1

ACP-WGF30IP08
/
International Civil Aviation Organization / ACP WG-F/30
IP 08

AERONAUTICAL COMMUNICATIONS PANEL (ACP)

30th MEETING OF WORKING GROUP F

ICAO Regional Offices, Bangkok, Thailand

13-19 March, 2014

Agenda Item 10: / Any Other Business

L-Band and C-Band Air-Ground Channel Measurement & Modeling

for Over-Sea Conditions

(Prepared by David Matolak, Kurt Shalkhauser and Robert Kerczewski)

(Presented by Robert Kerczewski)

SUMMARY
This information paper is intended to inform ICAO WG-F of the progress of L-Band and C-Band air-ground channel model development based on data gathered through a propagation flight measurements campaign performed by NASA. The L-Band and C-Band frequencies being modeled will support the line-of-sight Control and Non-Payload Communications (CNPC) for unmanned aircraft systems. The essential elements of the first channel model based on the measured data have been quantified for the specific terrain case of over sea/salt water conditions, and these are described in this paper.

1.INTRODUCTION

1.1At previous meetings of Working Group F, information papers have provided descriptions of plans by the National Aeronautics and Space Administration (NASA) to conduct a series of flight tests to measure the propagation characteristics of the air-ground (AG) channel at two frequency bands intended for Unmanned Aircraft Systems (UAS) Control and Non-Payload Communications (CNPC) and use the resulting data to enable the development of detfailed AG channel models under NASA’s UAS Integration in the National Airspace System Project (ACP-WG-F/25 WP 13, ACP-WG-F/26 IP8); the propagation measurement system and channel modelling approach were described (ACP-WG-F/28 IP3); and the first four sets of flight tests were described (ACP-WG-F/29 IP 5). The AG propagation flight measurements were made for two bands intended for use by UAS line-of-sight CNPC, the L-band (960-977 MHz) and C-band (5030-5091 MHz).

1.2Since WG-F/29, two additional flight test sets were completed in the Cleveland, Ohio, USA and Telluride, Colorado, USA areas. These two sets of tests complete the currently planned propagation measurement flight campaign. Some additional flight tests are being considered with the transmitter on the aircraft and separated receivers on the ground.

1.3Processing and analysis of very large sets of flight test data have been on-going since the first data was collected with the goal of developingof AG channel models for several different terrain types including urban, suburban, flat rural, hilly rural, desert, mountain, fresh water (over lake) and salt water (over sea). The analysis of the measurement data and development of channel model components(propagation path loss, root-mean square delay spread (RMS-DS), and the correlation coefficient of the primary received signal components on four antennas (two antennas for C-band, two for L-band)) has been completed for the over sea terrain. The results are presented in the following sections.

2.Air-Ground Channel Measurement and Modeling for Over-Sea Conditions

2.1Results for AG channel characteristics were recently compiled for flight paths over the ocean, with the ground site (GS) located on the coast. The channel characteristics were obtained from measured data taken in the two frequency bands currently planned for use by unmanned aircraft systems (UAS): the L-band (960-977 MHz) and C-band (5030-5091 MHz). These results are for the first of a set of several flight tests conducted in distinct GS environments. Here we summarize measured results for propagation path loss, root-mean square delay spread (RMS-DS), and the correlation coefficient of the primary received signal components on the four antennas (two antennas for C-band, two for L-band).

3.Over-Sea Propagation

3.1The propagation of radio signals in over-sea environments has been studied for decades, for a wide range of frequencies, for both communications and radar applications. The results here are new in that they (i) contain simultaneous measurement data for both L-band and C-band, (ii) contain measurements with two spatially-separated antennas for each band, and (iii) employ a wider measurement bandwidth than all previously known measurements.

3.2Despite the fact that the ocean usually offers a very open setting, meteorological conditions can yield atypical propagation effects such as ducting and attenuation due to hydrometeors. For the L-band measurement frequency, hydrometeor attenuation is negligible, and only in very heavy rainfall would the C-band signal be appreciably affected; in any case no precipitation was in effect in the area during the flight test, so attenuation from hydrometeors was essentially zero. Attenuation from atmospheric gases is also negligible in these frequency bands. Based upon flight altitude and measured path loss, no ducting was in effect. Thus the AG channel measurements were taken in “clear sky” conditions.

4.Over-Sea Measurement Environment and Flight Characteristics

ACP-WG-F/29 IP 05 provided a description of general flight test approach and key parameters such as altitude, flight track and terrain details and ground station field of view for the several sets of flight tests as well as the propagation data collection methodology. The oversea flight measurements took place near Oxnard, California, USA on 11 June 2013. With the ground station located on shore, the test flight tracks were performed over open salt water terrain that included a few stationary structures (oil platforms) and various types of watercraft. The aircraft altitude was 2650 feet MSL for the over sea measurements. Six of the twelve flight tracksare shown in Figure 1.a. Figure 1.b shows the view toward the ocean from the GS.

Figure 1.a - Recorded Flight Tracks (six of twelve shown) for Over Sea Propagation Measurements near Oxnard, California, USA

Figure 1.b – View from ground site toward aircraft flight path over ocean, Oxnard, California, USA

5.Over-Sea Measurement Data and Analysis: Example Results

5.1In Figure 2 we show geometric traces for one of the flight tracks (designated FT1), in which the aircraft flew directly toward the GS. The figure shows the straight flight path in Cartesian coordinates, and altitude, azimuth, and elevation angles vs. link distance. Data (not shown) were also taken for aircraft heading, pitch, and roll angles, all computed from the aircraft’s flight data recorder. The pitch and roll angles during this flight were nearly constant at 4 degrees and 0 degrees, respectively.

(a) (b)

(c) (d)

Figure 2. Geometric traces for flight track 1: (a) flight track in ECEF coordinates; (b) altitude difference between aircraft and ground station; (c) azimuth angle; (d) elevation angle.

5.2For this flight track, Figure 3 shows measured and analytical path loss vs. link distance, for the two C-band receivers (Rx1 and Rx2, in (a)), and the two L-band receivers (b). The measured path loss for both bands is close to that of the free-space value, but slightly larger for short distances, and slightly less in C-band for large distances. This is most likely attributable to aircraft antenna pattern effects (depending upon aircraft attitude, some mild obstruction shadowing or reflections from the airframe may also occur at the short distances—this is currently under study). Some of the “lobing” structure of the theoretical two-ray model can be seen in the measured data at the larger values of distance, and this is clearest for L-band. The curved-earth two ray model (including spherical earth divergence and sea-surface roughness effects) fits the measured path loss data better than the flat-earth two-ray and free space models, although the free space model predicts the mean value well for these link distances. More complex models (e.g., Longley-Rice, TIREM, etc.) might produce slightly better agreement but require much more local environment data than the two-ray model. Work is being done to augment the curved-earth 2-ray model to account for additional multipath and airframe effects.

(a) (b)

Figure 3. Measured results of C-band path loss vs. distance for all receivers, FT1: (a) C-band, (b) L-band.

5.3The RMS-DS is the most common measure of the temporal dispersion of the channel; it is roughly the “spread” of the channel impulse response. Figure 4(a) shows RMS-DS vs. distance for C-band Rx1 for FT1. In this figure, the RMS-DS is consistently larger than the single path value of approximately 10 ns only for the higher elevation angles at the shortest link distances. The RMS-DS is approximately 50-60 ns at the shortest link distances, and decreases to 10 ns as link distance exceeds ~2 km. This is as expected, since at larger link distances, the relative delay of the surface reflection decreases to below the C-band resolution of 20 ns. Several small “bumps” in the RMS-DS vs. distance plot are present, and these are indicative of likely reflecting objects on the sea-surface, or large ocean waves. We also include moving-average values of RMS-DS on the plot (averaging window length 1000) to help distinguish actual changes in RMS-DS from isolated RMS-DS “spikes” due to noise.

5.4Figure 4(b) plots a sequence of power delay profiles (PDPs) for the link distance near 2.5 km. The line-of-sight (LOS) component is largest, with its power normalized to 0 dB, located at the 5th chip position (chip duration is 20 ns). The sea-surface reflection at this link distance is at a relative delay of approximately 45 ns. The multipath components (MPCs) indicated at approximately 22 chips from the LOS component (relative delay ~ 440 ns) produce the larger RMS-DS values. These MPCs are also resolvable by the 5 MHz bandwidth L-band signal. The source of these MPCs is not known, but one hypothesis is the presence of a large ship in the ocean between the GS and aircraft; numerous boats and ships were indeed present in the vicinity during the flight tests. Another possibility is one of several offshore drilling platforms. For either of these cases, with the aircraft velocity of near 90 m/s, a short-term reflection as in Figure 4(b) is indeed possible. Yet another possibility for the source of the reflection is an isolated large ocean wave. These “intermittent” MPCs have been reported in over-sea measurements conducted by other investigators. We have also observed an even less frequent, and weaker, “intermittent 4th MPC.”

(a)(b)

Figure 4. Measured dispersion results for C-band, Rx1, FT1: (a) RMS-DS vs. distance, (b) sequence of PDPs.

5.5The final example result we provide is the correlation coefficient between the amplitudes of the first-arriving component signals on two separate antennas. The antennas are arranged at corners of a rectangle of dimensions approximately 0.6 m by 0.4 m. Each band’s antennas are on opposite diagonal corners, so antenna separations are greater than 1.3 wavelengths for L-band and 6 wavelengths for C-band. Table 1 shows correlation coefficient statistics for correlation window lengths of 500, 1000, and 5000 PDPs. The mean value of C-band spatial correlation is over 0.7, and that of L-band increases from 0.16 to 0.42 as the vector length n increases. The coefficient between receivers is denoted FiGj in Table 1, where Fi is F-band Rxi and Gj is G-band Rxj, with F, G either C or L, and i,j either 1 or 2. These large intra-band correlations are expected in this straight flight track where relative antenna orientation is constant.

Table 1. Statistics of Cross Correlation Coefficients for Oxnard, Flight Track 1 (straight path toward GS).

C1C2 / L1L2 / C1L1 / C1L2 / C2L1 / C2L2
Vector Length / 500 / 1000 / 5000 / 500 / 1000 / 5000 / 1000 / 1000 / 1000 / 1000
Mean / 0.72 / 0.72 / 0.72 / 0.16 / 0.22 / 0.42 / -0.01 / 0.01 / -0.01 / 0.02
Max / 0.98 / 0.98 / 0.99 / 0.73 / 0.77 / 0.87 / 0.68 / 0.80 / 0.63 / 0.66
Min / -0.70 / -0.78 / -0.24 / -0.45 / -0.53 / -0.24 / -0.76 / -0.67 / -0.55 / -0.70
Standard Deviation / 0.36 / 0.37 / 0.36 / 0.19 / 0.18 / 0.23 / 0.22 / 0.20 / 0.19 / 0.20

6.Over-Sea AG Channel Results Summary

6.1For path loss, the curved-earth two-ray model (+spherical divergence and surface roughness) provides a reasonable fit to the measured data, altered by several dB at the shortest link distances by aircraft antenna pattern effects. This two-ray model also accounts for the majority of measured RMS-DS results of a few tens of nanoseconds, except for the occasional intermittent reflections from surface objects. These intermittent reflections yield RMS-DS values up to several hundred nanoseconds. For portions of the flight path that were over a harbor area highly populated with boats, the channel was found to be more “continuously dispersive,” with RMS-DS reaching approximately 250 ns. The correlation coefficient between signals on the same-band antennas is generally large for the straight flight paths for C-band (above 0.6), whereas for the L-band signals and for the oval-shaped flight paths the correlation is generally small (below 0.4); analytical work to confirm these correlation behaviors is also underway. Inter-band correlations are typically very small, and are well modeled as zero-mean Gaussian in distribution, with a standard deviation mostly less than 0.2. Hence the over-sea channel effects in the two bands can be considered uncorrelated, which will allow for good diversity gains in dual-band systems. Comprehensive modeling for this over-sea channel is currently underway.

7.Next steps

7.1Processing of the flight measurement data and development of channel models for the other terrain types measured is underway and will be reported at future WG-F meetings as it is completed.

8.action by the meeting

8.1ACP WG-F is invited to note the information provided.