The Influence of Phased-array Antenna Systems on LEO Satellite Constellations

The Influence of Phased-array Antenna Systems on LEO Satellite Constellations

In this era of low earth orbit (LEO) satellite constellations, the phased array has become the antenna implementation of choice. Previously thought to be an expensive solution for satellite systems, the requirement to provide multiple beams makes the phased-array antenna implementation a modest cost on a

per-beam basis. Selecting a phased-array antenna for the satellite system based on cost/performance influences the satellite and constellation design in a manner that is not so obvious. This influence is primarily a function of scan loss (reduced performance at angles off boresight), which is not a significant factor in more limited-scan, parabolic-reflector antenna systems. Antenna system cost and performance (as influenced by scan loss) are significantly affected by satellite and constellation parameters, including the antenna system boresight, the preferred altitude of the satellite constellation and, lastly, the optimum number of satellites in the constellation. Propagation effects due to rain and, consequently, the allocated frequency band, further enhance the effects of scan loss and compound this influence on constellation size. This influence is most significant at K-band and above.

Paul Chiavacci
Raytheon Systems Co.
Sudbury, MA

The era of phased-array antennas in space has developed very quickly. Phased-array antennas are the antennas of choice for the LEO constellations for the mobile satellite services (MSS) currently being deployed, including Iridium, Globalstar and ICO-Global. (ICO-Global is a medium earth orbit constellation but is usually categorized with the so-called "Big LEO" MSS systems). In addition, these antennas are the preferred configuration for the wideband (Ka-band) data systems presently in the design stage for fixed-satellite services, such as Teledesic and SkyBridge. The interest in phased arrays for these systems is driven by their flexibility and ability to provide multiple simultaneous beams. (The discussion in this article does not distinguish between a multibeam array - an array of phased elements provided by a fixed beamformer typically utilizing a Butler matrix, Rotman lens or other technique for phasing arrays of elements - and a true phased-array antenna provided by a phase shifter per element per beam.)

If a single beam or a few beams are required (as is the case for the present geostationary earth orbit satellites), this requirement is inexpensively and efficiently handled using dish antennas with one to a few feeds. When tens or hundreds of beams are required, dish antennas are no longer an efficient and cost-effective means of implementation.

The current LEO constellations require many tens of beams per satellite for MSS and hundreds of beams per satellite for the wideband data systems. Because of their LEO orbit, these antennas must scan over very wide angles (±60° from nadir is not unusual). This scan requirement as well as the multiple beams mitigate for the phased array, thus making it the only practical solution.

A 1000-element phased array is capable of providing hundreds of beams within the wide scan coverage requirements. If an antenna cost based on a prorated cost-per-element of $1000 is assumed, a 1000-element phased array would cost on the order of $1 M. (Prorated element costs found in the literature vary from a few hundred to a few thousand dollars^1,2 ). Contrast this expense with the cost of a 100-beam, space-qualified dish antenna system. A space-qualified traveling-wave-tube amplifier feed combination will cost upwards of $100 K, and providing the required scan coverage would require multiple dish antenna structures. If it was practical to put the size and weight of such a package in space, a conservative cost of this configuration would be in excess of $10 M. Therefore, size, weight, cost, flexibility and performance (reliability, fail soft) mitigate strongly in favor of phased-array antenna implementation for LEO satellite constellations. The remainder of this article addresses some of the considerations that influence the LEO satellite constellation when a phased array is the antenna implementation of choice.

Array Boresight Determination

The principle criterion for determining the boresight (direction normal to the array face) of the array is cost. Reducing array costs while maintaining performance is paramount. For a LEO satellite constellation, the coverage area is a revolutionary cone about the nadir direction (the direction from the satellite to the earth's center). This area is most cost-effectively covered utilizing a nadir-pointing array. Although there are advantages to having the array boresight at the outer perimeter of the satellite's coverage area, the multiple

arrays required to implement this configuration are expensive. (The cost advantage for a nadir-pointing phased-array antenna has been identified by E. Brookner.³ )

For example, the scan loss (the inherent loss in gain of a phased-array antenna for scan angles removed from the antenna boresight) of an array increases as the cosine of the scan angle. A typical formulation used in practice is (cosine J)k, where J is the scan angle off boresight and the exponent k is a number between 1.2 and 1.5.⁴ Using this formulation, a nadir-directed array is penalized when it scans out to the limits of coverage relative to a multi-aperture antenna with each aperture boresighted to the perimeter of satellite coverage. In practice, this penalty is not as extreme as might be expected. Although (cosine J)k at a 60° scan with k = 1.3 produces a 3.9 dB scan loss, the maximum scan loss (to the corner of each array's coverage) for a three-aperture antenna with each aperture (array) directed optimally at the outer edge of coverage is 2.3 dB.

For an antenna comprising four arrays, the maximum scan loss (to the corners of coverage) is 1.3 dB. This result is listed in Table 1 along with the normalized element count (normalized to maintain the defined link performance at the limits of coverage) for each configuration. As can be seen, more than twice as many array elements and corresponding transmit and receive modules are required to provide a three- or four-array configuration with the same performance as a single nadir-pointing array. Since the cost of a phased-array antenna is proportional to the element count, a multi-aperture antenna is a more expensive implementation than a single-aperture solution for the large scan angles required from LEO satellites.

Phased-Array Influence on Constellation Altitude

Having identified a nadir-pointing array as a cost-effective solution for satellite antenna implementation, a most favorable altitude of operation next must be determined. Two factors that influence the power density (power per unit area) that the phased-array antenna can deliver to a given location on the earth are range and scan angle to the specified location. A scan loss will be incurred at scan angles off nadir for a nadir-pointing array for reasons discussed previously. For any given location in the satellite's coverage area, increasing altitude will result in increased range loss but, conversely, increasing altitude will result in reduced scan loss to all locations in the coverage area with the singular exception of the nadir point. For those points on the outer perimeter of the coverage area, this reduced scan loss can dominate the increased range loss to that point, resulting in a lower overall transmission loss and improved performance.

The outer perimeter of the coverage area is of most interest since it is at that location (maximum range) that the range loss is greatest and performance is most critical. This point is also where the link budget reaches its limit. Minimizing total transmission loss (range loss plus scan loss) at that point provides for the most cost-effective system implementation.

Figure 1 defines the transmission loss to the edge of coverage for a 10-satellite constellation that provides for double coverage at the equatorial plane. In other words, on average, a user at the equator is in view of two satellites. Coverage will improve at higher latitudes such that more than two satellites (on average) will be in view. Double coverage is a desirable and often-designed-to condition to allow for satellite beam failures as well as the potential blockage of the line of site to a single satellite.

The data plot displays the total transmission loss and minimum grazing angle (the angle between the horizon and line of sight to the satellite) as a function of satellite altitude for a user on the perimeter of coverage for a satellite constellation with 10 space vehicles (SV) per orbital plane. As can be seen, transmission loss is minimized at an altitude of 1500 km for a constellation providing double coverage with 10 satellites per orbital plane. Above that altitude, the increase in range loss at increasing altitudes dominates the reduction in scan loss for a constant coverage area on the earth's surface. Below that altitude, the increase in scan loss at lowering altitudes required to support edge-of-coverage service dominates the reduction in range loss.

The minimum transmission loss shown in the data is a broad minimum -- less than 0.1 dB over the 1250 to 1800 km range. Therefore, significant performance is not sacrificed for small changes in constellation altitude. Under such conditions, higher altitudes are favored to achieve higher minimum grazing angles.

The data also reveal that although the transmission loss varies by less than 0.1 dB, the grazing angle changes appreciably - from approximately 17° at 1250 km altitude to over 24° at 1800 km altitude. Higher grazing angles are desirable to allow the ground-based user a line of sight over near-in obstructions. Higher grazing angles also reduce multipath, which is the undesirable phenomenon in communication systems that produces time dispersion of the received signal and degradation of performance. However, the major problem in operating LEO satellites above 1000 km is the rapidly increasing radiation environment above this altitude.

Figure 2 shows a similar plot of total transmission loss, as a function of altitude, for a 15-satellite-per-orbit constellation. In this case, the minimum transmission loss occurs at approximately 1100 km. Again, a broad transmission-loss minimum is evident from a 950 to 1300 km altitude, while the grazing angle changes from 24° to 31° over this range.

A similar plot of transmission loss vs. altitude for a 15-satellite-per-orbit constellation but with a coverage requirement of 2.5 satellites in view (on average) from a point in the coverage area is shown in Figure 3 . In this case, the minimum transmission loss occurs at approximately 1200 km altitude, but changes slowly (< 0.1 dB) between 1050 and 1400 km. Minimum transmission loss occurs in this general altitude region (1100 to 1500 km) for other typical conditions of satellite orbital densities and multiple coverage conditions.

This altitude regime is also an appropriate maximum altitude for radiation conditions. A typical satellite in the altitude regime about 1300 km will experience a radiation environment of 10 kilorads (Si) per year with 100 mils of aluminum shielding. Radiation conditions above this altitude regime require additional shielding (representing additional weight) or radiation-hardened components (representing additional cost). A constellation altitude in the range of 1100 to 1500 km provides an optimum altitude for active phased-array antennas for both of these reasons.

It should be noted that the technology of choice for active components in a spaced-based phased-array antenna is GaAs. GaAs is necessary to provide efficient components (for example, amplifiers and phase shifters) at the required 12 to 30 GHz frequency regime of these satellite systems. Although GaAs is an inherently radiation-tolerant material, the computer and control system needed for the array as well as application-specific ICs used for phase shifter control and the like at the element level are primarily silicon devices. These silicon devices limit the radiation tolerance of the phased-array antenna subsystem.

Phased-array Antenna Influence on Constellation Size

Now that the orientation and optimum operational altitude regime for a phased-array antenna satellite system have been defined, it is interesting to investigate how these parameters influence constellation size. To do so, it is necessary to define the coverage area of interest for the satellite constellation and the coverage density that is desired. The coverage density is defined in this analysis as the minimum number of satellites, on average, that would be in view of a user on the ground.

Given this constant coverage area and defined satellite altitude, the number of satellites in the constellation is varied and the performance of these different-sized constellations is analyzed. This analysis is accomplished by comparing the link margin for the various constellations at the perimeter of the individual satellite footprint. This case is the most stressful.

The analysis allocates a satellite coverage area that is inversely proportional to the number of satellites in the constellation. For the case of double coverage, each satellite covers an area equal to 2/N of the total constellation coverage area where N is the number of satellites in the constellation. No attempt to define the type of orbit (for example, polar, Walker, etc.) is necessary to approach this condition; that exercise is left for the constellation designers.

The performance at the limit of satellite coverage (satellite coverage perimeter) is influenced by three parameters: range loss, scan loss and rain loss. In general, these parameters will improve at the edge of satellite coverage (losses reduced) as the number of satellites in the constellation increases.

An analysis based on this approach has been performed for the conditions of earth coverage between 60° north and south latitude for a constellation providing double coverage in this area. The altitude chosen for the constellation was 1200 km. These parameters are representative of many of the LEO constellations being proposed in Federal Communications Commission (FCC) filings.

Figure 4 shows the results of link improvement vs. constellation size for a Ka-band (20 GHz) system operating at 1200 km altitude. (The actual FCC allocation for LEO constellations operating in the Ka-band is 18.8 to 19.3 GHz for space-to-earth transmission.) The results have been normalized to a 45-satellite constellation and link improvement is plotted for both rain and clear conditions.

For a satellite-user link in the clear (without rain attenuation), such as the one in the example, performance improves modestly in a monotonic fashion as the number of satellites increases. To achieve a 3 dB improvement in performance (edge of coverage link margin), the satellite constellation must increase in size from 75 satellites to approximately 150 satellites. In other words, doubling the size and cost of the constellation produces a 3 dB performance improvement.

When rain attenuation is included, a size (cost) vs. performance trade-off becomes more apparent. A distinct knee exists in the curve between a 75- and 150-satellite constellation. Below a 75-satellite constellation, each additional satellite's performance increases significantly; above a 150-satellite constellation, each additional satellite's performance improves by a significantly lesser amount. Increasing the satellite constellation from 75 to 90 satellites produces an increase in performance in excess of 2 dB in rain conditions. An increase of 15 satellites to a larger satellite constellation (from 135 to 150 satellites) produces only a 1 dB increase in performance. As the constellation size continues to grow, a greater increment in the number of satellites is required to produce a 1 dB increase in performance.

At this point in the design, the communication system designer must answer the question "How much is 1 dB of performance worth?" The answer to this question is a complex and somewhat subjective trade-off of investor tolerance for increased investment vs. the potential future capacity and revenue stream that improved performance (under the assumed conditions) will produce. With the initial investment for such satellite communication systems in the tens of billions of dollars (Ka-band LEO systems), the answer to this question is not for the faint of heart.

Conclusion

A phased-array antenna system may be the most cost-effective antenna system implementation for satellite systems that require multiple simultaneous beams. This statement is true for the recently developed "Big LEO" satellite constellations such as Iridium, Globalstar and ICO-Global. It is also true for the wideband data communication systems currently on the drawing boards, such as Teledesic and SkyBridge.

For a specified requirement (providing the required link margin at all points in the coverage area), the most cost-effective phased-array antenna system implementation over all scan angles within a reasonable satellite coverage area is a nadir-pointing aperture. Multiple apertures (arrays) can provide modestly better performance but at significantly higher cost.

Due to the effects of scan loss at angles removed from the array antenna's boresight, the antenna system achieves its optimum performance (minimum loss at the limits of coverage) at satellite altitudes in the regime of 1100 to 1500 km. This altitude is also the one at which the radiation environment becomes a serious concern for off-the-shelf radiation hard components. Above this altitude, additional shielding or special-purpose radiation hard components are required. Both approaches have significant cost impact.

Lastly, satellite constellations providing multiple coverage at the desired altitude for optimum performance experience rapidly improving performance with increasing numbers of satellites for small constellations. As the satellite constellation becomes larger, the improvement increment for the addition of a single satellite is reduced significantly. This effect is more pronounced at higher frequencies (Ka-band) in rain conditions. A knee in the performance curve exists in the area of 90 to 135 satellites per constellation; performance with fewer satellites is appreciably degraded and performance improvement with additional satellites is not significantly improved.

Of course, the cost of the satellite constellation is influenced strongly by the number of satellites in the constellation, and the future benefits to the consumer must be traded vs. the cost to the investors. At this time, it is necessary for the system designer, in conjunction with the investors, to ask, "How much is 1 dB of future performance worth?" 

References

1.         Elliot D. Cohen, "Trends in the Development of MMICs and Packages for Active Electronically Scanned Arrays (AESAs)," Proceedings of the 1996 IEEE International Symposium on Phased Array Systems and Technology, Boston, MA, October 18-25, 1996.

2.         J.B.L. Roa, G.V. Trunk and D.P. Patel, "Two Low-cost Phased Arrays," Proceedings of the 1996 IEEE International Symposium on Phased Array Systems and Technology, Boston, MA, October 18-25, 1996.

3.         E. Brookner "Nadir vs. Triad Antenna System" (internal memorandum) (EB: 95:57), Raytheon System Co., Sudbury MA.

4.         E. Brookner, internal communication.

Paul Chiavacci received his BSEE and MSEE from  Pennsylvania State University and has over 20 years of experience in the design and development of communication and radar systems. The majority of his career has been with Raytheon Co. where he now performs as chief engineer for Space Systems. In this capacity, he is responsible for the future direction of Raytheon's involvement in space-based communication system antennas. Prior to this assignment, Chiavacci was deputy program manager for the IRIDIUM Main Mission Antennas. He is a senior member of IEEE and AIAA and a member of Eta Kappa Nu and Sigma Xi.