Antenna aperture size reduction using subbeam concept in multiple spot beam cellular satellite systems

Transcription

1 RADIO SCIENCE, VOL. 44,, doi: /2008rs004052, 2009 Antenna aperture size reduction using subbeam concept in multiple spot beam cellular satellite systems Ozlem Kilic 1 and Amir I. Zaghloul 2,3 Received 22 October 2008; revised 30 January 2009; accepted 25 February 2009; published 6 May [1] Multiple beam configurations in satellite systems enable the reuse of frequency in beams separated by a sufficient distance to address bandwidth limitation issues. Phased array antennas can simultaneously produce a large number of spot beams directed toward specific areas and therefore are a natural platform for multiple-beam satellite communications. This paper discusses a method to generate a large number of spot or cellular beams for multiple-beam satellite systems using smaller apertures. The basic spot beam is divided into a number of smaller subbeams with lower crossover levels that operate within the same bandwidth as the spot beam. The frequency reuse configuration can be maintained, while the overall aperture size is reduced. The aperture reduction is shown to have little or no effect on the cochannel interference. The technique is applicable to all satellite orbits as well as high-altitude platforms. Citation: Kilic, O., and A. I. Zaghloul (2009), Antenna aperture size reduction using subbeam concept in multiple spot beam cellular satellite systems, Radio Sci., 44,, doi: /2008rs Introduction 1 Department of Electrical Engineering, Catholic University of America, Washington, D. C., USA. 2 Bradley Department of Electrical and Computer Engineering, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA. 3 U.S. Army Research Laboratory, Adelphi, Maryland, USA. Copyright 2009 by the American Geophysical Union /09/2008RS [2] Multibeam reflector systems and multiple-beam phased array antennas are used on board satellites to produce a large number of spot beams. The multiple spot beams can be simultaneously generated by a phased array using a microwave beam former that uses phase shifters to apply appropriate phase shifts to each of the radiating elements in the array, or a digital beam former that adds the required phase shifts to the signals in the onboard processor. [3] To completely cover a satellite service area with spot beams, it is customary to use a network of contiguous beams, with the beam contours defined at levels of 3- to 4-dB down from the beam peaks. The phased array antenna therefore needs to be sized to accommodate the peak gains required in order to satisfy the edge gain requirements. This results in a large aperture for the satellite antenna as the aperture area is directly related to the peak gain achieved. This has been demonstrated in a number of mobile satellite systems that use large reflectors or sizable direct radiating phased arrays [Evans, 1998; Miller, 1998]. If a phased array is employed, a large number of radiating elements are needed to populate the aperture, and complex beam formers with high power and mass requirements are used [Schuss et al., 1999]. It was shown that the size of the phased array can be reduced, and consequently simplified, if the array is used as a feed in a near-field Gregorian dual reflector system [Zaghloul and Sorbello, 1984; Schuman and Pflug, 1990]. This has the effect of magnifying the array aperture through the dual reflector system, but still requires a large reflector as the radiating aperture that is sized to produce the narrow spot beam. [4] This paper discusses a method to generate the large number of spot beams for multiple-beam satellite systems by allowing the use of smaller apertures. The technique does not interfere with the frequency reuse scheme in the system or the cochannel interference levels. Rather, it divides the spot beams into subbeams with lower crossover levels, thus allowing a smaller aperture that is commensurate with lower peak gain values. Reducing the antenna aperture size and still maintaining the cellular beam configuration is a desirable feature in the satellite system. However, the aperture reduction should not compromise the low cochannel interference that is a key for the frequency reuse performance in the cellular spot beam system (O. Kilic et al., Beam interference modeling in cellular satellite communication systems, paper presented at General Assembly, 1of9

2 Figure 1. Frequency reuse for multiple spot beam coverage for reuse index equal to 7. International Union of Radio Science, Toronto, Ontario, Canada, August 1999). [5] Although the concept presented in this paper is applied to cellular satellite systems, it can also be applied to high-altitude platforms and terrestrial wireless systems in order to reduce the antenna size at the base stations. Cell splitting is known in terrestrial systems, but it uses higher crossover levels between the microcells and results in an increase in the antenna aperture. Such microcell zone concept [Lee, 1991] is generally used to increase the cell capacity, and results in a more efficient system. 2. Conventional Spot (Cellular) Beam System [6] Owing to the limitations on the available bandwidth, mobile satellite communication systems often reuse the same frequency in spot beams that are sufficiently separated from each other, so that interference is minimal. In a frequency reuse scheme, the total available bandwidth is allocated to a small number of cells. These cells form a cluster, which are then repeated in the coverage area. It is conventional to represent each spot beam by a hexagon as shown in Figure 1. There are only a discrete set of possible cluster sizes to accommodate a contiguous coverage. The number of spot beams in a hexagonal cluster satisfies the following equation: N ¼ i 2 þ j 2 þ ij ð1þ where N is the possible number of beams in a cluster and i and j are nonnegative integers. Only certain cluster sizes (i.e., N = 1, 3, 4, 7, 9,...) can tessellate; i.e., form a contiguous coverage when repeated as described by Macdonald [1979]. A coverage area with a frequency reuse index, or cluster size, of seven corresponding to i = 2, j = 1, is shown in Figure 1. The spot beams are represented by hexagons which are circumscribed by the circular contour of a spot beam. The numbers 1 to 7 denote beams that use different frequency bands. The clusters are circumscribed with the larger circles that represent the total bandwidth reuse area The antenna aperture size is a function of the required peak gain in the spot beam, which in turn depends on the required edge gain and the beam contour level between the beams that is conventionally set to be 3 4 db. [7] Because of the large number of spot beams (cells), digital beam formers (DBF) have become the beam former of choice. While the number of inputs to the DBF corresponds to the large number of array elements, the number of outputs corresponds to the number of spot beams, or cells. The demultiplexing of composite signals and the assignments to the spot beam ports are performed within the DBF processor. The complexity of the DBF and the payload in general is a function primarily of the number of elements in the transmitting and receiving antenna arrays, and to a lesser extent on the number of spot beams. 3. Subbeam Concept [8] The subbeam design reduces the peak gain requirements at the beam center while maintaining the edge gain requirements. In order to do this, each spot beam in the coverage area is divided into a number of smaller beams defined by contour levels less than the typical 3 db. A group of these smaller subbeams, i.e., subbeam clusters, represent each spot beam. To satisfy the contiguity requirements of the coverage, the subbeams intersect at the boundaries of the spot beam. Figure 2 demonstrates the subbeam concept, where the hexagons correspond to Figure 2. Division of a spot beam into subbeams, frequency reuse index, No = 7, subbeam cluster size, Ns = 4. 2of9

3 Figure 3. Various subbeam configurations. the spot beams in the coverage area. The numbers at the center of each hexagon denote the frequency band assigned in each beam. A frequency reuse index (or spot beam cluster size) of seven is depicted in Figure 2. The large circles encompassing sets of seven hexagons represent the spot beam clusters based on this reuse factor. The subbeams are shown for the center spot beam of each cluster. A subbeam cluster size of four is assumed in the demonstration. [9] As a result of this approach, the antenna is required to generate four times more beams than the conventional spot beam design. However, the subbeams do not need to satisfy the peak gain requirements of the spot beam. Since only the edge gain needs to be satisfied, the subbeam peak gain is lower and the overall antenna size is reduced as explained in the following sections Spectrum Subdivision With the Subbeams [10] The frequency reuse configuration is conserved with the subbeam approach. The allocated bandwidth per spot beam is shared among the subbeams in a subbeam cluster. As Figure 2 suggests, each subbeam in a subbeam cluster uses the bandwidth allocated to the spot beam they represent. While each subbeam is designated a smaller portion of the bandwidth, the bandwidth allocation for the entire spot beam remains the same. Hence, the capacity per spot beam is conserved. Furthermore, this approach makes it possible to have a nonuniform distribution of the available bandwidth in a spot beam to address local high-traffic areas. Depending on the subbeam cluster size, some areas are served by more than one subbeam from different spot beams, adding to the flexibility of traffic assignment Gain Reduction Concept of the Subbeam Approach [11] The amount of reduction in the antenna size depends on the edge gain relative to the peak, and beam width requirements of the coverage. The subbeam cluster generates an equivalent coverage to the spot beam by using a higher number of smaller beams. This enables using smaller antennas. The number of beams in a subbeam cluster, N s needs to be chosen such that a contiguous coverage is achieved. This results in values such as N s =1,3,4,7... etc. [12] The relationship between the relative edge gain of the spot beam and the relative edge gain of the subbeams can be determined as a function of the number of subbeams that make up a subbeam cluster. In the derivations that follow, the subscripts o and s denote the spot beam and subbeams, respectively. The subbeam edge contours are defined by x s db down relative to the peak gain of G s db. The spot beam contours are assumed to be x o db lower than the peak gain, which is denoted by G o db. Since the edge gain requirements are satisfied for both cases, the gain relationship can be written as follows: G o x o ¼ G s x s ð2þ [13] This equation can be rewritten in terms of gain reduction as: DG G o G s ¼ x o x s ð3þ [14] It is observed that the reduction in gain is directly proportional to the difference of the beam contour levels of the spot beam and the subbeams. While the typical contour level to define a spot beam is 3 4 db, the subbeam configuration is flexible, and the choice is based on increasing the gain reduction. As equation (3) suggests, defining the subbeams at a low level would increase the reduction in gain. However, the subbeam cluster size, N s, constrains the beam width of the subbeams, and the value of x s cannot be assigned arbitrarily. The choice will depend on the edge gain of the spot beam as well as N s Defining the Subbeams to Achieve Desired Gain Reduction [15] A contiguous coverage of the subbeam clusters can be obtained by satisfying equation (1), in which case the number of subbeams that define a spot beam are restricted to values such as 1, 3, 4, 7,..., etc. Different cluster arrangements that do not satisfy equation (1) may also be utilized. Examples of the geometry for different cluster sizes are shown in Figure 3. 3of9

4 Figure 4. Beam width at arbitrary contour levels for large arrays. [16] In Figure 3, the circle at the center corresponds to the spot beam, while the surrounding circles denote the subbeam contours. The beam width ratio, q s /q o depends on the subbeam cluster geometry, and is equal to 1, 1, and 0.5 for subbeam cluster sizes of 1, 3, 4 and 7, respectively. The higher subbeam clusters approximate the spot beam better and reduce the overlap of subbeams from different clusters. Higher cluster sizes also reduce the subbeam peak gain that is needed to satisfy the spot beam edge gain requirement. [17] An empirical relationship has been obtained to relate the beam width of a phased array at an arbitrary contour level, x to its HPBW by using the definition of the array factor as: q X q 3 ¼ 0:59x 0:4806 ð4þ where x is in dbs [Zaghloul et al., 2000]. This empirical relationship assumes isotropic radiating elements, which combine in space to form the directive pattern, and has been verified for large array sizes for broadside radiation. The expression is determined by fitting a curve to different size square arrays that range from to elements with half wavelength spacing between them. The results of the simulations and the agreement with the empirical relation above are shown in Figure 4. [18] Using equations (4) and (2), the contour levels of the spot and subbeams can be related to their beam widths, q s and q o as follows: 9:612 log x s þ x o x s ¼ 20 log q s ð5þ x o q o Equation (5) can be rewritten in terms of gain reduction, using (5) as 9:612 log x s þ DG ¼ 20 log q s ð6þ x o q o where DG = G o G s is the gain reduction achieved by using the subbeam concept and the beam width ratio on the right is a constant determined by the subbeam cluster configuration. Thus, for a given subbeam configuration, the peak gain reduction and the ratio of the relative edge gains are uniquely related. In other words, if two of the parameters DG, x o and x s are given, the third is uniquely determined for a specific subbeam cluster configuration. [19] The relation between reduction in gain and beam contour levels is demonstrated in Figure 5. Two cases are Figure 5. Contour levels of spot beams and subbeams as a function of gain reduction. 4of9

5 Figure 6. Number of elements in the reduced aperture array versus subbeam cluster size. plotted for comparison. The blue line corresponds to the subbeam cluster size of 4, while the red line represents the subbeam cluster size of 7. The dashed lines show the contour level for the subbeams as a function of gain reduction, and the solid line describes the relation between spot beam contour levels and the gain reduction. It is observed that, for a desired amount of gain reduction, the beam contour levels for both subbeam and spot beam are uniquely defined for each subbeam cluster size. [20] Since the subbeam approach is offered as a solution to reduce the antenna aperture size, it is worth investigating the improvement by assuming that the spot beam contour levels and the peak gain are already defined for the system to be replaced. For instance, if a spot beam width at 4 db contour levels is being replaced with subbeams of lower contour levels, Figure 5 suggests that a gain reduction of 3dB can be achieved for N s =4 by defining the subbeam contours at 1 db down from the subbeam peak gain. Similarly, for N s = 7, a gain reduction of 3.6 db is achieved by defining the subbeams at 0.4 db down from their peak gain. [21] This example implies that for a given spot beam configuration, the larger subbeam clusters can offer higher gain reduction, which is associated with a corresponding reduction in the antenna aperture size and cost. However, since the higher cluster sizes define beams at lower contour levels, a legitimate concern is to make sure the energy in the main beam of these antennas do not adversely affect the cochannel interference levels. An investigation is made in section 5 to show that this approach, does not necessarily adversely affect the cochannel interference. [22] The reduction in peak gain results in a reduction in the aperture size and correspondingly a reduction in the number of elements in the array. This is illustrated in Figure 6 where a or 784-element array for a conventional spot beam (corresponds to Ns = 1) is replaced with arrays of fewer elements as the number of subbeams increase in the cluster. It is observed that the limit case is approached for subbeam cluster sizes of 7, where less than 300 elements are required. [23] The reduction in the receive and transmit antenna aperture sizes result in fewer components. However, the generation of subbeam clusters requires further demultiplexing and increase in the size of the digital traffic router. Corresponding multiplexing on the transmit side is also required. However, the added complexity occurs in the software-driven processor on board the satellite, with minimal addition to the hardware. 4. Example [24] To demonstrate the advantages of the subbeam approach, a low Earth orbiting satellite system is considered. The altitude of the satellite is 9400 km and the view angle for the coverage area from the satellite is ±30 degrees. A peak gain of 40 dbi is required for the spot beams, which are defined at 4 db relative to the peak contour levels; i.e., 36 dbi edge gain. The frequency of operation is 1.9 GHz. [25] In order to achieve this performance with the standard approach, a element phased array with circular horns of 8 db peak gain would be needed. This assumes radiating element separation of half wavelength, which corresponds to an aperture size of 6 6 = 36.0 m 2. The broadside radiation pattern of this array is shown in Figure 7. The HPBW for the beams generated by this array is 1.5 degrees, and the spot beam size, defined at 4 db, is 1.8 degrees. A total number of 170 beams need to be generated to illuminate the coverage area. [26] If the same system is designed using the subbeam concept, the array size can be reduced. Since the spot 5of9

6 Figure 7. Broadside radiation pattern of the conventional array for spot beam (28 28 elements). beam contour is set at 4 db, the gain reduction that the subbeam can achieve is determined from Figure 8 for different subbeam cluster sizes. For Ns = 7, the gain reduction achieved for x o = 4 db is 3.6 db. This means that the subbeam array is required to generate a peak gain of 36.4 db and the edge gain requirement of 36 dbi is met with x s = 0.4. An element array satisfies the peak gain requirement with element spacing of half wavelength, corresponding to aperture size of m = m 2. The broadside radiation pattern of the subbeam array, assuming isotropic radiating elements is shown in Figure 8. The resulting antenna size corresponds to a 60% reduction in the number of radiating elements and aperture area. However, = 1190 beams will have to be generated instead of 170 beams with this method. Since the payload complexity depends on the number of radiating elements, as indicated above, this compromise is advantageous. 5. Interference Analysis [27] The subbeam array example shown above was designed based on Ns = 7 to achieve a relatively higher gain reduction and smaller antenna aperture. While higher subbeam clusters improve gain reduction, two effects of this choice need to be considered in the design: Figure 8. Broadside radiation pattern for the subbeam array. 6of9

7 Figure 9. Contour plot of subbeam pattern and cochannel beams. (1) the number of beams in the system is increased, and (2) since lower gain beams are generated, they tend to be broader. The first causes some complexity in the DBF design that can be tolerated. However, the latter can impact the cochannel interference levels among beams and subbeams that use the same frequency segment. [28] The effects of the broader subbeams on interference are studied in this section. Figure 9 shows the contour plots overlaid with the cochannel beam locations and tiers for spot and subbeam configuration for Ns = 4 for frequency reuse number (No) of 3. The white circles in Figure 9 (left) denote the cochannel spot beams, while in Figure 9 (right) they correspond to the cochannel subbeams. The blue circle to the lower right of the center in the subbeam plot demonstrates the spot beam which is being replaced by the subbeams. Only one of the subbeams (shown in white) and its cochannel beams are shown for legibility. The potential interference from the center beam into the cochannel beams can be visualized in Figure 9. The broader subbeam pattern has the potential to contribute to interference at higher levels as a high sidelobe can coincide with multiple cochannel beams. However, it is difficult to conclude what the overall effects would be simply by observation. [29] A code has been developed to automate the interference calculations for the subbeam and spot beam coverage on Earth for different reuse cases. The performances in terms of the carrier to interference ratio, C/I and interference are shown in Figure 10, where the x axis denote the frequency reuse number. The different color bars correspond to spot beam (Ns = 1), and subbeams for Ns = 4 and Ns = 7 cases. The C/I at a point is computed as the ratio of the antenna directivity for the beam that contains the observation point to the total sidelobe energy received at that point by the cochannel beams. Figure 10. (a) C/I comparison and (b) interference comparison. 7of9

8 Figure 11. C/I performance comparison within the spot beam. All power levels from the cochannel beams are added to calculate the total interference at a point. [30] It is observed in Figure 10a that the C/I at the center of the beam (where carrier power is max) is better for the spot beam configuration for large cluster sizes (i.e., N > 9). However, in certain cases such as No = 3 and No = 9, the performance of the subbeam approach is better or comparable to the spot beam configuration. It should be noted that this comparison presents a best case scenario for the spot beams as the carrier power is at its peak value. The total interference shown in Figure 10b indicates that subbeams may have lower interference values relative to the spot beam cases of the same frequency reuse number. A large cluster size separates cochannel cells, and the highly directive spot beam antenna tends to have lower sidelobe levels at these separations. [31] To better understand the subbeam performance, a better analysis is calculating C/I as one moves within the spot beam to be replaced by the subbeams as depicted in Figure 11. The spot beam shown at the center with the blue circle is replaced by 4 subbeams. The C/I is calculated at discrete distances from the center (point d) and at angular positions (point a). Figure 11 (right) shows the C/I for the spot beams (blue lines) and subbeams (red lines) for different d and a values. The x axis shows the ratio of the distance from the center to the spot beam radius, R. It is observed that partitioning the spot beam into subbeams performs superior toward the spot beam edge because the carrier power in the subbeam gets stronger. This is typically where the spot beam performs worst and a significant advantage for the subbeam approach results. The main region where the spot beam seems to perform better is at the center of the beam as expected due to the high carrier power levels in the spot beam. In the regions near the beam center, it is observed that the subbeam approach performs worst where the spot beams have the best performance. As expected, this is due to the high carrier power the spot beam has at the center. Typically the C/I value at the beam center has a higher margin as the systems are designed for the edge gain and the degraded performance of the subbeam at the center may not pose a problem. To mitigate the relatively poor performance at the center, it would be possible to optimize the subbeam antennas to perform better at the beam centers. If necessary a further step could be to modify the frequency allocation. However, this could result in lower system capacity, and should be attempted only if the antenna size and cost is the primary concern. Overall, the findings suggest that the subbeam approach holds promise as its C/I is mostly not degraded compared to the spot beam approach, while its peak gain is lower and thus requires smaller antenna aperture. 6. Conclusions [32] A method was presented to reduce the array antenna aperture on board the satellite for multiple spot beam cellular coverage. The overlapped spot beams have edge gain that is determined based on link budget, coverage requirement, frequency band allocations, and corresponding channel capacity allocations. The edge gain relative to the peak gain of the spot beam determines the size of the required antenna aperture. Dividing the spot beam into a number of subbeams that overlap at the same gain level as the spot beam but with lower gain taper over the subbeam results in lower peak gain and 8of9

9 consequently smaller aperture. The frequency band allocated to the spot beam is divided among the constituent subbeams and can be assigned with enough flexibility to meet projected traffic demands. The reduction in the antenna aperture translates into significant reductions in number of array elements, RF components, and A/D and D/A converters. These savings are obtained at the expense of more complex digital beam forming and traffic routing algorithms. Analysis has shown that in spite of the smaller aperture and the broader beams of the subbeams, the cochannel interference between subbeams using the same frequency segment is not adversely affected. In fact, the subbeam approach can perform better where the spot beams have the worst performance. Although presented in this paper for satellite-based transponders, the subbeam concept can be applied to wireless communication systems where interference scenarios may be more favorable due to the larger differences in the path lengths between the desired and interfering signals. References Evans, J. V. (1998), Satellite systems for personal communications, Proc. IEEE, 86(7), Lee, W. C. Y. (1991), Smaller cells for greater performance, IEEE Commun. Mag., 29(11), Macdonald, V. H. (1979), The cellular concept, Bell Syst. Tech. J., 58(1), Miller, B. (1998), Satellites free the mobile phone, IEEE Spectrum, 35(3), Schuman, H. K., and D. R. Pflug (1990), A phased array feed, dual offset reflector antenna for testing array compensation techniques, in Proceedings of International Symposium on Antennas and Propagation, Dallas, Texas, May, pp , IEEE Press, Piscataway, N. J. Schuss, J. J., et al. (1999), The IRIDIUM main mission antenna concept, IEEE Trans. Antennas Propag., 47(3), Zaghloul, A., and R. Sorbello (1984), 20-GHz phased array fed reflector antenna with distributed MMIC modules, in Proceedings of International Symposium on Antennas and Propagation, Boston, Massachusetts, June, pp , IEEE Press, Piscataway, N. J. Zaghloul, A. I., O. Kilic, and A. E. Williams (2000), Aperture size reduction using sub-beam design concepts for multiple spot beam satellites, in Proceedings of International Symposium on Antennas and Propagation, Salt Lake City, Utah, July, pp , IEEE Press, Piscataway, N. J. O. Kilic, Department of Electrical Engineering, Catholic University of America, Washington, DC 20064, USA. (kilic@cua.edu) A. I. Zaghloul, Bradley Department of Electrical and Computer Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA. 9of9