Progress In Electromagnetics Research, PIER 67, 263–274, 2007


M. Dessouky, H. Sharshar, and Y. Albagory
Department of Electronics
and Electrical Communications Engineering
Faculty of Electronic Engineering
Menouf, Egypt

Abstract—Cellular communications using high altitude platforms will
predominate the existing conventional terrestrial or satellite systems.
Radio coverage of HAPs is an important issue that affect the cellular
system design, therefore a deep analysis of the HAP cell footprint
will be presented in two coverage models and the cell parameters are
also deduced for subsequent cellular design. In the design stage, a
cell generating algorithm is introduced to define their locations more
accurate. When deploying the cellular system with such algorithm, the
cells have proper overlap and interspacing.


High altitude platforms (HAPs) has been developed wherein a
large scale airship is positioned at a predetermined altitude
approximately 20 km high in the stratosphere where it can be used for
telecommunications, broadcasting and environmental measurements.
Using HAPs in mobile communications is a promising scheme because
it takes the advantages of using satellites but at lower altitudes [1–
6]. On the other hand the conventional terrestrial system suffers from
coverage problems which are mostly eliminated using HAPs. In the
terrestrial cellular structure, the geometry of the formed cells can be
determined by considering its hexagonal shape and it is easy to find
the other cells locations, but in the case of using HAPs, the cell will
be defined by the half power contour on the ground which can be
considered as an ellipse. In dealing with the footprint analysis, the
cell parameters must be determined such as the major and minor axes
and there variations with the utilized antenna beamwidths and beam
direction [3–6]. In [3], the coverage analysis is well defined based on an

or any other suitable antenna that gives the desired directional pattern. Directional antennas may be in the form of parabolic reflectors. the flat ground approximation model is presented.. Denoting the distance EF as bF which is given by      Bθ Bθ bF = h tan θo + − tan θo − (1) 2 2 where the subscript F stands for flat ground approximation and h is the platform altitude in km. two coverage analysis models are presented and compared. this assumption will not expect the proper cell parameters for those cells at the coverage edge (i. horn antennas. Section 4 compares between the two coverage models and finally Section 5 concludes 2.e. and Albagory assumption that the earth surface is flat as the platform height is much smaller than the earth radius. 1. cells of lower elevation angles). is located by an angle from the platform given by     −1 Bθ bF θc = tan tan θo − + (2) 2 2h . In this figure. and the projection of the beam on the ground will be an ellipse that has a major axis EF and minor axis HK. The cell as depicted in Fig. but on the other hand. As the HAP station is located at an altitude about 20 km high. in Section 2. which is very small compared with the earth’s radius. 8] which has a widespread use. 2 is defined by the coverage beam that has a direction of θo and cross section beamwidth of Bθ and Bφ . Ground cells also can be formed by directing a beam using phased arrays [7. C. The cell center point. Therefore in this paper. while in Section 3 the curved ground model is introduced. FLAT EARTH APPROXIMATION FOR HAP RADIO COVERAGE The HAPs wireless communication system utilizes the directional as well as phased antenna arrays to construct its ground cells. Sharshar. Any of the formed beams is constructed by a number of antenna elements therefore any element failure in the array will slightly distort the beam pattern (the beam will have slightly larger beamwidths) and this can be an advantage compared with the use of directional antennas. The use of directional antennas has some advantages such as its practical availability and simplicity but on the other hand a failure in one of them results in a coverage hole due to the absence of the beam used in forming its cell. we can approximate the earth as a flat surface as shown in Fig.264 Dessouky. the footprint of a beam formed by any of the mentioned antennas onboard the HAP is shown. The paper is arranged as follows.

PIER 67. Flat ground approximation geometry. . and the cell minor axis distance HK can be denoted as aF and can be given by   Bφ aF = 2h sec (θc ) tan (3) 2 These two quantities (i. 2007 265 P Bφ S E C Earth H K F Figure 1. the minor and major axes) will define the cell shape and this assumption can be used for smaller and moderate coverage areas but when the coverage area increases the approximation error will increase and can’t no longer be used.e.Progress In Electromagnetics Research. HAP cell footprint. P Bθ h θo θc S E C F Earth Figure 2..

the major axis will be the arc between the two ground central angles γ1 and γ2 which can be deduced as     −1 h Bθ Bθ γ1 = sin 1+ sin θo − − θo + (4) R 2 2 and     −1 h Bθ Bθ γ2 = sin 1+ sin θo + − θo − (5) R 2 2 . we take into consideration the earth curvature which good predicts the cell footprint. 3. which depicts the geometry used to define the cell parameters.266 Dessouky. Sharshar. and Albagory Bθ h θo θc S A E B C D F Earth γ2 γo γ1 G Figure 3. In this figure. CURVED EARTH MODEL FOR HAP COVERAGE CELLS In the following section. A side view is shown in Fig. 3. Curved earth coverage geometry.

Progress In Electromagnetics Research. ac . the cell major axis. PIER 67. will be   Bφ aC = HK = 2P C tan (14) 2 or   Bφ aC = 2h sec (θc ) tan (15) 2 . bC . can be defined as bC = EF = R(γ2 − γ1 ) (10) or      −1 h Bθ bC = R sin 1+ sin θo + R 2      −1 h B θ − sin 1+ sin θo − − Bθ (11) R 2 and in this case the value of θc will be   −1 BC θc = tan (12) PB or        tan(γo )  θc = tan−1   (13)   h  2 1 + (cos(γ1 ) + cos(γ2 )) − 1   R therefore the cell minor axis. 2007 267 where the cell center has a ground center angle given by 1 γo = (γ1 + γ2 ) (6) 2 and we can get the distance P B (see Appendix A) as follows:   1 P B = h + R 1 − (cos(γ1 ) + cos(γ2 )) (7) 2 and the distance BC (see Appendix A) will be 1 BC = R(cos(γ1 ) + cos(γ2 )) tan(γo ) (8) 2 therefore the platform-to-cell center slant distance will be  PC = P B 2 + BC 2 (9) from the above equations.

7a and b. On the other hand. The same analysis is done for the cell minor axis as depicted in Figs. 6a and b and the error (absolute and relative) for the two models is shown respectively in Figs. The variations in this figure indicates the increase in the footprint with increasing both the beam direction and the beamwidth. the error may approach about 2% of the major axis for beamwidth of 20◦ at a beam direction of about 60◦ which corresponds to 700 meter difference. 5(a). In this figure. ANALYSIS AND COMPARISON FOR THE TWO COVERAGE MODELS Figs. and Albagory which can be also given by    2 Bφ h 1 aC = 2R tan 1+ − (cos(γ1 ) + cos(γ2 )) 2 R 2 1/2 1 + (cos(γ1 ) + cos(γ2 ))2 tan2 (γo ) (16) 4 4. 5a while the relative error in the cell major axis between the two models may be defined as: bC − bF εb = × 100% (17) bC where its variation with both the beam direction and beamwidth is shown In Fig. One can therefore utilize the simple equations used in the flat ground model for the range of moderate and acceptable error (such as for cells near the coverage center) while for larger error we can utilize the better curved earth model thus optimizing the use of both models. Sharshar. The difference (or the absolute distance error) between the two quantities in km is shown in Fig. the error is much smaller for the inner coverage cells and for cells of narrower beamwidth. For example a beamwidth of 5◦ generates cells that have an error not exceeding 12 meters for direction no more than 40◦ as depicted in Fig.268 Dessouky. This large difference between the two expected major axis values for the two models will affect the system design especially for the cells at the coverage border or edges. 4a and b depicts the variations of the cell major axes with the beam direction θo at different beamwidths Bθ for both coverage models using directional antennas. . 5b.

bC variation with beam direction at different beamwidths.Progress In Electromagnetics Research. bF variation with beam direction at different beamwidths. PIER 67. 2007 269 35 30 Bθ = 5 o Bθ = 10 o Ground Major Axis b F in km 25 Bθ = 15 o Bθ = 20 o 20 15 10 5 0 0 10 20 30 40 50 60 Beam Direction θο in Deg Figure 4a. . 35 30 Bθ = 5 o Bθ = 10 o Ground Major Axis b C in km 25 Bθ = 15 o Bθ = 20 o 20 15 10 5 0 0 10 20 30 40 50 60 Beam Direction θο in Deg Figure 4b.

5 % Normalized Ground Major Axis Error Bθ = 5 o 2 Bθ = 10 o Bθ = 15 o Bθ = 10 o 1.5 1 0. Absolute error variation with beam direction at different beamwidths. .5 0 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 5b. and Albagory 0 10 Ground Major Axis Approximation Error in km Bθ = 5 o -1 Bθ = 10 o 10 Bθ = 15 o Bθ = 20 o -2 10 -3 10 -4 10 -5 10 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 5a. Sharshar.270 Dessouky. 2. Relative error variation with beam direction at different beamwidths.

aC variation with beam direction at different beamwidths. aF variation with beam direction at different beamwidth. . PIER 67. 2007 271 18 16 Bφ = 5 o B φ = 10 o 14 B φ = 15 o Ground Minor Axis a F in km 12 B φ = 20 o 10 8 6 4 2 0 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 6a. 18 16 Bφ = 5 o Bφ = 10 o 14 Bφ = 15 o Ground Minor Axis a C in km 12 Bφ = 20 o 10 8 6 4 2 0 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 6b.Progress In Electromagnetics Research.

0.1 0 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 7b.2 0. and Albagory 0 10 Ground Major Axis Approximation Error in km Bφ = 5 o -1 10 B φ =10 o B φ = 15 o B φ =20 o -2 10 -3 10 -4 10 -5 10 0 10 20 30 40 50 60 Beam Direction θo in Deg Figure 7a.5 0.3 0. Sharshar.272 Dessouky.4 0.8 B φ = 10 o B φ = 15 o 0.6 0. . Relative error variation with beam direction at different beamwidths.7 B φ = 20 o 0. Absolute error variation with beam direction at different beamwidths.9 Bφ = 5 o % Normalized Ground Minor Axis Error 0.

(A4) can rewritten as 1 P B = h + R − (GA + GD) (A5) 2 or 1 P B = h + R − (R cos(γ1 ) + R cos(γ2 )) (A6) 2 or   1 P B = h + R 1 − (cos(γ1 ) + cos(γ2 )) (A7) 2 and BC will be BC = GB tan(γo ) (A8) . The first model approximates the earth as a flat surface which simplifies the equations used for coverage analysis but this approximation will be more erroneous especially for cells at the coverage edges of the HAP. 2007 273 5. Therefore a composite cell footprint equations may be used to both simplify the analysis for inner cells and for more accurate results for the outermost cells. if the earths radius is R. PIER 67. From Fig. APPENDIX A. CONCLUSION High altitude platforms is an emerging technology for different types of communications especially mobile cellular radio. then the distance from the earth’s center to point A will be GA = R cos(γ1 ) (A1) and the distance connecting the earth’s center to point D will be GD = R cos(γ2 ) (A2) and from the same figure we have 1 GB = (GA + GD) (A3) 2 therefore we can get the distance P B as P B = h + R − GB (A4) substituting GB with Eq. 3. Two coverage models are presented which explores the cell footprint. (A3). therefore Eq.Progress In Electromagnetics Research.

M. “Optimization of beams directions for high altitude platforms cellular communications design. B. Hadhoud. “Modeling and investigating the rotational motion effects of the high altitude platforms. 3.. Steele.. M. 686–700. Massa. Donelli. M. Sharshar. No. Italy. El-Jabu. 3. . Caorsi.” IEEE Transactions on Wireless Communications. (A2) in Eq. Capstick. Nofal. PIER 60. “Two- dimensional switched-beam smart antennas for stratospheric platform cellular communications.” 23rd National Radio Science Conference (NRSC 2006). and M.. J.274 Dessouky. “Cellular communications using aerial platforms. Konefal. Albagory. Mouhamadou. F. Sharshar.” 22nd National Radio Science Conference (NRSC 2005). Grace. S. 2006. M. and R. Veh. C. and T. (A3) in Eq.” Electronics & Communications Engineering Journal. M. “Optimizing an array of antennas for cellular coverage from a high altitude platform. (A1) and Eq. “Broadband communications from a high-altitude platform: the European HeliNet programme. Nofal. Vol. March 15–17. 2001.. Croatia. C. 4. May 2001. and Y. and Y. Vol. Albagory.” Progress In Electromagnetics Research. 6. D. 2005. Dubrovnik. June 2001. Menoufia. P. Rammal. 50. M. Egypt. 279–285. M.” Proceeding of The IEEE International Conference on Software.” IEEE Trans. D. and T. 7. Egypt. Thornton. Technol. H.. October 9–14. M. M. Split. J. Ancona. T. 2.” Progress In Electromagnetics Research. and A. May 2003. Albagory. “Linear antenna synthesis with hybrid genetic algorithm. 95–106. and Computer Network (SoftCOM 2001). M. Sharshar.. 8. “Smart antenna array patterns synthesis: Null steering and multi-user beamforming by phase control. Tozer. PIER 49. M. and Albagory substituting with Eq. 484–492. and Y. Dessouky. 2006. 1–22. Spillard. (A8) we have 1 BC = (GA + GD) tan(γo ) (A9) 2 and finally substituting Eq. Dessouky.. Bari. Cairo. H. Pastorino. DeNatale. H. Thornton. Tozer. 2004. Vaudon. Telecommunications.. (A9) we have 1 BC = R(cos(γ1 ) + cos(γ2 )) tan(γo ) (A10) 2 REFERENCES 1. Grace. 2. Dessouky. 5. C. March 15–17. Nofal.