Columbia, USA. {sun55@email., matolak@}sc.edu. AbstractâAs use of unmanned aircraft systems (UAS) is expected to dramatically increase in the coming ...
2014 IEEE Military Communications Conference
Over-Harbor Channel Modeling with Directional Ground Station Antennas for the Air-Ground Channel Ruoyu Sun, David W. Matolak Department of Electrical Engineering University of South Carolina Columbia, USA {sun55@email., matolak@}sc.edu
Abstract—As use of unmanned aircraft systems (UAS) is expected to dramatically increase in the coming years, and two new frequency bands (960-977 MHz in L-band and 5030-5091 MHz in C-band) have been allocated for future control and nonpayload communications for UAS civil applications, characterization of the air-ground channel becomes essential. Directional ground station (GS) antennas are expected to be used, hence the UAS is likely to fly into the GS backlobe area in some conditions. In this paper, dual-band single-input multipleoutput (SIMO) data is employed to estimate channel characteristics. This data was taken over a harbor in Oxnard, CA where the aircraft was outside the half power main beam of the GS antennas. We analyze and compare results with those taken within the main beam. Although the results depend on specific antenna pattern, the statistical models provide a useful evaluation of this AG channel. In the backlobe area the path loss is 15 dB larger (in C-band) and 20 dB larger (in L-band) than free space path loss, the fading is rapid and severe (worse than Rayleigh), and the root mean square delay spread (RMS-DS) is on average 35 ns with maximum 400 ns, whereas the RMS-DS in the main beam over water setting is only 10 ns. The number of multipath components (MPCs), MPC excess delay, and the path loss of the line of sight (LOS) component all increase in the backlobe area.
Glenn Research Center (GRC) conducted a measurement campaign for the two CNPC bands in multiple representative ground station (GS) environments including over sea [7], [8] and fresh water [9], urban, suburban, rural, desert, hilly [10] and mountainous [11]. In this paper we focus on a segment of data in an overharbor setting in Oxnard, CA near the Pacific Ocean, where the aircraft (AC) was outside the half power main beam of the GS antennas. Although many papers have reported results for the over water AG channel, e.g., [7]-[9] and [12]-[14], no one has studied results for the backlobe area. In [15], the authors provided path loss and narrowband shadow fading models for the backlobe area in a cellular GSM-R system under a base station (BS) tower with a directional BS antenna. We define two regions as illustrated in Fig. 1: (1) the AC is out of the main beam of the GS antenna where the GS transmitter antenna gain Gt is only very approximately known; (2) the AC is within the half power main beam where Gt is almost a constant. Additional details are discussed in Section III.
Keywords—air-ground channel; unmanned aircraft system; path loss; RMS delay spread
I. INTRODUCTION Use of unmanned aircraft systems (UAS) is expected to grow dramatically in the next decade [1]. In order to secure safety and reliability, control and non-payload communication (CNPC) link specifications are being designed by academia, government, and industry with rigorous performance requirements. Two spectral bands have been allocated for CNPC, including L-band from 960 MHz to 977 MHz and Cband from 5030 MHz to 5091 MHz [2]. The special committee 228 (SC-228) of the Radio Technical Commission for Aeronautics (RTCA) is leading development of the CNPC specifications [3], which will also affect policies of the International Civil Aviation Organization (ICAO) [4].
Fig. 1. Definition of two regions, not to scale.
The backlobe area and flight areas with large elevation angles are often neglected by researchers, and these areas are usually avoided by communication system designers. However, unless the GS has an antenna beam directed upward, even if sectored antennas are applied, the AC will still be out of the antenna main beam at large elevation angles. The poor coverage, severe fading and large delay spread that can occur in backlobe areas can dramatically degrade the performance of CPNC systems operating in the backlobe area. Backlobe channel characteristics can also be used in estimating intersector interference in sectored antenna schemes. In this paper we report measurement based channel models for the backlobe
The air-ground (AG) wireless channel is an essential element for any new aeronautical communication system. Existing studies for the AG channel are sparse, and for different frequency bands and environments [5], [6]. NASA’s 978-1-4799-6770-4/14 $31.00 © 2014 IEEE DOI 10.1109/MILCOM.2014.69
382
area in an AG channel. Despite the dependence of the results on specific antenna patterns, the empirical statistical models provide interesting practical results.
OxnardCA***06-11-2013***FT2 x 10
Section II describes the measurement environments and equipment. In Section III we provide detailed data analysis and results for path loss, fading and root mean square delay spread (RMS-DS). The conclusions appear in Section IV.
Tx (Ground) Rx (Aircraft) Backlobes C-band Backlobes L-band Rx start point Rx end point
6
3.565 3.5645
Z
3.564 3.5635
II. MEASUREMENTS
3.563
A. Measurement Sites The dual-band measurements were conducted over the Pacific Ocean near Oxnard, CA on 11 June 2013. Twelve flight tracks (FTs) were measured. FT2 includes the backlobe area. The GS was located on a parking lot 250 m from the coast line. A photo of the GS taken from the beach toward Oxnard is shown in Fig. 2. More photos and information for the measurement sites are described in [7].
3.5625 -4.606 -2.578
-4.608 x 10
-2.58
6
-2.582
-4.61 -2.584
Y
-4.612
-2.586
x 10
6
X
(a)
(b) Fig. 3. Flight route and backlobe areas for C- & L-band in (a) ECEF coordinates; (b) Google Earth®.
Fig. 2. View of GS from beach toward Oxnard, CA, 11 June 2013.
The GS antennas were elevated 20 m above the ground, oriented 283 from geographic north with zero elevation angles. In the backlobe area, the altitude difference between AC and GS ranged from 780.6 to 783.8 m. The azimuth angle ranged from -40 to 60 degrees relative to geographic north. The elevation angle was between 21.2 and 48.5 degrees. Each of the four receivers recorded approximately 70,000 PDPs in region 1.
The whole route of FT2 is over 40 km, but the route outside the half power main beam of the GS antennas is only approximately 2.5 km in L-band and 3.5 km in C-band. Fig. 3 (a) shows the flight route in earth-centered earth-fixed (ECEF) coordinates based on GPS information; the pink circle indicates the point where the channel sounder began recording. The link distance d between GS and aircraft at the starting point is 2163 m. The green and brown lines are the backlobe areas of the C- and L-band antennas, respectively. The link distance at the boundary of the 3 dB main beam is 1857 m in C-band and 1175 m in L-band. The minimum link distance dmin is 1043 m. The blue line is the route for the rest of FT2, within the main beam for both bands. Fig. 3 (b) is the flight route in Google Earth®. The green and brown balloons are boundary of the GS antennas’ main beam. In region 1, the earth surface reflections are on the ground. Additional potential reflectors include houses of no higher than three levels, water vehicles in the harbor, the harbor itself, ground vehicles, and the ground itself (areas of sand).
B. Measurement System The measurement system, termed the channel sounder, is a customized dual-band, single-input multiple-output (SIMO) direct sequence spread spectrum (DS-SS) stepped correlator manufactured by Berkeley Varitronics Systems, Inc. [16]. The transmitter antenna, mounted on a tower as shown in Fig 2, transmits an L-band (968 MHz) and a C-band (5060 MHz) signal simultaneously. The detailed description is reported in [11]. The transmit power for both bands is 10 W. In order to extend the range for C-band, we employed a 7 dB gain high
383
power amplifier to increase the transmit power to 47 dBm. Low noise amplifiers (LNAs) of gain 30 dB were used at the C-band receiver inputs. The L-band receivers used 15.5 dB gain LNAs. The channel sounder transmits at 50 MChips per second in C-band and 5 MChips per second in L-band, which corresponds to a delay resolution of 20 ns in C-band and 200 ns in L-band. The output of the channel sounder is a sequence of power delay profiles (PDPs) with update rate of approximately 3 kHz. The chip length for each PDP is 1023, which allows a maximum delay span of 20.46 microseconds in C-band and 204.6 microseconds in L-band.
Azimuth Plane 10
5
0
-5
-10 0
90
180
270
360
C-band GS Antenna Gain (dBi)
C-band GS Antenna Gain (dBi)
The patterns of the directional Tx (GS) antennas are shown in Fig. 4, with main beams directed to zero azimuth/elevation angles. The maximum gains are 6 dB in C-band and 5 dB in Lband. Two L-band receivers and two C-band receivers are located on S-3B Viking aircraft. The locations of Rx antennas are shown in Fig. 5, the front to rear and left to right distances between Rx antennas are 1.29 and 1.32 m, respectively. The Rx antennas are omni-directional monopoles with gain 5 dB. They are S65-5366 series manufactured by Sensor Systems® Inc. [17].
0 -10 -20 -30 -40 -50 90
60
-30
-60
-90
In region 1, the aircraft lies in the backlobes of the GS antennas where it is possible that the line of sight (LOS) is heavily attenuated in a null but some other multipath components (MPCs) are in a backlobe. The power of these MPCs can thus be relatively stronger than in the main beam. Compared with the over-water results in the main beam, results in region 1 show fast fading and large delay spread. In this paper, we focus on region 1, and provide comparison with the results in region 2.
(d)
10 5 0 -5 -10
270
Azimuth Angle (degree)
360
L-band GS Antenna Gain (dBi)
L-band GS Antenna Gain (dBi)
0
(b)
(c)
(e)
30
Elevation Angle (degree)
Azimuth Plane
180
In post processing, we applied the dynamic antenna gains according to the elevation/azimuth angles and the orientation of the GS antennas. The patterns of the Rx antennas were measured on a metal ground plate in an anechoic chamber, and not while mounted on the aircraft. The metallic body of the aircraft has strong impact on the patterns, which introduces uncertainties into our measurements, especially in region 1 as shown in Fig.1, where the elevation angle is large. Both Gr and Gt have resolution of one degree. The backlobes are always narrow, the depth and location of the nulls are sensitive to resolution and the method with which the patterns are estimated. Hence Gr and Gt are only very approximately known at large elevation angle.
Elevation Plane
(a)
90
III. DATA ANALYSIS AND RESULTS
10
Azimuth Angle (degree)
-15 0
Fig. 5. S-3B Viking aircraft with four receiver antennas on bottom.
Elevation Plane 10 5
A. Path Loss The path loss in dB vs. link distance for the two C-band receivers is shown in Fig. 6. As indicated in Fig. 2, the link distance between AC and GS first decreased and then increased (hence in Fig. 6 two loss values may occur for a single distance value). The pink circle and black square are the Rx start point and end point within the backlobe areas, respectively, which correspond to the points shown in Fig. 2. Compared to the path loss within the main beam, as the distance extends to 5 km (Fig. 6 (b)), the path loss in the backlobe areas is up to 15 dB larger than the expected free space path loss, and has large variation. The standard deviation (SD) of the measured path loss difference from free space path loss is 6.8 dB for Rx1 and
0 -5 -10 -15 -20 90
60
30
0
-30
-60
-90
Elevation Angle (degree)
(f)
Fig. 4. GS antenna patterns. (a) C-band azimuth plane; (b) C-band elevation plane; (c) C-band 3D pattern in linear scale; (d) L-band 3D pattern in linear scale; (e) L-band azimuth plane; (f) L-band elevation plane.
384
Similarly, the path loss in L-band is shown in Fig. 7. The path loss in the main beam shows clear two-ray behavior [7], but the path loss in the backlobe area does not have two-ray lobes, and this is likely due to the presence of more MPCs; here the measured path loss is up to 20 dB larger than free space path loss. The SD of the measured path loss with respect to free space path loss is 5.8 dB for Rx1 and 6.4 dB for Rx2. In region 2 the SD is approximately 2.3 dB.
7.3 dB for Rx2. In contrast, in region 2 the SD is less than 2 dB. OxnardCA***06-11-2013***FT2***C-band Rxs
Path Loss (dB)
130
120
110
Measured PL Rx1 Measured PL Rx2 Free Space PL Rx start point Rx end point
100
90 1043
1200
1400
1600
1800
2000
B. Narrowband Fading We extracted amplitude fading data from the path loss measurements; fading amplitude distributions are shown in Fig. 8. Generalized Extreme Value (GEV), lognormal, Weibull and Rayleigh distributions are plotted. The probability density function of the GEV distribution is:
2163
Link Distance (m)
(a) OxnardCA***06-11-2013***FT2***C-band Rxs
Path Loss (dB)
130
(1)
120
110
100
90
where ξ is the shape factor. The parameters of these four distributions are listed in Table I. Rice and Nakagami distributions do not fit the data. Due to the complexity of the backlobe patterns combined with multipath, the distributions show two or three modes, and often fading is severe, or worse than Rayleigh [18].
Measured PL Rx1 Measured PL Rx2 Free Space PL Rx start point Rx end point 1500
2000
2500
3000
3500
4000
4500
5000
Link Distance (m)
(b)
TABLE I.
DISTRIBUTION PARAMETERS FOR AMPLITUDE FADING.
Fig. 6. Path loss of C-band receivers. (a) Back lobe areas only; (b) including backlobe areas and part in region 2 up to 5 km.
Path Loss (dB)
110
Lognormal Weibull Rayleigh
100
Measured PL Rx1 Measured PL Rx2 Free Space PL Rx start point Rx end point
90
80 1043
ξ μ σ μ σ β σ
Generalized Extreme Value
1200
1400
1600
1800
2000
Amplitude GEV Lognormal Weibull Rayleigh
30
2166
Link Distance (m) Density
25
(a)
20 15
L-band Rx1 Rx2 0.58 0.62 0.02 0.02 0.01 0.01 -3.45 -3.45 0.66 0.74 1.69 1.41 0.03 0.04
Amplitude GEV Lognormal Weibull Rayleigh
40
Density
OxnardCA***06-11-2013***FT2***L-band Rxs 120
C-band Rx1 Rx2 0.43 0.59 0.02 0.02 0.02 0.02 -3.45 -3.45 0.78 0.84 1.44 1.38 0.04 0.04
30 20
10
OxnardCA***06-11-2013***FT2***L-band Rxs
0
0.05
0.1
0
0.15
0.05
Amplitude
110
1500
2000
2500
3000
3500
4000
4500
Amplitude GEV Lognormal Weibull Rayleigh
30
5000
Link Distance (m)
20
10
(b)
0
Amplitude GEV Lognormal Weibull Rayleigh
30
20
10
0.02 0.04 0.06 0.08 0.1 0.12 0.14
Amplitude
(c)
Fig. 7. Path loss of L-band receivers. (a) Back lobe areas only; (b) including backlobe areas and part in region 2 up to 5 km.
0.15
(b) 40
Density
90
80
40
Measured PL Rx1 Measured PL Rx2 Free Space PL Rx start point Rx end point
0.1
Amplitude
(a) 100
Density
Path Loss (dB)
10
5
120
0
0.05
0.1
0.15
0.2
Amplitude
(d)
Fig. 8. Histogram of received signal envolope amplitude of (a) C-band Rx1; (b) C-band Rx 2; (c) L-band Rx1; (d) L-band Rx2.
385
few MPCs present have delay up to 1820 ns. The power of the strongest (LOS) component increases from the range -40 to -62 dBm up to -31 dBm when the AC goes from the backlobe area to the main beam.
C. RMS Delay Spread in C-band The RMS-DS is the most widely used measure of the time dispersion of the channel. We denote RMS-DS as , and RMS-DS is computed as follows [19]:
,
where k denotes the kth MPC’s amplitude, L is the number of MPCs in the PDP (1023 as indicated in Section II.B), and k denotes the delay of the kth MPC. The mean energy delay is computed as
.
The C-band RMS-DS along with moving averaged RMSDS over 1000 PDPs is shown vs. distance in Fig. 9. The RMSDS in the main beam is approximately 10 ns with some “bumps” no larger than 50 ns as seen in Fig. 9 (b) and [7].1 The RMS-DS in the backlobe area has mean 35 ns, maximum 220 ns for Rx1 and mean of 37 ns, maximum of 391 ns for Rx2. OxnardCA***06-11-2013***FT2***C-band Rxs
(a)
RMS-DS (ns)
400
300
200
Measured Rx1 Measured Rx2 Moving Averaged Rx1 Moving Averaged Rx2 Rx start point Rx end point
100
0 1043
1200
1400
1600
1800
2000
2163
Link Distance (m)
(a) OxnardCA***06-11-2013***FT2***C-band Rxs
RMS-DS (ns)
400 Measured Rx1 Measured Rx2 Moving Averaged Rx1 Moving Averaged Rx2 Rx start point Rx end point
300
200
100
0
1500
2000
2500
3000
3500
4000
4500
5000
Link Distance (m)
(b)
(b)
Fig. 9. RMS-DS of C-band receivers (a) Back lobe areas only; (b) including backlobe areas and part in region 2 up to 5 km.
Fig. 10. Sequence of PDPs for C-band Rx1 (a) from start point to dmin; (b) from dmin to 2.5 km (inlcude part of region 1 & 2).
Fig 10 (a) is the sequence of PDPs of C-band Rx1 from the start point to the minimum link distance dmin. Fig. 10 (b) shows the sequence of PDPs from dmin to approximately 2.5 km. The RMS-DS decreases substantially after the half power main beam boundary at approximately 1857 m. Many MPCs are present in the backlobe area, most within 1000 ns delay, but a
An example individual PDP is shown in Fig. 11; the link distance of this 17150th PDP is 1287.8 m, which lies in the segment before dmin (Fig. 10 (a)). The strongest MPC (likely the LOS) is rotated to a delay value of 100 ns. MPCs are present with a maximum excess delay of 540 ns, with MPC power from 10 to 30 dB smaller than that of the LOS. The instantaneous RMS-DS is 61.6 ns.
1
Due to the present of filters in the channel sounder, the minimum RMS-DS (measured through a back-to-back connection) [20] that can be measured is 10 ns in C-band.
386
[4]
OxnardCA***06-11-2013***FT2***C-band Rx1 -35
[5]
-40
Chip Power (dBm)
-45
[6]
-50
[7]
-55 -60
[8]
-65 -70
[9]
-75 -80
0
100
200
300
400
500
600
700
[10]
Delay (ns) Fig. 11. Example of individual PDP for C-band, 17150th PDP at d=1287.8 m.
[11]
IV. CONCLUSION In this paper, we reported on measured results for the airground channel when the aircraft is outside the half power main beam of the GS antennas for an over harbor environment in Oxnard, CA near the coast line of the Pacific Ocean. The path loss is up to 15 dB in C-band (5060 MHz) and 20 dB in Lband (968 MHz) larger than the free space path loss. The path loss variation is also much larger than that within the main beam. The received signal amplitude fading is fast and worse than Rayleigh. The C-band RMS-DS is up to 400 ns with mean of 35 ns, whereas the RMS-DS in the main beam (for the flight over the Pacific Ocean) is essentially near the measurement lower limit of 10 ns with a few values no larger than 50 ns. The sequence of PDPs and example individual PDPs show that the power of the LOS component in the backlobe area is 10-30 dB smaller than that in the main beam and has rapid variation. The number of MPCs and their excess delay within the backlobe are much larger than those in the main beam as well. The excess delay is up to 1.7 microseconds.
[12]
[13]
[14]
[15]
[16] [17] [18]
REFERENCES [1]
[2] [3]
[19]
International Telecommunications Union (ITU), “Characteristics of unmanned aircraft systems and spectrum requirements to support their safe operation in non-segregated airspace,” Report ITU-R M.2171, December 2009. International Telecommunications Union, website, www.itu.int, 13 May 2014. Radio Technical Commission for Aeronautics, website www.rtca.org, 13 May 2014.
[20]
387
International Civil Aviation Organization, website, www.icao.int, 13 May 2014. D. W. Matolak, “Air-Ground Channels & Models: Comprehensive Review and Considerations for Unmanned Aircraft Systems,” Aerospace Conference, 2012 IEEE, Big Sky, MT, 3-10 March 2012. D. W. Matolak, “AG Channel Measurements & Modeling: Initial Analysis & Flight Test Planning,” (Report #2) NASA Grant #NNX12AD53G, 8 June 2012. D. W. Matolak, R. Sun, “Initial Results for Air-Ground Channel Measurements & Modeling for Unmanned Aircraft Systems: Over-Sea,” Aerospace Conference, 2014 IEEE, vol., no., pp. 1-15, Big Sky, MT, 1-8 March 2014. D. W. Matolak, R. Sun, “AG Channel Measurement & Modeling Results for Over-Sea Conditions,” (Report #6) NASA Grant #NNX12AD56G, 3 Dec. 2013. D. W. Matolak; R. Sun, "Air-Ground Channel Characterization for Unmanned Aircraft Systems:The Over-Freshwater Setting," Integrated Communications, Navigation and Surveillance Conference (ICNS), 2014, vol., no., pp. 1-9, Herndon, VA, 8-10 April 2014. D. W. Matolak, R. Sun, “Air-Ground Channel Characterization for Unmanned Aircraft Systems: the Hilly Suburban Environment,” to appear, IEEE Vehicular Technology Conference (VTC Fall), Vancouver, Canada, 14-17 Sept. 2014. D. W. Matolak, R. Sun, "Air-ground channel measurements & modeling for UAS," Integrated Communications, Navigation and Surveillance Conference (ICNS), 2013, vol., no., pp.1-9, 22-25, Herndon, VA, 23-25 April 2013. Y. S. Meng, Y. H. Lee, “Measurements and Characterization of Air-toGround Channel Over Sea Surface at C-Band with Low Airborne Altitudes,” Vehicular Technology, IEEE Transactions on, vol. 60, no. 4, pp. 1943-1948, May 2011. M. D. Rice, R. Dye, K. Welling, “Narrowband Channel Model for Aeronautical Telemetry,” Aerospace & Electronic Systems, IEEE Transactions on, vol. 36, no. 4, pp. 1371-1377, October 2000. Q. Lei, M. D. Rice, “Multipath Channel Model for Over-Water Aeronautical Telemetry,” Aerospace & Electronic Systems, IEEE Transactions on, vol. 45, no. 2, pp. 735-742, April 2009. R. He, A. F. Molisch, Z. Zhong; B. Ai, J. Ding, R. Chen, Z. Li, "Measurement based channel modeling with directional antennas for high-speed railways," Wireless Communications and Networking Conference (WCNC), 2013 IEEE, vol., no., pp. 2932-2936, Shanghai, China, 7-10 April 2013. Berkeley Varitronics Systems, Inc., website, www.bvsystems.com, 13 May 2014. Sensor Systems®, Inc., website, www.sensorantennas.com, 13 May 2014. D. W. Matolak; J. Frolik, "Worse-than-Rayleigh fading: Experimental results and theoretical models," Communications Magazine, IEEE , vol.49, no.4, pp.140-146, April 2011 J. D. Parsons, The Mobile Radio Propagation Channel, 2nd ed., John Wiley & Sons, New York, NY, 2000. D. W. Matolak, R. Sun, “AG Channel Measurements & Modeling: Initial Channel Sounder Laboratory & Flight Tests,” (Report #3) NASA Grant # NNX12AR56G, 29 January 2013.