orbit for hundreds of years. Recent studies of the interaction of satellite constellations with the space debris environment have concluded that "the debris environment cannot sustain the long-term operation of large constellations”. In order to avoid this situation LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 24 22 September 2015 each satellite in the constellation must be de-orbited while still under ground control – or risk creating a massive debris field in the constellation’s operational orbit. With a very simplified design to reach low cost targets, there is a significant conflict with the requirement to achieve security of control to guarantee compliance with the universal obligation to respect space resources and limit debris. In the case of systems operating at around 1200 Km altitude the satellites would have to be de-orbited at the end of life - i.e. deliberately brought back into the Earth’s atmosphere under control. New systems such as the Gossamer Orbit Lowering Device (GOLD) may offer a solution for satellites orbiting up to 1200km altitude, but this system has never been flown. More importantly, since it requires a suitcase sized container it adds significantly to any small satellites accommodation requirements and mass – again pushing up launch costs. In fact de-orbiting may be the biggest hurdle faced by constellations. Most countries that co-ordinate spectrum applications to the ITU require the issuance of a Space License as a pre-cursor to launch. The Country issuing the license provides a State guarantee for unlimited third party liability. Thus if satellites crash in orbit causing catastrophic debris, or crash to Earth on de-orbit destroying life and property, the Country of issue provides unlimited liability cover. It is difficult to see an established nation willing to provide such unlimited liability in a scenario where the risks are so extreme. It is doubtful that such a risk would meet with tax payer approval. 4.10 In-Orbit Hazards – Radiation and Space Debris Near-Earth environment changes drastically with altitude in LEO orbits. Radiation and atmospheric drag have to be taken into consideration when deciding the altitude of a constellation. Drag and radiation have a huge influence in the analysis of satellite mass and lifetime. If the altitude is too low (i.e. 400 km), drag becomes a major environmental parameter which influences the satellite operational lifetime. Its mass will influence the requirements for Station-Keeping to maintain the desired altitude. However, higher altitudes that are above Earth’s atmosphere, where there is no natural radiation protection the spacecraft would suffer from a harsher radiation environment. The Van Allen belts are normally divided in two different regions; inner and outer belts. The limits of each belt are not well defined and different altitude values can be found depending on the source. As a general guideline, the harsher radiation environment is situated between 600 km and 7000 km. In addition, at low altitudes and high inclinations, the dominant feature of the radiation environment is the region known as the South Atlantic Anomaly (SAA). Because of the offset and tilt of the geomagnetic axis relative to the Earth’s rotation axis, this is a region of enhanced radiation in which parts of the radiation belt are brought to lower altitudes. The shape and particle density of the SAA varies on a diurnal bases, with greatest particle density corresponding roughly to local noon. At an altitude of approximately 500 km, the SAA spans from -50° to 0° latitude and from -90° to +40° longitude. This LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 25 22 September 2015 means that for low altitudes, the only way a constellation can avoid the SAA is having all its satellites in an equatorial plane. Figure 10 below shows a comparison of the radiation environments encountered by typical GEO, MEO and LEO satellites. In this analysis, the satellites are at a respective altitude of 35778 km (typical for GEO), 8061 km (similar to O3b) and 1200 km (OneWeb’s purported altitude). Figure 10 shows the combined average dose rate calculated over a period of 6 hours, where the LEO satellite has been programmed to pass over the SAA. It can be seen that the LEO radiation environment at this altitude is comparable to, if not worse than, the MEO environment, and both are around 10 times worse than the GEO environment. Satellites can use components that are ‘radiation hardened’. These components are designed, created and tested using a process that ensures a certain level of robustness against damage and malfunctions due to radiation. Creating rad-hardened components capable of withstanding the most hostile environments requires extensive testing and development, leading to overall life cycles from design to incorporation within a spacecraft of up to 10 years. As such, heavily rad-hardened components often lag behind the most recent developments, employing older technology. Usually satellite manufacturers and operators choose a compromise, analysing the potential risks to their satellites before selecting components with a suitable level of rad-hardening. As can be seen in Figure 10, LEO manufacturers and operators will either be forced to use decade-old technology to protect against radiation, or subject their spacecraft to excessive levels of radiation which can lead to component malfunction and failure. Figure 9: SAA location LEO: Roar or Whimper Low Earth Orbit Broadband Constellations: Technical and Economic Truths 26 22 September 2015 Figure 10: Combined average dose rate for typical GEO, MEO and LEO satellites vs. aluminium shielding thickness One of the most important characteristics of any constellation is collision avoidance. After a collision the debris cloud created from the satellite fragments would remain in the same orbit being a threat to the rest of the constellation and other spacecraft. The potential for secondary collisions increases, which, in turn, continues to increase the amount of debris and the possibility of making the orbit “uninhabitable”. Intra-plane phasing and separation between same orbital plane satellites are parameters that have to be taken into consideration to avoid colli

LEO Broadband Constellations .pdf

Comments

You must sign in to comment