Experimental Measurements of Maritime Radio Transmission Channels

Experimental measurements of maritime radio transmission channels

Yvon-Marie Le Roux, Jacky Ménard, Claude Toquin, Jean-Pierre Jolivet, Fabien Nicolas Institut Télécom - Télécom Bretagne, UMR CNRS 3192 Lab-STICC, Brest, France

Email: [email protected]

Abstract-This paper focuses on experimental measurements, the propagation channel: masking effects or reflected paths due carried out by Télécom Bretagne, concerning both maritime radio to the coastal relief and the islands, tide effects that modify the transmission channels and new radio technologies (mainly WiMAX, 802.16e). The aim of these studies is to reinforce the refraction conditions of the indirect path. This paper proposes quality and the robustness of such transmissions. Several and estimates also the efficiency of vertical space diversity for measurements were set up, with specific experimental devices maritime radio links. developed and implemented by Télécom Bretagne, to characterize of the propagation channel in maritime environment and to study the communication performances of WiMAX. The II. MEASUREMENTS experimentations were carried out at frequencies around 3.5 GHz (Licensed WiMAX Band in France) and 5.8 GHz (Free Band in A. Measurement Locations France). A previous paper [1] published by the same authors The measurements presented were carried out in the Brest showed that there is a good agreement between the theoretical harbour during two different campaigns realized in March values of field attenuation computed with the well known two-ray 2008 and April 2010 with similar boats. The navigation model and the measurements. New measurements presented in conditions were rather good. Indeed the climate was gentle and this present paper confirm the possible presence of deep fading and bring to light the benefit to exploit the channel diversity the waves not very high. Figure 1 represents the locations thanks to the use of two receiving antennas separated in height. where the measurements were carried out and an example of Furthermore, a comparison between the propagation two routes covered by the boat with transmitting equipments characteristics and the performances of a WiMAX transmission (in yellow the forward route and in white the backward route). are analyzed in order to have a better comprehension of the Geographical representations are from Google Earth. Point 1 functionality of the system. Some results concerning maritime HF and UHF transmission channels are also proposed in this paper. indicates the receiver location for all the experimentations.

I. INTRODUCTION In the wireless world, the needs in wireless broadband communications concern quite particularly the ground zones not provided by ADSL. In maritime environment like harbours and coastal areas, needs are also important but the specificities of this type of propagation channel complicate the transposition of the ground solutions such as WiMAX and require serious studies. Similar studies [2] and [3] were already carried out, but they did not take into account all the environment characteristics. As mentioned in [3] and [4], a basic model being able at a first order to apply to the 5 Km propagation in maritime environment is the two-ray model. This model takes into account one direct path between the transmitter and the receiver and another path refracted by the sea. For this model, it is possible to calculate the level of the Figure 1. Example of routes (Max Tx-Rx distance = 27 km). received field at a given point, according to: the distance between the transmitter and the receiver, the relative heights of B. Measurement Equipments the transmitter and the receiver from the sea, the wave At transmitting site (the boat), two kinds of signals may be polarization, the working frequency, the conductivity, the generated: a continuous sine wave or a pulsed one. The system relative permittivity of the sea and the transmitted power. involves several generators providing the different carrier The present study aims at evaluating the levels of accuracy frequencies 968 MHZ, 3.5 GHz or 5.4 GHz. Installation of of this model and at examining other causes of modification of antennas was different for the two campaigns, because of the configuration of the boats and according to the expected III. MEASUREMENT RESULTS measurements. For the first campaign [1], two horn Tx antennas were A. Theoretical Results installed on the boat, one to 12 meters above water and the As indicated in the introduction a basic model being able to other one between 2.5 meters and 5 meters above water. For apply to the propagation in maritime environment is the two- the second one, omnidirectional narrow frequency range Tx ray model. It takes into account a direct way between the antennas where set up at 14 meters above the sea. At the transmitter and the receiver and a way refracted by sea surface. receiving site, several systems were available, namely a We consider this model for simulate the propagation over the spectrum analyzer controlled by a computer and a second sea, the figure 4 shows simulation examples. system constituted by PXI formatted cards. A controller, a Field attenuation, horizontal polarization RED, vertical polarization GREEN, signal analyzer, a frequency converter and a mass storage 3.5 GHz, hTx = 12 m, hRx = 52 m, sea 1/d slope BLUE, 1/d**2 slope CYAN device constituted this last one. This acquisition system can 20 record the received signal with 80 dB dynamics and a 20 MHz 0 bandwidth centred on a programmable frequency included between a few tens kHz and 6 GHz. -20

For the second campaign described here, an additional dual -40 channels acquisition system was used to store the signals provided by two Rx antennas separated in height by about 4.8 -60 Field (dB) meters. These Rx antennas were 12 dB gain horns with 120° -80 horizontal aperture and 30° vertical aperture, followed by a filter and a low noise amplifier. -100

A GPS equipped a PC on the boat. The time and the position -120 data were recorded for each trip. These data make it possible to

3 4 readjust the recordings of the signal received according to the 10 10 Length Tx-Rx (m) real distance between the transmitter and the receiver. These Figure.4. Simulation with two-ray model: level of the received field according data were confirmed by the equipment of the ship. The GPS to the length Tx-Rx information was also used for clock synchronization of the Tx an Rx devices. When looking at simulated results obtained from this The following pictures represent the devices used for the theoretical model we can deduce the following considerations: experimentations. • the average slope of path loss is proportional to the transmitter-receiver distance approximately up to the radio electric horizon, as for a connection in free space, • the average slope of path loss is proportional to the Figure.2. Transmitting devices squared of the transmitter-receiver distance, approximately beyond the radio electric horizon, • the horizontal polarization presents major and recurrent fading (from 20 to 50 dB) depending on the transmitter- receiver distance, • the vertical polarization presents major and recurrent fading (from 10 to 30 dB) depending on the transmitter- receiver distance, • the fading of horizontal and vertical polarizations are phase opposite until a distance of few hundred meters, • the distance interval between fading for both the Figure.3. Receiving dev ices horizontal and vertical polarizations exceeds the hundred meters for transmitter-receiver distances greater than approximately 1 km.

These characteristics were verified during experiments and Throughout this paper, the presented results concern the the measurement results are presented in the following chapter. following frequencies: 868 MHz, 3.5 GHz and 5.4 GHz.

B. Experimental Results Figures 5 and 7 represent examples of travel during which measurements were carried out.

1 k m

5 k m

Figure 7. Example of travel where the measurements were carried out (18/04/2010. 07h36-08h05 U.T).

Figure 5. Example of travel where the measurements were carried out 18 04 2010 07 36 00 - 08 05 00 f = 5.447 GHz a) hTx = 14 m, hRx = 47.5 m (17/04/2010. 10h11-12h45 U.T). -20

-40 Examples of result are shown on figures 6 and 8, comparing -60 the levels of the theoretical values, defined by the two-ray field (dB) -80 model, and the measured levels of the received signal provided 2000 3000 4000 5000 by two antennas separated in height by 4.8 meters. The third b) f hTx = 14 m, hRx = 42.7 m -20 graph shows the result obtained when the most powered signal -40 of the two antennas is chosen during the travel. The gain is -60 significant, but not optimal since the difference of height field (dB) -80 should be in this case of approximately 12 m. 2000 3000 4000 5000 Max[a),b)] -20

17 04 2010 10 11 00 - 12 45 00 f = 5.447 GHz -40 a) hTx = 14 m, hRx = 50.9 m -20 -60

field (dB) -40 -80 2000 3000 4000 5000 -60 length Tx-Rx (m) -80 field (dB) -100 4000 7000 10000 13000 Measurement Theorical Value Free space attenuation

b) hTx = 14 m, hRx = 46.1 m -20 Figure.8. Comparison of the measured received signal level with simulated -40 values. In blue the measured signal, in green the 2-ray model. -60 -80 There is a good agreement between the theoretical values of field (dB) -100 field attenuation and the measurements. On figure 6, the level 4000 7000 10000 13000 differences at the beginning of the travel are mainly due to Max[a),b)] -20 antenna effects. Indeed, the Tx and Rx antennas has only 30 -40 degrees vertical apertures, and the receiving antenna at point A -60 was located at about 50 meters above the sea, whereas the -80 field (dB) transmitting antenna was onboard a small ship. -100 4000 7000 10000 13000 The presented results confirm the benefit to exploit the length Tx-Rx (m) channel diversity as expected with simulations. The Measurement Theorical Value Free space attenuation fluctuations in the signal on the figure 8 is due to the use of a transmitting omnidirectional antenna and to the presence of Figure.6. Comparison of the measured received signal level with simulated additional reflections on the coasts and show the limit of the values. In blue the measured signal, in green the 2-ray model. two ray model when nearby relief is formed by stiffs and high cliffs. A four ray model would be more convenient in this case.

C. Communication system In this paragraph, a comparison between the performances of a WiMAX transmission are carried out, on the same routes as previously (18/04/2010. 07h34-08h05 U.T), in order to have a better comprehension of the functionality of the system. Effectively, the following figure shows measurement results for the level of the RSSI (Received Signal Strength Indication ) and the throughput from CPE. This throughput was measured by the “iperf” software. In this experimentation, the WiMAX system configurations are a fixed constellation (QPSK3/4), a throughput of 5Mbps in upload and download and a frequency of 5.6 GHz. 1 Km

18 04 2010 07 34 25 - 08 04 51 UT -75 5000 Figure.9. Example of way with an important masking effect (25/03/2008. 15h20-15h50 U.T).

-80 4000 Throughput (kbit/s) 25 03 2008 A

-85 3000 -40

D -90 2000 -50 RSSI (dBm)

-95 1000 -60 Field (dB)

-100 0 -70 1000 2000 3000 4000 5000 Length Tx-Rx (m) -80 C RSSI Throughput B

Figure.11. RSSI and upload throughput according to the length Tx-Rx. -90 5 10 15 20 25 30 Time (Minutes) Figure 11 presents the level of the RSSI looking like the Figure.10. Level of received field with an important masking effect. two-ray model previously described, with fading and attenuation of the level, according to the distance. Concerning E. UHF frequency the throughput, the figure shows an agreement between the level of the RSSI and the throughput. Indeed, the first part of The following figures show measurement results of the the graph shows that the WiMAX system has the RSSI level received field level with a frequency of 868 MHz. Figure 12 sufficient for providing a constant throughput. After that, the represents the locations where measurements were carried out. graph indicates fading corresponding to the two-ray model. For the throughput, it appears a time delay after each increase of the RSSI curve, and this is due to a synchronization time 5 Km required for the WiMAX system to begin to work.

D. Masking effects Other experimentations were realized to understand the masking effects. The following figures show some obtained results. Figure 10 shows the measurement results of levels of received field on a way with an important masking effect. The passage behind and near from point B causes a brutal fading of the field level (approximately 40 dB). The cause of this is the important height of this obstacle (76 meters). The passage in the vicinity causes an important fading. Figure.12. Example of travel where the measurements were carried out (20/04/2010. 09h44-10h58 U.T).

Figure 13 shows the received field level according to the or both connection ends, being able to create complex length Tx-Rx. Multipath effects due to the use of an situations of multi-paths. omnidirectional Tx antenna and to coastal reflections may be noticed. This may approximately be described by a four path To take benefit from these diversities of the propagation model. channel in maritime environment, the following solutions on the systems could be examined: 20 04 2010 09 44 00 - 10 58 00 f = 0.868 GHz • adaptive throughput and/or emission power according to hTx = 12 m, hRx = 45.5 m 0 the reception level and/or the signal to noise ratio (thus depending from distance and masking effects), -10 • taking advantage of the natural channel diversity: -20 - using two transmitting and/or two receiving antennas -30 with orthogonal polarizations (vertical and horizontal),

-40 with a reception strategy to be determined (combining, best level, …). This solution would be useful until -50 approximately 1000 m (in harbour and harbour field (dB) -60 approach). Beyond 1000 m, vertical polarization is

-70 preferable, - take advantage from channel horizontal diversity by -80 using two (or more) receiving antennas approximately -90 distant of half of the maximum interval between two

-100 fading, with a reception strategy to be determined 2000 4000 6000 8000 10000 12000 length Tx-Rx (m) (combining, best level, …). But the effectiveness of the solution depending on the Tx-Rx distance is

Measurement 2 ray model proportional to the size of the ships concerned. - exploit the channel vertical diversity by using two (or more) reception antennas separated by the height from Figure.13. Level of received field according to the length Tx-Rx with a approximately half of the maximum interval between frequency: 868 MHz. In blue the measured signal, in green the 2-ray model. two fading in altitude, with a reception strategy to be determined (recombination, best level, …). Vertical and IV. DATA ANALYSIS AND SYSTEM IMPACTS polarization diversities could be combined, • The theoretical results and the experimental validations show mechanically or electronically control the Tx-Rx antenna that the propagation channel in maritime environment has a pointing. three dimensional diversity: • a polarization diversity which is present until distances V. RELATED MARITIME PROJECTS ranging between 10 meters and 500 meters. Experimental studies in cross polarizations must be The following projects are labelled by the “pôles de made to consolidate this observation, compétitivité’ “MER” and/or “Images et Réseaux”. • the vertical polarization mode is better adapted starting A. Ex’treme project from distances of a few hundred meters, This project proposes the implementation of a broadband • a horizontal space diversity (and temporal for a connectivity between the ground and the ships on sea. It will transmitter and/or a receiver in mobility). This diversity, offer broadband services, mainly to increase the safety on the inversely proportional to the Tx-Rx distance is thus sea and to exchange information at short distances from the reduced with this distance, coast. • a vertical space diversity proportional to the relative Because the already existing IP Mobile services are not yet height variations of the transmitting and receiving really dedicated to the maritime field, the Ex’reme project aims antenna. at providing IP services in these areas. The results confirm also that the masking effects are more or The partners of this project are: Alcatel Business Systems, less important according to the distance relatives to the masks. Thomson-GrassValley, IFREMER, C2 Innovativ’Systems, Moreover, the influence of the sea states is certain on two Morgan’Conseil, Chantiers de l’Atlantique, and Télécom levels and request still to be quantified in experiments: Bretagne. • pointing defaults of the receiving and/or transmitting antennas [5] and [6], B. ImaginLab project • modification of the refraction conditions of the indirect ImaginLab project is an experimental regional platform, way. adopted by FUI 6. It will make it possible to test new products Studies and experimental complements are necessary to and services. The infrastructure design and installation are refine the description of the channel behaviour, in particular for carried out jointly by the pole Images & Réseaux and the UEB the presence of obstacles (coastal relief, large ship …) near one (Université Européenne de Bretagne), this later one having maritime band (4 to 26 MHz) in order to obtain a useful radio delegated its mission to Télécom Bretagne. data rate up to 22 kbps. The infrastructure is made up of three inter-connected For coverage from 40 to 250 NM, low frequencies (4 to 8 MHz) technological platforms: will be used, promoting the ground-wave propagation, as • “Internet du futur” platform, located in Lannion, proved by experimental measurements made by Telecom • “Images en mobilité” platform, located in Rennes, Bretagne and KENTA Electronics. • “haut débit sans fil” platform, located in Brest. For long-distance connections over 200 NM, 8 to 26 MHz These studies are related with Telecom Bretagne wave frequencies will be used. In this case, the ionospheric wave propagation studies in different environments (urban, suburban, propagation is unavoidable. white zones and maritime areas). VI. CONCLUSION AND PERSPECTIVES

C. Palmyre project This paper presented experimental researches carried out by The Palmyre project consists in a realization of a modular the CaPSyS team of Télécom Bretagne on the new radio platform for the evaluation of both systems and new technologies in the maritime environment. This communication concepts in frequency bands going from HF to characterization of the propagation channel is necessary to microwaves, within realistic experimental situations. understand the behaviour of radio waves in a maritime This platform is being developed by a hardware and software environment. Moreover, with this study, the performances of infrastructure group. WiMAX systems were simultaneously analyzed. It is thus Partners of this project are teams from the “ Institut possible to take advantage of this analysis to propose some d'Électronique et de Télécommunications” in Rennes, improvements of the transmission systems for better the “Laboratoire d'Électronique des Systèmes Temps Réel” performances. from University of Bretagne Sud, the UMR LabSTICC from Télécom Bretagne develops new testing systems, in Télécom Bretagne and the “Université de Bretagne particular multi-channel receiving and transmitting systems up Occidentale” in Brest. to the Ku frequency band, in order to explore the channel diversities quoted previously (MIMO spatial and polarization D. Navtrack project diversities) and for a better understanding of the propagation The general concept of this project is constituted of two mechanisms in operational transmission environments technical guidelines: (terrestrial as well as maritime). WiMAX (or others) • real-time localisation of all the actors of a boat race, transmission systems are also being operated simultaneously using a radio transmission (868 MHz) to transmit GPS with channel sounding equipments, in order to compare the data toward ground. A suited cartography traces the propagation channel characteristics with transmission trajectories with environmental data (wind, currents, performances.

tide…) These studies have been partially funded by: Brittany • the second aims at transmitting on line the images of the Regional Council, Finistere Council, Brest Metropole boat race by a radio transmission at 5.8 GHz. Oceane and Europe (FEDER). Partners of this project are teams from DETI, Ecole Navale, AGESSI, Technopole Brest Iroise, Télécom Bretagne and REFERENCES SeatizenPro. [1] Yvon-Marie Le Roux, Jacky Ménard, Claude Toquin, Jean-Pierre Jolivet, Fabien Nicolas, "Experimental measurements of propagation E. The SEANET project characteristics for maritime radio links", 9th International Conference on The recent project SEANET is intended to set up an “ad Intelligent Transport Systems Telecommunications,(ITST), IEEE, 20-22 october, 2009. hoc” communication network for broadband communication [2] J. Joe, S.K. Hazra, S.H. Toh, M.W. Tan, J. Shankar, V.D. Hoang, M. between boats as well on sea as in coastal area. It is being Fujise, “Path Loss Measurements in Sea Port for WiMAX”, Wireless carried out at frequencies around 3.5 GHz (Licensed WiMAX Communications and Networking Conference, 2007. WCNC 2007. IEEE, 11-15 March, 2007. Band in France) and Ku frequency. [3] Rosario G. Garroppo, Stefano Giordano, Davide Iacono, Alessandro The partners of this project are: THALES, SATIMO, DETI, Cignoni, Matteo Falzarano, “Wimax Testbed For Interconnection Of ESTAR, Technopôle Brest-Iroise, ENIB and Télécom Mobile Navy Units In Operational Scenarios”, Military Communications Conference. MILCOM 2008. IEEE, 16-19 Nov, 2008. Bretagne. [4] Yattoun, I.; Labia, T.; Peden, A.; Landrac, G.; Ney, M.; Resibois, M.; Bonnin, J.M.; Baghdadi, A.; Montavont, N.; Fujise, M.; Le Roux, Y, “A Millimetre communication system for IVC”, ITST 07, 7th International F. The IPBC project Conference on ITS, 6-8 June 2007. The IPBC system is planned to allow data transmission on HF [5] Su Wen, Peng-Yong Kong, Jaya Shankar, Haiguang Wang, Yu Ge, Chee-Wei Ang, “A novel Frame work to simulate Maritime Wireless maritime bands 4 to 26 MHz in a radio transmission channel Communication Networks”, OCEANS 2007, Sept. 29 2007-Oct. 4 2007. 10 to 20 kHz bandwidth. [6] Chee-Wei Ang and Su Wen, “Signal Strength Sensitivity and Its Effects The aim of this system is to allow ships to access the Internet on Routing in Maritime Wireless Networks”, Local Computer Networks, rd network for digital mail by a radio link operating in a HF 2008. LCN 2008. 33 IEEE Conference. See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/224108326

Experimental measurements of propagation characteristics for maritime radio links

Conference Paper · November 2009 DOI: 10.1109/ITST.2009.5399326 · Source: IEEE Xplore

CITATIONS READS 15 178

5 authors, including:

Yvon-Marie Le Roux Institut Mines-Télécom

37 PUBLICATIONS 158 CITATIONS

SEE PROFILE

All content following this page was uploaded by Yvon-Marie Le Roux on 19 November 2015.

The user has requested enhancement of the downloaded file. Experimental measurements of propagation characteristics for maritime radio links

Yvon-Marie Le Roux, Jacky Ménard, Claude Toquin, Jean-Pierre Jolivet, Fabien Nicolas Institut Télécom - Télécom Bretagne, UMR CNRS 3192 Lab-STICC, Brest, France Email: [email protected]

• the tide effects that modifies the refraction conditions Abstract: This paper presents studies, carried out by Télécom of the indirect path. Bretagne, concerning new radio technologies (mainly WiMAX, The paper will show various propagation characteristics 802.16e) in the maritime environment. The aim of these studies is observed during our experiments. to reinforce the quality and the robustness of such transmissions. Several measurements were set up to characterize the propagation channel in maritime environment in order to study II. MEASUREMENTS the communication performances of WiMAX. This characterization was obtained by specific experimental devices, A. Measurement Locations developed and implemented by Télécom Bretagne. The The measurements were carried out in the Brest harbour. The experimentations were carried out at frequencies of 3.5 GHz navigation conditions were good. Indeed the climate was (Licensed WiMAX Band in France) and 5.8 GHz (Free Band in gentle and the waves not very high. Figure 1 represents the France). Similar studies [1] and [2] were already carried out, but locations where the measurements were carried out and an they did not take into account all the environment characteristics, namely tides and coast relief masking. Our study looks at these example of two routes covered by the boat with transmitting parameters for a better understanding of the propagation in equipments (in yellow the forward route and in white the operational transmission environments. In this paper, backward route). The point 1 (IFREMER site) indicates the measurements were carried out by measuring the mean received receiver location for all the experimentations and the point 2 power both for several distances over the sea and for a coastal (Le Caro site) notices one other spot, in order to have a point- point-to-point link during long periods. Part of the work to-point fixed connection above the sea. The Brest harbour presented in this paper has been obtained during the Ex’treme presents various types of environments, also defined in paper project. [4] in which four types of environments and five groups of measurements are mentioned. Index Terms: propagation channel measurements, maritime environment, two-ray model, WiMAX.

I. INTRODUCTION As mentioned in [2] and [3], a basic model being able at a first order to apply to the propagation in maritime environment is the two-ray model. This model takes into account one direct path between the transmitter and the receiver and another path refracted by the sea. For this model, it is possible to calculate the level of the received field at a given point, according to: the distance between the transmitter and the receiver, the relative heights of the transmitter and the receiver from the sea, the wave polarization, the working frequency, the conductivity, the relative permittivity of the sea and the transmitted power. 5 Km The present study aims at evaluating the levels of accuracy of this model and at examining two other causes of Fig. 1. Example of routes (Max Tx-Rx distance = 27 km). modification of the propagation channel:

• the masking effects due to the coastal relief and B. Measurement Equipments islands situated between the transmitter and the At the transmitting site (the boat), two kinds of signals may receiver, be generated: a continuous sine wave or a pulsed one. The system involves two generators, one for the carrier frequencies 3.5 GHz or 5.8 GHz and the other one for the pulse shaped

signals. The pulsed wave is built with a switch and a band pass transmitter and the receiver and a way refracted by sea surface. filter. Two horn antennas are installed on the boat, the first at a We consider this model for simulate the propagation over the height of 12 meters above sea level and the second one at a sea, the figures 4 and 5 show simulation examples. height of 3 meters above sea level. At the receiving site, two Field attenuation, horizontal polarization RED, vertical polarization GREEN, systems were available, namely a spectrum analyzer controlled 3.5 GHz, hTx = 12 m, hRx = 52 m, sea 1/d slope BLUE, 1/d**2 slope CYAN by a computer and a second system build with PXI cards. A 20 controller, a signal analyzer, a frequency converter and a data 0 storage device constituted this last one. This acquisition -20 system can record the received signal with 80 dB magnitude range and a 20 MHz bandwidth centred on a programmable -40

frequency included between a few tens kHz and 6 GHz. -60

These two systems received the signal transmitted by means Field (dB) -80 of a horn antenna followed by a filter and a low noise amplifier. This antenna was placed at the point number 1 with -100

a height of 52 meters. -120 The antennas were 12 dB horn antennas with 120° horizontal

0 1 2 10 10 10 aperture and 30° vertical aperture. A GPS equipped a PC on Length Tx-Rx (m) the boat. The time and the position data were recorded for Fig. 4. Simulation with two-ray model: level of the received each trip. These data make possible to readjust the recordings field according to the length Tx-Rx. of the signal received according to the real distance between

the transmitter and the receiver. These data were confirmed by Field attenuation, horizontal polarization RED, vertical polarization GREEN, 3.5 GHz, hTx = 12 m, hRx = 52 m, sea 1/d slope BLUE, 1/d**2 slope CYAN the equipment of the ship. The GPS information was also used 20 for clock synchronization of the Tx an Rx devices. 0 The following pictures represent the devices used for the experimentations. -20

-40

-60 Field (dB)

-80

-100 Fig. 2. Transmitting devices.

-120

3 4 10 10 Length Tx-Rx (m) Fig. 5. Simulation with two-ray model: level of the received field according to the length Tx-Rx.

From the simulated results we can notice what follows: • the average slope of path loss is proportional to the Tx- Rx distance approximately up to the radio electric horizon, as for a connection in free space, • the average slope of path loss is proportional to the Fig. 3. Receiving devices. squared of the Tx-Rx distance, approximately beyond the radio electric horizon, • the horizontal polarization presents major and recurrent fading (up to 60 dB) depending on the Tx-Rx distance, • the vertical polarization presents major and recurrent fading (up to 30 dB) depending on the Tx-Rx distance,

• the fading of horizontal and vertical polarizations are Throughout this paper, all the presented results concern the phase opposite until a distance of few hundred meters, 3.5 GHz frequency. • the distance interval between fading for both the horizontal and vertical polarizations exceeds the hundred EASUREMENT RESULTS III. M meters for transmitter-receiver distances greater than approximately 1 km. A. Theoretical Results As indicated in the introduction a basic model being able to B. Experimental Results apply to the propagation in maritime environment is the two- These preceding characteristics were verified during ray model. It takes into account a direct way between the experiments and some measurement results are presented now. Geographical representations have been obtained from some obtained results. Figure 8 shows a trip with an important “Google Earth”. masking effect and figure 9 shows the measurement results Figure 6 represents an example of route during which received field levels on this way. measurements were carried out.

1 Km 1 Km

Fig. 6. Example of route where the measurements were carried Fig. 8. Example of route with an important masking effect. out (Max Tx-Rx distance = 14 km). 25 03 2008

A The following figure shows measurement results for the level of the received field, on this “mixed” course, corresponding to -40 the two-ray model previously described but involving also D some partial or total masking effects. -50 Results shown figure 7 compares the levels of theoretical values from the two-ray model, and measured ones of the -60 Field (dB) received signal for a distance from A to C. -70

26 03 2008 Simulation : f = 3.6 GHz, h Tx = 12 m, h Rx = 52 m -80 C -10 B Measurement Theorical value -20 -90 5 10 15 20 25 30 Time (Minutes) -30 Fig. 9. Received field level with an important masking effect.

-40 Sailing behind and near from point B causes a sudden -50 fading of the field level (approximately 40 dB). The cause of Field(dB) this is the important height of this obstacle (76 meters). Other -60 experimentation is shown with the figures 10 and 11. It is a long way where an important cliff is located between the -70 A => C B starting point and the arriving point.

-80 3 4 10 10 Length (m) Fig. 7. Level comparison of the measured received signal with simulated values. In blue: measurements, in green: simulations

There is a good agreement between the theoretical values of field attenuation and the measurements. The level differences at the beginning of the travel are mainly due to antenna effects. Indeed, the Tx and Rx antennas has only 30 degrees vertical apertures, and the receiving antenna at point A was located at 52 meters above the sea, whereas the transmitting antenna was onboard a small ship. The sharp level fluctuation at the end of the record (position B) is a masking effect due to a small 1 Km island.

For a better understanding of the masking effects, other Fig. 10. Example of measurement location with an important experimentations were realized. The following figures show masking effect. 27 03 2008 Tide at BREST harbour 4/07/08 13h - 6/07/08 12h -10 8 C -20 7

-30 6

-40 B 5 -50 4

Field (dB) -60 Height (m) 3 A -70 2 -80 1 -90 0 5 10 15 20 25 30 0 Time (Minutes) 16 20 0 4 8 12 16 20 0 4 8 12 Fig. 11. Level received field with an important masking effect. Hours (U.T) Fig. 13. Tide according to Universal Time. Here two effects are opposed: increase of the field level with the reduction of the transmitter-receiver distance and the Obviously, these tide effects also exist for radio links masking effect growing when the boat gets closer the coastal between a boat and the ground and should be taken into point. In this configuration, it is the first that prevails slightly. account. At the point B, the mask effect causes a reduction of 25 dB as previously indicated to the passage behind the island. IV. DATA ANALYSIS AND SYSTEM IMPACTS C. Tide effects The theoretical results and the experimental validations show In the two-ray model, the relative heights of Tx and Rx that the propagation channel in maritime environment has antennas are important characteristics in calculation equation. diversity at three dimensions: • In our situation, the effective antenna heights of course varied a polarization diversity which is present until with the tide effects, which imply a modification of the indirect distances ranging between 10 meters and 500 meters. path. To evaluate this phenomenon the figure 12 presents the Experimental studies in cross polarizations must be results of the field level received over a radio link between Tx made to consolidate this observation, • and Rx on the ground, but above sea. Figure 13 presents the the vertical polarization mode is better adapted hours of tides corresponding to the measurement period. starting from distances of a few hundred meters, • These figures show the strong incidence of tides on the a horizontal space diversity (and temporal for a point-to-point fixed connection above sea. We note that for transmitter and/or a receiver in mobility). This this link geometry, the maximum of the field levels were diversity, inversely proportional to the Tx-Rx obtained for tide heights from approximately 1 and 6 meters. distance is thus reduced with this distance, The minimum of the field levels were obtained for tide heights • a vertical space diversity proportional to the relative from approximately 2.5 meters. height variations of the transmitting and receiving Other descents of the field levels occur near 7 meters of tide antenna. heights, indicating that in the event of strong tide (over 8 The results confirm also that the masking effects are more or meters), a minimum of field level would again be reached. less important according to the distance relatives to the masks. Other complementary results confirm the influence of the sea Moreover, the influence of the sea states is certain on two state but this influence requires to be better quantified by new levels and request still to be quantified in experiments: studies and experiments. • pointing defaults of the receiving and/or transmitting antennas [5] and [6], Ifremer -> Le Caro 4/07/08 13h - 6/07/08 12h • modification of the refraction conditions of the -60 indirect way. Studies and experimental complements are necessary to -65 refine the description of the channel behaviour, in particular for the presence of obstacles (coastal relief, large ship …) near -70 one or both connection ends, being able to create complex situations of multi-paths.

Field (dB) -75 To take benefit from these diversities of the propagation

-80 channel in maritime environment, the following solutions on the systems could be examined: • throughput and/or emission power adaptability according -85 16 20 0 4 8 12 16 20 0 4 8 12 to the reception level and/or the signal to noise ratio Hours (U.T) Fig. 12. Received field level according to the Universal Time. (thus depending from distance and masking effects), • taking advantage of the natural channel diversity: o using two transmitting and/or two receiving B. ImaginLab project antennas with orthogonal polarizations (vertical ImaginLab project is a test and experimental regional and horizontal), with a reception strategy to be platform, adopted by FUI 6. It will make it possible to test new determined (combining, best level, …). This products and services. The infrastructure design and solution would be useful until approximately installation are carried out jointly by the pole Images & 1000 m (in harbour and harbour approach). Réseaux and the UEB (Université Européenne de Bretagne), Beyond 1000 m, vertical polarization is this later one having delegated its mission to Télécom preferable, Bretagne. o take advantage from channel horizontal diversity The infrastructure is made up of three inter-connected by using two (or more) receiving antennas technological platform. The “Internet du futur” platform, approximately distant of half of the maximum located in Lannion, works on IMS (IP Multimedia Subsystem) interval between two fading, with a reception architecture. The “Images en mobilité” platform, located in strategy to be determined (combining, best level, Rennes, will be used to diffuse mobile television services. The …). But the effectiveness of the solution “haut débit sans fil” platform, located in Brest, will make it depending on the Tx-Rx distance is proportional possible to test high-speed services on radio links, such as to the size of the ships concerned. applications intended for the professionals or accessible o exploit the channel vertical diversity by using two services from a boat located near the coasts. (or more) transmitting antennas and/or two (or In this project, Télécom Bretagne uses its knowledge in the more) reception antennas separated by the height wireless transmission domain. These studies will relate to the from approximately half of the maximum interval wave propagation studies in different environments (urban, between two fading in altitude, with a reception suburban, white zones and maritime areas). strategy to be determined (recombination, best level, …). Vertical and polarization diversities C. Palmyre project could be combined. The solution would be The Palmyre project consists in a realization of a modular effective for connections between two points at platform for the evaluation of both systems and new ground and above the sea, for connections communication concepts in frequency bands going from HF to between a ground point and a ship, and for microwaves, within realistic experimental situations. connections between two ships, with however a This platform is being developed by a hardware and software respectively decreasing efficiency for these three infrastructure group cases. Partners of this project are teams from the “ Institut • control the Tx-Rx antenna pointing. d'Électronique et de Télécommunications” in Rennes, the “Laboratoire d'Électronique des Systèmes Temps Réel” from V. LINKED MARITIME PROJECTS University of Bretagne Sud, the “Micro-ondes, Electronique and Signal & Communications departments” from Télécom A. Ex’treme project Bretagne and the “Université de Bretagne Occidentale” in This project proposes the implementation of a broadband Brest. connectivity between the ground and the ships on sea. It will D. Navtrack project offer broadband services, mainly to increase the safety on the sea and to exchange information at short distances from the The general concept of this project is constituted of two coast. technical guidelines: • Moreover this project is based on complementary radio the first consists in locating in real-time all the boat technologies to ensure IP services towards the passengers and race actors, using a radio transmission (868 MHz) to the members of crew. The Ex’treme project is pioneer in the transmit GPS data, which are decoded on ground. A sector of the broadband services on the sea. This is because the suited cartography traces the trajectories. The virtual existing IP Mobile service offers are not yet really dedicated to images are enriched by environmental data (wind, the maritime field. currents, tide…) • The public concerned is mainly: pleasure boats, passenger the second aims at transmitting on line the images of liner, fishing vessels and ships of service. Moreover a great the boat race by a radio transmission at 5.8 GHz. range of services will be proposed to these users: information Partners of this project are teams from “DETI”, “Ecole on safety, the management and diffusion of an alarm, video Navale – IRENav”, “AGESSI”, “Technopole Brest Iroise”, transmissions, video monitoring, videoconference and “Electronique department from Télécom Bretagne” in Brest applications of collaborative work, Internet access... and from “ SeatizenPro” in Villeneuve-Loubet. The partners of this project are: Alcatel Business Systems, Thomson-GrassValley, IFREMER, C2 Innovativ’Systems, VI. CONCLUSION AND PERSPECTIVES Morgan’Conseil, Chantiers de l’Atlantique, and Télécom This paper presented experimental researches carried out by Bretagne. the CaPSyS team of Télécom Bretagne on the new radio technologies in the maritime environment. This characterization of the propagation channel is necessary to understand the behaviour of radio waves in a maritime VII. REFERENCES environment. Moreover, with this study, the performances of WiMAX systems can be analyzed according to the [1] J. Joe, S.K. Hazra, S.H. Toh, M.W. Tan, J. Shankar, V.D. Hoang, M. characteristics enumerated above. It is thus possible to take Fujise, “Path Loss Measurements in Sea Port for WiMAX”, Wireless Communications and Networking Conference, 2007. WCNC 2007. advantage of this analysis to propose some improvements of IEEE, 11-15 March, 2007. the transmission systems for better performances. [2] Rosario G. Garroppo, Stefano Giordano, Davide Iacono, Alessandro Télécom Bretagne develops new testing systems, in Cignoni, Matteo Falzarano, “Wimax Testbed For Interconnection Of particular multi-channel receiving and transmitting systems, in Mobile Navy Units In Operational Scenarios”, Military Communications Conference. MILCOM 2008. IEEE, 16-19 Nov, 2008. order to explore the channel diversities quoted previously [3] Yattoun, I.; Labia, T.; Peden, A.; Landrac, G.; Ney, M.; Resibois, M.; (MIMO spatial and polarization diversities) and for a better Bonnin, J.M.; Baghdadi, A.; Montavont, N.; Fujise, M.; Le Roux, Y, “A understanding of the propagation mechanisms in operational Millimetre communication system for IVC”, ITST 07, 7th International transmission environments (terrestrial as well as maritime). Conference on ITS, 6-8 June 2007. [4] Konstantinos N. Maliatsos1, Student Member IEEE, Panagiotis Loulis1, WiMAX (or others) transmission systems will also be operated Michail Chronopoulos, Philip Constantinou1, Member, IEEE Panagiotis simultaneously with channel sounding equipments, in order to Dallas2, Michail Ikonomou, “Measurements and Wideband Channel compare the propagation channel characteristics with Characterization for Over-the-sea Propagation”, Wireless and Mobile transmission performances. Computing, Networking and Communication, WiMob 2006. IEEE International Conference, 19-21 June 2006. [5] Su Wen, Peng-Yong Kong, Jaya Shankar, Haiguang Wang, Yu Ge, These studies have been partially funded by: Brittany Chee-Wei Ang, “A novel Frame work to simulate Maritime Wireless Regional Council, Finistere Council, Brest Metropole Communication Networks”, OCEANS 2007, Sept. 29 2007-Oct. 4 Oceane and Europe (FEDER). 2007. [6] Chee-Wei Ang and Su Wen, “Signal Strength Sensitivity and Its Effects on Routing in Maritime Wireless Networks”, Local Computer Networks, 2008. LCN 2008. 33rd IEEE Conference.

View publication stats Journal of Communications Vol. 10, No. 5, May 2015

Maritime Radio Propagation with the Effects of Ship Motions

Fang Huang, Yong Bai, and Wencai Du College of Information Science and Technology, Hainan University, Haikou, 570228, China Email: [email protected]; [email protected]; [email protected]

Abstract—For designing maritime wireless transmission ship motions with a 3D antenna gain model [8]. Therein, system, the radio propagation over sea surface needs to be the numerical results of the receive power was given only known first. One of the distinct characteristics of maritime radio for the up-down ship motion at 2.4GHz carrier frequency propagation is the impact of ship motions due to the fluctuation with a fixed transmission distance. Nevertheless, it still of sea waves. This paper establishes a radio propagation model lacks the study of maritime radio propagation with the with the integration of the effects of ship motions. Using such an integrated radio propagation model, the maritime radio effects of ship motions under different transmission propagation characteristics are analyzed and discussed under distances, different carrier frequencies, and different different transmission distances, different carrier frequencies, motion types. To further investigate the maritime radio and different motion types. propagation with the effects of ship motions, this paper first improves the traditional two-ray propagation model Index Terms—Radio propagation, Maritime communications, Channel modeling by taking into account the earth curvature, and then establishes a radio propagation model by integrating the 3D antenna gain model of ship motions with the I. INTRODUCTION improved two-ray propagation model. Using such a model, the maritime radio propagation characteristics are In the ship-to-ship and the ship-to-shore wireless analyzed and discussed under different transmission communications over the sea, the radio signal propagates distances, carrier frequencies, and different motion types. over the sea surface. Hence, the radio propagation over The rest of this paper is organized as follows. Section sea surface needs to be studied first for designing the II describes the maritime radio propagation path. Section maritime wireless transmission system. One of the III improves traditional two-ray propagation model by distinct characteristics of maritime radio propagation is taking into account the earth curvature. Section IV the impact of ship motions due to the fluctuation of sea establishes the radio propagation model by integrating a waves. The angle between the transmit antenna and the 3D model antenna direction gain model of ship motions receive antenna changes with ship motions, which results with the improved two-ray propagation model. Section V in the fluctuations of the received power at the receive presents the numerical results of the radio propagation antenna. In previous investigations on maritime radio with the effects of ship motions using the established propagation models, several deterministic models such as model. Section VI concludes this paper. the Free Space Loss (FSL) model and the Plain Earth Loss (PEL) model based on Friis transmission formula II. MARITIME RADIO PROPAGATION PATH and two-ray tracing method have been commonly used as references for the open-sea environment [1]-[4]. In Frist, we discuss the radio propagation path over sea addition, the distance between the transmitter and the surface. For over-the-sea transmission to and from ships, receiver can be far in the maritime radio transmission the effect of earth curvature on the radio propagation environment, and the effect of the earth curvature cannot needs to be taken into account. Fig. 1 illustrates the radio be ignored. Another deterministic path-loss model for the propagation path over sea surface between a terminal on open-sea environment was proposed in accordance with ship and a base station on land. The ship-to-shore radio measurements at 2 GHz with a maximum distance of 45 propagation distance can be divided into three segments km [5]-[7]. The proposed model accounts for different according to the distance between the RF transmitter and effects including effective reflection, divergence, and receiver: segment A, which is from T (the point of the diffraction due to rough sea and earth curvature. However, base station) to RA (the sightline of the base station) with the abovementioned studies have not considered the length d1; segment B, which is from RA to RB (the effects of ship motions. Hubert et al. presented a maritime sightline of the terminal) with length d2; and segment C, radio link channel simulator and studied the impact of which is the shadow area beyond RB [9]．Segment A is a line-of-sight path through free space, with no obstacles nearby to cause reflection or diffraction. In the segment B Manuscript received January 12, 2015; revised May 13, 2015. and C, diffraction arises because of the curved way in Corresponding author: Wencai Du, email: [email protected] doi:10.12720/jcm.10.5.345-351 which waves propagate.

A B receiver, Rd can be calculated according to the time delay, d1 d2 RA RB T C then the value of  can be calculated by d Hr 3 Ht 222 1 ht R e  h r  R e  R d Hr   cos (5) RC 2 h R h R  t e r e 

Re In [14], l1 can be calculated by the flowing formula when we assume the height and size of the antenna is much smaller than the earth radius,

2l3 3 ll 2  ( l 2  2 Rhhl (  ))  2 Rhl  0 (6) 1 1e r t 1 e r

The value of 1 and 2 can be calculated by

  lR/ (7) Fig. 1. Three distance segments of maritime radio propagation 11e Assume that the antenna heights of base station and  ()/l l R (8) 21e terminal are Ht, and Hr, respectively. Using trigonometry, Then, R and R can be calculated by it can be calculated that [10] 1 2

2 2 2 22 d R () H  R (1) R11 Re ( R e  h t )  2 R e ( R e  h t )cos (9) 1 e t e d2 R H H 2 (2) 1 e t t R R22 ( R  h )  2 R ( R  h )cos (10) 22e e r e e r where Re is the earth radius, and Re=8500km. Because

RHet , we can get Finally, in [13], the grazing angle considering the earth curvature can be calculated by d 2 R H (3) 1 et 1 hRr 2 Similarly, g sin  (11) RR2 2 e

d 2 R H (4) 2 er

Rd III. IMPROVED TWO RAY PROPAGATION MODEL R1 ht R2 hr The reflection effect from the sea surface and the g l scattering effect due to the roughness of sea waves result Tangent plane of in multipath components of the transmitted signal at the reflection point l1 l2 receive antenna. The random phase and amplitudes of different multipath components cause fluctuations in signal strength [11]. In [12], the roughness factor judgment, Rayleigh judgment, coherent reflection R coefficient method and specular and diffuse reflection e coefficients method are used to analyze the reflection characteristics of electromagnetic waves over the sea. It is  concluded that the sea surface can be assumed smooth 12 mirror surface when the sea state is seven and the grazing angle is from 0 to 1.3 degrees, or the sea state is six and Earth center the grazing angle is from 0 to 2.5 degrees. This paper assumes that the sea surface is calm, and only considers Fig. 2. Two-ray model considering the earth curvature the effect of specular reflection. Thus, a two-ray model can be used for analyzing the radio propagation over sea IV. INTEGRATED MARTIME RADIO PROPAGATION MODEL surface. As shown in Fig. 2, a two-ray model of radio WITH SHIP MOTIONS propagation includes a direct path and a specular reflection path. The direct signal path is the line of sight A. Ship Motion Modeling (LOS) signal propagation between the transmitter and the Surface fluctuations are generally divided into three receiver. The specular reflection path is produced by the types: waves caused by wind, tidal caused by gravity and reflection from the smooth sea surface [13]. centrifugal force, and the tsunami caused by tectonic. To make the model more precise, we modify the two- This paper only considers the wave motions caused by ray model considering the earth curvature as shown in Fig. wind [15]. The directions of ship motions can be up and 2. In general, the distance between the transmitter and the down, left and right, front and rear. It can be described

with the motion model (,,,,,)x y z    of six degrees of where  and  are elevation angle and azimuth angle, freedom [16]. A 3D coordinate system (,,)x y z , in which respectively. As shown in Fig. 5, U is a normalized vector O is the Earth center, is shown in Fig. 3. Heave is a which corresponds to the strength ratio emitted (or translation along z axis (up and down), and roll is a received) along U and U components. A database of rotation about x axis (left and right), and pitch is a antenna gains for simulation can be created using HFSS, rotation about y axis (front and rear). an electromagnetic (EM) simulator. z k A1

u ur

u

y  u j A1 x u  Fig. 3. ship in 3D coordinate system

The variation range of the heave motion is Hmax which is the crest-to-trough wave height. The variation ranges of i both the pitch motion and the roll motion can be A1 Fig. 5 Antenna coordinate system expressed by max , which is the antenna maximum angular deviation from vertical direction, as show in Fig. In the coordinate system, the direct-path channel 4. Without real ship motion records, approximate matrix CD can be calculated by geometrical relations can be derived for max . It can be calculated by 1 ABBB CGGUUD   ... R  d  Hmax (14) max =arcsin (12) ABABA 2 2 2 u u  u  u    U    jR2/ s+ H max e d ABABA    u u  u  u    U   where s is the wavelength of sea wave. Using the fundamental mode of the Pierson-Moskowitz spectrum where u and u present the unit vectors along the for fully developed regular wind waves [17], a complete angles of  and . data set is deﬁned for the studied case. It includes the i period T  7s, and λ  131.4m. At the sea condition s s u // B Douglas 5, Hmax is about 5.7m. According to (12), we can A =7.7 .  get θmax °  r i R1 R2 u u  P   max Hmax

wavelength s Fig. 6. Unit vector definition for specular reflection

Fig. 4. Variation ranges of sea waves The reflected-path channel matrix CR can be calculated by B. Integrated Propagation Model with the Effects of Ship Motions 1 ABBB CGGUUR   ... Then we analyze the change of antenna gains of the RR12 transmitter and receiver when ships move on the sea uB u r u B u r  0 surface. A radiation vector function G is introduced to //     // ... (15) B r B r FD r  account for the polarization state and the antenna gain u u// u u 0  along any direction [18]. G(,) depends on the realized A r A r A u u//// u  u   U   gain for a given set of departure (emitting antenna) or      ej2 ( R12 R )/ uA u r u A u r   U A  arrival (receiving antenna) angles,      

r r U (,) where u and u are shown in Fig. 6;  is the factor GUGG(,)(,)    (13) //  FD  U (,) of energy dispersion [19] and it can be calculated by

1/2 1/2 1 22RRRR CGGUUABBB ... 1 2 1 2 (16) R  FD r FD 11   RR RRRRRR( )sin ( ) 12 e1 2 g e 1 2 (23)  0 U A //  j2 ( R12 R )/ For predicting the specular reflection coefficient r , A e 0  U Ament presented in [20] that it can be calculated by  

22 r exp  2(2 g )  I0  2(2  g )  (17) V. NUMERICAL RESULTS In this section, we study the maritime radio g  (hg sin  ) /  (18) propagation characteristics under different transmission where  h is the rms surface height variation, g is the distances, different carrier frequencies, and different grazing angle of incident,  is the wavelength, and I0 motion types (heave, roll, and pitch) by using the represents the modified Bessel function of zero order. integrated radio propagation model presented in Section Next, the power strength at the receiver can be IV. calculated as We assume the antennas at the transmitter and receiver 2 for wireless communications are both half-wavelength P  2 P( dB ) 10logr  10log ( C  C ) dipole antennas with maximum gain 2.1 dBi. The heights 10 10 DR (19) Pt 4 of transmitter and receiver are both 10m. With such  assumptions, the range of segment A is 13 km, and the 20log10 (CCDR ) 4 segment B is 23 km according to (1) to (4). Furthermore, we assume that the transmitter is on the shore, when the In segment A, two-ray radio propagation model is used receiver is located in the sea, and the distance between for analysis. In segment B, one-ray model can be used them is set to 1 km, 5 km, 10 km and 20 km, respectively. since the reflected path can be ignored. The propagation Finally, we suppose that the sea condition is Douglas 5, path losses for segment A and B can be written as and the surface fluctuation cycle is 7 seconds.

La 147.5582  20lg f   ( f , d ) fc=700MHz fc=2.4GHz (20) -85 -90 147.5582  20lgf  20lg CDP  C RP -90 -95 -100 -95 Lb147.5582  20lg f  20lg C DP (21) -105 We also can obtain the direct-path channel -100 -110 Power(dB) Power(dB) matrix C without motions by the following equations, -105 D -115

A -110 -120 1 U C GABBB G U U e jR2/d (22) D  A -115 -125 R U 0 5 10 15 20 0 5 10 15 20 d  Rd (km) Rd (km) In the same way, we obtain the reflected-path channel Fig. 7. Received power versus transmission distance when ship is motionless matrix CR without motions by the following equations,

fc=700MHz fc=2.4GHz Rd=1km Rd=5km Rd=7km -80 -90 Rd=10km Rd=13km -90 -100 Rd=15km Rd=17km -100 -110 Rd=20km

Power(dB) -110 Power(dB) -120

-120 -130 0 0 20 20 10 10 Rd (km) 10 Rd (km) 10 t (s) t (s) 20 0 20 0

Fig. 8. Received power versus time and transmission distance with ship heave (up and down) motion

Using the integrated radio propagation model, we Fig. 7 shows the received power versus the present the numerical results of maritime radio transmission distance for fc=700MHz and fc= 2.4GHz propagation under different transmission distances, without ship motions. Figure 8 shows the received power different carrier frequencies (700 MHz and 2.4 GHz), and versus time and the transmission distance for fc=700MHz three motion types (heave, roll, and pitch). The numerical and fc= 2.4GHz under ship heave (up and down) motion. results are shown in Fig. 7 to Fig. 11. The horizontal distance between the transmitter and

receiver is from 1km to 20km. It can be seen that with the impact of ship motions on radio propagation become increase of transmission distance, the fluctuations of smaller at far transmission distance. received power become smaller, which implies that the fc=700MHz fc=2.4GHz -80 -90 Rd=1km Rd=1km -90 (a) -100 (a) -110 -100 0 5 10 15 20 25 0 7 14 21 -95.5 -100 Rd=5km Rd=5km -96 (b) -110 (b) -120 -96.5 0 5 10 15 20 25 0 5 10 15 20 25 -101.98 -113 Rd=10km Rd=10km -101.99 (c) -114 (c)

-115 Power(dB)

Power(dB) -102 0 5 10 15 20 25 0 5 10 15 20 25 -113.1069 -123.8092 Rd=20km Rd=20km -113.1069 (d) -123.8092 (d)

-113.1069 -123.8092 0 5 10 15 20 25 0 5 10 15 20 25 t (s) t (s) (i) (ii) Fig. 9. Received power versus time at different transmission distances with ship heave (up and down) motion

fc=700MHz fc=2.4GHz -95 -100 Rd=1km Rd=1km -95.5 (a) -101 (a)

-96 -102 0 5 10 15 20 25 0 5 10 15 20 25 -95.92 -111.4 Rd=5km Rd=5km -95.93 (b) -111.6 (b)

-95.94 -111.8 0 5 10 15 20 25 0 5 10 15 20 25 -101.99 -113.7 Rd=10km Rd=10km (c) -113.75 (c)

Power(dB)

Power(dB) -101.995 -113.8 0 5 10 15 20 25 0 5 10 15 20 25 -113.106 -123.808 Rd=20km Rd=20km -113.107 (d) -123.809 (d)

-113.108 -123.81 0 5 10 15 20 25 0 5 10 15 20 25 t (s) t (s) (i) (ii) Fig. 10. Received power versus time at different transmission distances with ship roll (left and right) motion

fc=700MHz fc=2.4GHz -95 -100.5 Rd=1km Rd=1km -95.5 -101 (a) (a) -96 -101.5 0 5 10 15 20 25 0 5 10 15 20 25 -95.92 -111.4 Rd=5km Rd=5km -95.93 -111.6 (b) (b) -95.94 -111.8 0 5 10 15 20 25 0 5 10 15 20 25 -101.991 -113.7 Rd=10km Rd=10km -101.9912 (c) -113.75 (c)

Power(dB) -113.8 0 5 10 15 20 25 Power(dB) 0 5 10 15 20 25 -113.1069 -123.8091 Rd=20km Rd=20km -113.1069 (d) -123.8092 (d)

-113.1069 -123.8092 0 5 10 15 20 25 0 5 10 15 20 25 t (s) t (s) (i) (ii) Fig. 11. Receiver power versus time at different transmission distances with ship pitch (front and rear) motion

Fig. 9, Fig. 10, and Fig. 11 show the receiver power and the power loss resulted from the ship motions of roll versus time at different transmission distances under ship and pitch is only about 0.2 dB. Therefore, the heave heave, roll, and pitch motions, respectively. In these motion has the largest impact on the transmission path figures, the propagation model equation of segment A is loss for maritime radio propagation compared to the other used when the transmission distance is 1km, 5km and two motion types. 10km; the propagation model equation of segment B is Comparing the sub-graph (d) in Fig. 9 to 11 where the used when the transmission distance is 20km, which is transmission distance locates in Segment B, we observe longer than 13 km (the boundary point of segment A and that the fluctuations of received power are all relatively B). Comparing the sub-graph (a) in Fig. 9, Fig. 10, and small under three motion types. Thus, the effects of ship Fig. 11, we observe that the power loss caused by ship motions on the received power are relatively small in motions of roll and pitch is lower than that caused by Segment B, and can be negligible. heave motions. As shown in Fig. 9-i(a) and Fig. 9-ii(a), Comparing with the received powers at carrier the power loss resulted from heave motion is about 9 dB, frequencies of 700MHz and 2.4GHz, we can see that the

fluctuations of received power become larger at higher [7] K. Yang, A. F. Molisch, T. Ekman, and T. Roste, "A deterministic carrier frequency. Thus, the impact of ship motions on the round earth loss model for open-sea radio propagation," in Proc. IEEE Vehicular Technology Conference (VTC Spring), 2013, pp. radio propagation is more significant at higher carrier 1-5. frequencies. For instance, as shown in Fig. 9-i-(a) and Fig. [8] W. Hubert, Y. M. Le Roux, M. Ney, et al., “Impact of ship 9-ii-(a), when the transmission distance is 1km, the motions on maritime radio links,” IEEE Trans. International fluctuation of received power is about 9dB at 700MHz, Journal of Antennas and Propagation, vol. 2012, pp. 1-6, and the fluctuation of received power is about 16dB at Nov.2012. 2.4GHz. [9] Y. S. Meng and Y. H. Lee, “Measurements and characterizations of air-to-ground channel over sea surface C-band with l ow airborne altitudes,” IEEE Transaction on Vehicular technology, VI. CONCLUSIONS vol. 60, no. 4, pp. 1943-1948, 2011. To study the maritime radio propagation with the [10] T. S. Rappaport, Wireless Communications: Principles and Practice, Prentice Hall, January, 2002, pp. 23-50. effects of ship motions, this paper first modified [11] F. P. Fontan and A. Rocha, “Estimation of the number of clusters traditional two-ray propagation model by taking into in multipath radio channel data sets,” IEEE Transactions on account the earth curvature, and then established a radio Antennas and Propagation, vol. 61, no. 5, pp. 2879-2883, 2013. propagation model by integrating a 3D model antenna [12] M. Dong, Y. B. Zhao, and S. H. Zhang, “The analysis of the direction gain model of ship motions with the improved multipath model under the VHF band and at sea,” Acta Electronic Sinica, vol. 36, no. 6, pp. 1373-1377, 2009. two-ray propagation model. Using such an integrated [13] X. Q. Hu, J. B. Chen, and Y. L. Wang, “Research on metre-wave radio propagation model, the received power with the radar height-finding multipath model,” Chinese Journal of Radio impact of ship motions can be obtained under different Science, vol. 23, no. 4, pp. 651-657, 2008. transmission conditions. From the numerical results, we [14] B. R. Mahafza, Radar Systems Analysis and Design Using analyzed the radio propagation at different transmission MATLAB, Boca Raton: CRC Press, 2000. [15] A. Fung and K. Lee, “A semi-empirical sea-spectrum model for distances and carrier frequencies under different motion scattering coefficient estimation,” IEEE Transaction on Oceanic types. We draw the conclusions that the impact of ship engineering, vol. 7, no. 4, pp. 166-176, 1982. motions on radio propagation becomes smaller with the [16] X. Yang and X. H. Wang, “Modeling and simulation research of increase of transmission distance, and the impact of ship six-degree-of-freedom fighter,” Journal of System Simulation, vol. motions on the radio propagation is more significant at 12, no. 3, pp. 210-213, 2000. higher carrier frequencies; Lastly, the heave (up and [17] L. Gardenal, B. Philibert, and R. M. Turner, “Study of the interaction of electromagnetic waves on a sea surface: down) motion has the largest impact on the transmission Improvement of the algorithm of low-altitude tracking of radar path loss for maritime radio propagation compared to targets,” in Proc. Canadian Conference on Electrical and other two motion types. Computer Engineering, vol. 2, pp. 693–696, 1994. [18] B. Uguen, L. M. Aubert, and F. T. Talom, “A comprehensive MIMO-UWB channel model framework for ray tracing ACKNOWLEDGMENT Approaches,” in Proc. IEEE Ultra-Wideband Conf., 2006, pp. This paper was supported by the National Natural 231-236. Science Foundation of China (Grant No. 61062006 and [19] R. J. Papa, J. F. Lennon, and R. L. Taylor, “Multipath effects on an azimuthal monopulse system,” IEEE Transactions on Grant No. 61261024) and the Special Social Service Aerospace and Electronic System, vol. 19, no. 4, pp. 585-597, July Project Fund of Hainan University, China (Grant 1983. No.HDSF201301). [20] A. R. Miller, R. M. Brown, and E. Vegh, “New derivation for the rough-surface reflection coefficient and for the distribution of sea- REFERENCES wave elevation,” in Proc. IEEE H(Microwaves, Optics and Antennas), vol. 131, no. 2, pp. 114-116,1984. [1] N. H. Lu, “Linearized, unified two-ray formulation for propagation over a Plane Earth,” in Proc. Sensor for Industry Fang Huang received the B.S. degree from Conference, 2005. Anhui Normal University, China, in 2012. [2] J. Joe, S. K. Hazra, S. H. Toh, W. M. Tan, and J. Shankar, “Path She is currently pursuing her M.S. degree at loss measurements in sea port for WiMAX,” in Proc. IEEE the College of Information Science & Conference on Wireless Communications and Networking Technology, Hainan University. Her research Conference, 2007. interests include radio channel modeling, and [3] ITU-R Recommendation P.1546-2, “Method for point-to-area wireless communications. predictions for terrestrial services in the frequency range 30 MHz to 3000 MHz,” Sep. 2005. [4] L. J. Zhang, H. G. Wang, R. Zhang, et al., "Radio wave propagation characteristics measurement and modeling over the Yong Bai received his B.S. degree from sea," in Proc. General Assembly and Scientific Symposium (URSI Xidian University, China, in 1992, and M.S. GASS), 2014, pp. 1-4. degree from Beijing University of Posts and [5] K. Yang, T. Roste, F. Bekkadal, and T. Ekman, "Experimental Telecommunications (BUPT), China, in 1995, multipath delay profile of mobile radio channels over sea at 2 and Ph.D. degree from Rutgers-The State GHz," in Proc. Antennas and Propagation Conference (LAPC), University of New Jersey in 2001. He was 2012, pp. 1-4. with PacketVideo Corporation from 2000 to [6] K. Yang, T. Roste, F. Bekkadal, et al., "Long-distance propagation 2002. He was with Motorola from 2002 to measurements of mobile radio channel over sea at 2 GHz," in Proc. 2004. He was with CEC Wireless from 2004 IEEE Vehicular Technology Conference (VTC Fall), 2011, pp. 1-5. to 2005. He was a senior researcher at DOCOMO Beijing

Communication Labs from 2005 to 2009. He is a professor at College of Wencai Du received the B.S. degree from Information Science & Technology, Hainan University since 2010. He Peking University, China, two M.S. Degrees acted as the Lead Guest Editor for EURASIP Journal on Wireless from ITC, The Netherlands, and Hohai Communications and Networking, Special Issue on Topology Control in University, China, respectively, and Ph.D. Wireless Ad Hoc and Sensor Networks. His current research interests degree from South Australia University, include mobile communications, and maritime communications. He is a Australia. He was a Post-doctor Fellow in member of the IEEE. Israel Institute of Technology (IIT), Haifa, Israel. He is Dean of College of Information Science & Technology at Hainan University and Director of Maritime Communication and Engineering of Hainan province. He has authored or coauthored 18 books and more than 80 scientific publications. He is currently members of the Editorial Board of Inverts Journal of Science and Technology, India. He has taken services on many professional conferences, including Conference Chair of IEEE/ACIS ICIS 2011, Conference Co-Chair of IEEE/ACIS SNPD 2010, London, Conference Chair of IEEE/ACIS SERA 2009, and Program Chair of IEEE/ACIS SNPD 2009, Daegu, Korea. His research interests include several aspects of Information Technology and Communication (ITC), including computer network and maritime communications.

Evaluation of Radio over Sea Propagation Based ITU-R Recommendation P.1546-5

Han Wang, Wencai Du, and Xing Chen College of Information Science & Technology, Hainan University, 58 Renmin Ave., Haikou, Hainan 570228, China Email: [email protected]; [email protected]; [email protected]

Abstract—The propagation prediction research has mainly fee, the data rate are far from the main user requirements. focused on studying the radio propagation models in urban or Consequently, a maritime communication needs to be rural land areas. In this paper, we extend a method for 950 MHz further developed. propagation prediction over sea based ITU-R Recommendation In recent years, there have been a lot of researches in P.1546-5. A gauge of the accuracy of the prediction Free Space variety models of propagation prediction, well-known model, correction Okumura-Hata model and the P.1546-5 model is presented. The P.1546-5 is evaluated using propagation models are Longley-Rice model [4], Durkin measurement results which were obtained by utilizing the pilot model and Okumura-Hata model [5]. They are all signal in a commercial GSM network over sea around Haikou, applicable for the land scenario. Research on radio China. The comparison is enabled by using hit rate metrics. propagation model over sea mainly utilize parabolic Measurement results show that P.1546-5 model underestimates equation method [6], [7], it takes into account phenomena the field strength about 6 dB on average for Haikou region in such as refection, refraction and diffraction. While in South China Sea. However, the P.1546-5 prediction model line-of-sight propagation, it is usually regarded as free provides better accuracy prediction of the path loss compared to space propagation [8], for non-line-of-sight propagation, the Free Space model and the correction Okumura-Hata model. the path loss is calculated according to the experience of We provide proposals to enhance the sea propagation prediction accuracy of ITU-R Recommendation P.1546. chart. However, the conclusions of the tradition methods are that none of the evaluated models are comprehensive Index Terms—Radio over sea propagation, Propagation enough to predict radio propagation over sea. Hence, an prediction, Hit rate metrics, ITU-R P.1546-5, Free space model, accurate prediction model for radio propagation over sea Okumura-Hata model needs to be further studied.

The International Telecommunication Union (ITU) has I. INTRODUCTION developed a new Recommendation on the method for point to area predictions for terrestrial services in the Hainan Province is located in the southernmost tip of frequency rang 30 MHz to 3000MHz. The latest China, it comprises about 200 square kilometers sea area Recommendation ITU-R P.1546-5 [10] is intended for which is about fifty-seven times its mainland area. use on tropospheric radio circuits over land paths, sea Numerous activities in the sea area such as offshore oil paths or mixed land-sea paths up to 1000km length for exploitation, maritime transportation and marine fishery effective transmitting antenna heights less than 3000 m. make the maritime communications more and more The method is based on interpolation/extrapolation from important. The current maritime communication models empirically derived field-strength curves as functions of mainly include signal sideband (SSB) short wave radio, distance, antenna height, frequency and percentage time. VHF radiotelephone, coast cellular mobile Currently, the ITU-R P.1546 is increasingly used as a communication network and maritime satellite benchmark propagation method to analyze the radio wave communication network [1]. Maritime VHF radio propagation. In [11], the Longley-Rice, ITU-R P.1546 telephone is mainly used for ship to shore and ship to ship and Hata-Davidson propagation models for DVB-T voice communication scenario while the transmission coverage prediction was compared. In [12], a new distance should be less than 20 nautical miles. Maritime approach for 1 km urban propagation model of the satellite communication system [2], [3], such as the recommendation ITU-R P.1546 was proposed. The ITU- Inmarsat-F system, Fleet-Broadband maritime data R P.1546 is also utilized to combine with other model to service, is suitable for ocean sea ship communications, analyze some special propagation scenario [13]. But the but the satellite communication system is relatively costly, study of ITU-R P.1546 are mainly concentrated on the the terminal equipment is expensive, accompanied with radio propagation in land scenario. higher maintenance and update costs and communication In this paper, we focus on the prediction over sea propagation. We extend the ITU-R P.1546-5 as a radio

Manuscript received January 20, 2015; revised April 10, 2015. over sea propagation model. Propagation measurement This work is supported by the National Natural Science Foundation results are compared with the prediction based of China, Grant No. 61162010. propagation model Recommendation ITU-R P.1546-5. Corresponding author email: [email protected]. doi:10.12720/jcm.10.4.231-237 Some conventional empirical models, such as the Free

Space model and the Okumura-Hata model, are used as h2 path loss prediction benchmark. Hit rate metrics are Correction Kh2 log' dB R introduced to complement conventional first order (1) statistics. where R' 10 m , K 3.2 6.2log(f ) , f is the The rest of this paper is organized as follows: Section h2 2 describes the Recommendation ITU-R P.1564-5 and its frequency (MHz). two main correction factors. Measurement procedure and When the receiving antenna is adjacent to sea for analysis methods are given in Section 3. The evaluation h2 10 m , an alternative method should be used, based of prediction models and measurements are presented in upon the path lengths at which 0.6 of the first Fresnel Section 4. Finally, the conclusion and a discussion for the zone is just clear of obstruction by the sea surface. For a future work are presented in Section 5. given frequency and antenna heights h1 and h2 , the path length which just achieves a clearance of 0.6 of the first II. RECOMMENDATION ITU-R P.1546-5 Fresnel zone over a smooth curved earth is given The ITU-R P.1546 model provides a set of curves and approximately by: tables of field strength as a function of frequency (100 DDfh MHz, 600 MHz, and 2 GHz), distance (1 km to 1000 km), D06  (2) transmitting antenna height (10 m to 1200 m), time DDfh variability (50%, 10%, and 1%), location variability (1% to 99%), and path type (land, cold sea, warm sea, and where Df  0.0000389 fh12 h , Dh 4.1 h12 h  , f mixed paths), at the height of the receiving antenna being is frequency, h and h are antenna heights above smooth equal to the representative height of ground cover. It has 1 2 earth (m). mentioned that if families of curves exist for regions with If the transmission distance is equal to or greater than different climates which experience substantially d , the correction should be calculated using equation (1) different prevailing radio propagation conditions, 10 accurate characterization of radio propagation in these with R' 10 m . If the transmission distance is less than regions may be attained using the methods found in this d10 , then the correction should be calculated as: Recommendation. There are some correction factors in the Recommendation, including correction for Correction 0 dB for ddh2 (3) transmitting antenna height, interpolation of field strength For d d d , as a function of distance and frequency, correction for h2 10 receiving antenna height, cluttered transmitter correction, logddh2  terrain clearance angle correction, correction due to Correction C10 dB (4) tropospheric scattering, etc. logdd10h 2  In this paper, we study on the prediction of radio over here d is distance at which the path just has 0.6 Fresnel sea propagation, especially considering interpolation and 10 extrapolation of field strength as a function of distance clearance for h2 10 m given in equation (2) calculated and frequency, correction of receiving antenna heights, as D06 f, h 1 ,10 , C10 is correction for the required value correction based on tropospheric scattering. Compared of h at distance d using equation (1) with R' 10 m , with version P.1546-4 [9], the latest version P.1546-5 2 10 which has addition correction for antenna height dh2 is distance at which the path has 0.6 Fresnel clearance difference is also considered in this paper. The two main for the required value of h2 given in equation (2) correction factors [10] of P.1546-5 shown in the calculated as D f,, h h  . The recommendation is not following are implemented to extend a sea propagation 06 1 2 model. valid for receiving antenna heights less than 3 m when adjacent to sea. In this paper, the receiving antenna we

A. Correction for Receiving Antenna Height h2 utilized is more than 3 m above the horizontal. For the sea path, the above correction for receiver For the sea path, the concept of transmitting antenna h 1 antenna height can be summarized by the flowchart is that it represents the physical height of the antenna shown in Fig. 1. above the surface of the sea. The Recommendation gives A correction is also required to take account of the correction for the situation “adjacent to sea” where the difference in height between the two antennas. receiving antenna is either over sea, or is immediately adjacent to the sea with no significant obstruction in the d direction of the transmitting station. For sea paths the Correction  20log dB (5) dslope notional value of h2 is 10 m. Where the receiving antenna is adjacent to sea for h2 10 m , the correction where d is the horizontal distance and dslope is the slope should be calculated using equation distance.

seen that as distance increase field strength decrease. Start Higher values of h1 can obtain higher field strength,

lower values of h1 curves are more sensitive to the change of distance. h2  10 m h2  10 m B. Correction Based on Tropospheric Scattering For sea paths, the influence of scattering to the D 06  f , h 1 ,10  , using eq(2) transmission is very significant. Taking account of tropospheric scattering, we calculate the path scattering

dd 10 dd 10 angle in degrees s , using

180d ' CR eq(1) with 10 m d h 2  D 06  f ,, h 1 h 2  , using eq(2)   (7) s  ka where d is the path length (km), a  6370 km , radius of dd h2 dd h2 4 the earth, k  , effective earth radius factor for median 3 C  0 eq(3) C  eq(4) refractivity conditions. Here, we ignore the effect of diffraction in the sea level. Terrain clearance angle correction which is given by the Recommendation is Finish appropriate for land path. In this paper, we consider the Fig. 1. Flowchart for receiver antenna height correction terrain clearance angel correction is zero. Calculate the field strength predicted for tropospheric

26 2 scattering, Ets , using dslope d 10  h a  h tter  h2  h rter  (6)

Ets24.4  20log d  10 s  L f  0.15 N0  G t (8) where ha is the antenna height above ground, htter and

hrter are the terrain heights in meters above sea level at where Lf 5log f  2.5 log f   3.3 is the the transmitter and receiving terminals respectively. Although the correction given by equation (5) is very frequency-dependent loss; N0  325 , median surface small except for short paths and high values of h1 , it is refractivity, N-units, typical of temperate climates; recommended that it is used in all cases to avoid making 0.7 Gtt 10.1 log 0.02  , time-dependent enhancement; an arbitrary decision as to precision. d is path length; f is required frequency. The Recommendation gives an equivalent basic transmission loss. The basic transmission loss equivalent to a given field strength is given by:

Lb 139.3  E  20log f dB (9)

where Lb is basic transmission loss, E is field strength for 1 kW e.r.p, f is frequency (MHz).

III. MEASUREMENT PROCEDURE AND ANALYSIS METHODS In order to acquire the path loss over sea propagation measurement data, a investigation team was sent by Fig. 2. P.1546-5 propagation curves for different transmitting antenna College of Information Science & Technology, Hainan heights h1 , frequency of 950 MHZ, warm sea paths, 50% of time and University, to conduct radio wave spectrum measurement locations and receiving antenna height h2 equals 5 m. and radio wave path loss measurement in the range of Hainan region in South China Sea. Fig.3 shows the Fig. 2 shows one set of curves that is applicable for sea survey boat at maximum speed of 5 m/s. paths and with a frequency equal to 950 MHz. A rigorous Radio wave propagation measurement was performed interpolation/extrapolation procedure is given to allow using a signal scanner made by Agilent Technologies Inc, consistent prediction for any input values within the machine module is N9342C. The scanner is controlled by specified range. The frequency f , transmitting antenna a laptop personal computer and includes a global positioning system (GPS) receiver and an Omni- h1 and receiving antenna height h2 are corrected. It is

directional antenna. During the measurement, the values, mi on a logarithmic [dB] scale. The prediction receiving antenna was placed on a boat at a height of error is expressed as approximately 5 m above horizontal. Table I shows the (10) main parameters of the scanner. ip i  m i , i  1,2,..., N where N is the number of sample. Hence, the mean of error is given as 1 N   i (11) N i1 The standard deviation can reflect the dispersion degree of data, it is a significant index to evaluate the accuracy of predictions. It can be calculated as

N 2 i     i1 (12) N 1 Fig. 3. Survey boat The correlation coefficient gives a measure of the

TABLE I: MAIN PARAMETERS OF THE SCANNER degree of linear relationship between two random variables and is calculated as [14] Parameters Values Parameters Values Impedance 50 Gate Delay 0.000018 N mii m p p Number of Points 461 Gate Length 0.000084 i re  (13) NN Sweep Time(s) 0.092527 Burst Level -20 22 mii m  p p Attenuation 0 Timer Period 0.001 ii Trigger Delay 0.000006 Timer Offset 0 where m and p are the means of the measured and predicted values, respectively. The measurement was started at nearby Haikou sea in Nov. 2014. Fig. 4 is the measurement path profile. The TABLE II: F IRST ORDER STATISTICS FOR PREDICTION MODELS First order transmitting pilot power, antenna height, and antenna P.1546-5 Free space Okumra-Hata statistics gain of the BSs were provided by China Mobile Hainan Mean error (dB) 5.754 8.645 17.925 branch. The measurement data originating from the two Standard deviation 2.268 4.084 4.799 BSs with Omni-directional antenna in nearby Haikou sea of error (dB) and Wenchang sea were analyzed and used for evaluating Correlation 0.907 0.897 0.894 the P.1546-5. coefficient In this paper, the first order statistics and hit rate metrics have been used to evaluate the measurement The first order statistics summary in Table II shows results. that the P.1546-5 has the smallest mean error and standard deviation of error in the three models. However, the Free Space model and Okumura-Hata model yield a slightly better correlation coefficient than the P.1546-5 model. For further analysis of the accuracy of the three models can be obtained by hit rate metrics. B. Hit Rate Metrics First order statistics are not always comprehensive to reflect the accuracy of prediction models. Assume a set of prediction path loss that closely match measurement results for most of the range, but which are subject to an

error in the predicted or measured locations, then the Fig. 4. The measurement path profile from Google Earth calculate of mean of error and standard deviation of error A. First Order Statistics will be large, although the model is accurate for the Mean of the error, standard deviation of the error and prediction of the overall number of locations. Hence, Hit the correction coefficient between the measurements and rate metrics were proposed by Owadally, Montiel, and predictions are expressed in the conventional first order Saunders [15] to complement conventional first order statistics. The mean of error is defined as the difference statistics. The measured data, mi and predicted data, pi , between the predicted values, pi and the measured are compared to the path loss threshold, LT . If the

magnitude of any measured path loss is less than or equal The correction Okumura-Hata prediction of the field to the magnitude of the path loss threshold, then define strength is given by: that the measurement result is in ‘coverage’. Otherwise, E69.82  6.16log f  13.82log H  the measurement result is in ‘outage’. The same applies   1 b (20) for the set of predicted path loss data. This can be a H21 44.9 6.55log H log d  expressed as a function as below: where H35 m, H  5 m, f  950 MHz, b  1 . Fig. 3 1 xL 12 Ux   T (14) shows the measurements and predictions path loss curves 0 others versus transmitting distance. where xp ii or m .Refer to [15], for a given path loss threshold, there are four kinds of hit rate metrics, they are Total Hit Rate (THR), Availability Hit Rate (AHR), Outage Hit Rate (OHR) and Coverage Area Accuracy (CAA). The THR gives a direct indication of the quality of a model, for it directly evaluate how often the predictions correctly predict the coverage state of any given location, and can be calculated as:

U mi U p i U m i U p i  THR L ii (15)  T   NNTT  where U is the complement of U , NT is the total Fig. 5. The measurement and predictions path loss curves versus number of points compared. transmitting distance. The prediction model parameters are f950 MHz, h  35 m, h  5 m . The AHR is the ratio of the number of locations where 12 both measurements and predictions are in a ‘coverage’ situation. The AHR is defined as

U mii U p  i AHR LT  100 (16) Up i  i The OHR is the ratio of the number of locations where both measurements and predictions are in an ‘outage’ situation relative to the number of predicted outage locations. The OHR is given by

U mii U p  i OHR LT  100 (17) Up i  i The CAA can also evaluate the accuracy of the model, Fig. 6. Models comparison using Total Hit Rate (THR) metric contrast with the THR, the specific location of coverage is less important than the overall area served. It is given From Fig. 5, it can be seen that the Okumura-Hata by model provides overall higher path loss prediction than Free Space model and P.1546-5.The P.1546-5 gives the approximate path loss values to measurement data. U mii U p  ii Within 2 km transmission, the three path loss curves are CAA LT  100    100 (18) NNTT very close. The measurement data show that the path loss values, in the transmission range of 10 km, have relatively large fluctuation. IV. EVALUATION BY MEASUREMENTS AND PREDICTIONS Fig. 6 and Fig. 7 show the THR curve and CAA curve, In this paper, the Free Space model and the correction respectively. In Fig. 6 and 7 the THR and CAA provide Okumura-Hata model [2] are used to compare with further insight into demonstrating that the P.1546-5 is correction P.1546-5. Here, the Free Space field strength more accurate than the Free Space model and Okumura- for 1 kW e.r.p is given by : Hata model. A high THR means that there is a good match between the predictions and the measurements. Fig Edfs 106.9 20log  (19) 6 shows that in a band of the path loss threshold, between

120 dB to 180 dB, the P.1546-5 outperforms the other The AHR and OHR metrics correspond with standard two models. While in a band of 90 dB to 120 dB, the deviation of the error. If the standard deviation of the three models have similar prediction accuracy. The CAA error and the mean error are low, the prediction and gives a concept of the number of locations at which the measurement curves are expected to be very close to each predictions lie on the same side of the path loss threshold other and to have a similar shape. Fig. 8 and Fig. 9 show as the measurements. As shown in Fig. 7, in the band of the AHR and OHR respectively, we can see that the 110 dB to 180 dB, the P.1546-5 does better than the other P.1546.5 have a higher OHR but lower AHR. From the two models. analysis data, there are a set of locations where predictions differ significantly from the measurements. Hence this causes the predictions and measurements to be different sides of the path loss threshold. This is why results in Fig. 8 and Fig. 9.

V. CONCLUSION In this paper, we extend a correction ITU-R recommendation P.1546-5 for radio over sea propagation. The correction P.1546-5 is compared with two traditional models and evaluated using measurements that were obtained by utilizing a commercial GSM mobile network. The measurements were carried out in Nov.2014 from nearby Haikou sea to Wenchang sea, China. Four hit rate

Fig. 7. Models comparison using Coverage Area Accuracy (CAA) metrics were used to evaluate the accuracy of the three metric. models. It has been shown that P.1546-5 provides better overall prediction of the path loss compared to the Free Space model and the Okumura-Hata model. Hit rate metrics give a complementary ways to analyze the accuracy of the models. Although a method to predict the field strength for sea paths has been provided in Recommendation ITU-R P.1546 since 2007, additional works are still needed to complete the model that can be wildly applied for various ocean scenarios. In this paper, the effect of diffraction is ignored, and there is no terrain clearance angel correction for sea paths in the Recommendation ITU-R P.1546. To develop reliable correction models, the effect of diffraction from obstacles on the sea should be considered, a lot of measurements from different types of terrain should be obtained and analyzed in future work. Fig. 8. Models comparison using Availability Hit Rate (AHR) metric

ACKNOWLEDGMENT This work is supported by the National Natural Science Foundation of China (Grant No.61162010). The authors would like to thank the editor and the anonymous reviewers for their valuable comments.

REFERENCES [1] IEEE 802.22 Working Group on Wireless Regional Area Networks. [Online]. Available: http://ieee802.org/22/ [2] K. B. Kim, M. Ali, J. H. Lee, et al., “Experimental study of propagation characteristic for maritime wireless communication,” in Proc. International Conf. Antennas and Propagation, Nagoya, 2012, pp. 1481-1484. [3] F. Clazzer, A. Munari, M. Berioli, et al., “On the characterization of AIS traffic at the satellite,” in Proc. OCEANS Conf. Oceanic Fig. 9. Models comparison using Outage Hit Rate (OHR) metric Engineering Society and Marine Technology Society, TaiPei, 2014, pp. 1-9.

[4] Longley-Rice Model. [Online]. Available: http://www.awe- Han Wang was born in Jiangxi Province, communications.com/Propagation/Rural/ITM/index.htm China, in 1986. He received the B.S. degree in [5] M. Farhoud and A. El-keyi, “Empirical correction of the okumura- electrical engineering from Hubei University hata model for the 900MHz band in Egypt,” in Proc. 3th ICCIT of Nationalities, China, in 2009 and the M.S. Conf. Communications and Information Technology, Belrut, 2013, degree in information and communication pp. 386-390. system from Hainan University, Haikou, [6] J. M. Collis, L. William et al., “Extension of the rotated elastic China, in 2013. He has worked in China parabolic equation to beach and island propagation,” IEEE Mobile Jiangxi branch as a network engineer Journal of Oceanic Engineering, vol. 34, no. 4, pp. 671-683, Nov. for one year. Now, he is pursuing the Ph.D. 2009. degree with the Department of College of Information Science & [7] F. J. Ryan, J. T. Johnson, and R. J. Burkholder, “A comparison of Technology in Hainan University. His research interests include propagation over sea surfaces using MOM and PWE methods,” maritime communication and information theory. presented at Radio Science Meeting, 2014 United States National Committee of URSI National Boulder, CO, Jan. 1, 2014. Wencai Du received the Bachelor of Science [8] J. Y. Chen, “Method of the recommendation ITU-R P.1546-3 for degree from the Peking University, Beijing, VHF field strength prediction over sea propagation,” Marine China, in 1978. He received the two M.S. Technology, vol. 1, no. 3, pp. 39-42, March 2009. degree from the Hohai University, Nanjing, [9] Method for Point to Area Predictions for Terrestrial Services in China, in 1986, and from ITC, Enschede, The the Frequency Range 30 MHz to 3000 MHz, ITU-R Netherlands, in1996. He received the Ph.D. Recommendation P.1546-4, Oct. 2009. degree at University of South Australian, in [10] Method for Point to Area Predictions for Terrestrial Services in 2000. He conducted ostdoctoral research at the Frequency Range 30 MHz to 3000 MHz, ITU-R Technion-Israel Institute of Technology (IIT), Recommendation P.1546-5, Sep. 2013 Israel, from March, 2001-March 2002. His research interests span the [11] S. Kasampalis, P. I. Lazaridis, Z. D. Bizopoulos, et al., areas of computer science and communication engineering. He is “Comparison of longley-rice, ITU-R P.1546 and hata-davidson especially interested in the computer networking, service computing, e- propagation models for DVB-T coverage prediction,” in Proc. service and maritime communication. Dr. Du has authored or co- IEEE BMSB Conf. Broadband Multimedia Systems and authored 18 books and more than 80 scientific publications. Dr. Du has Broadcasting, Beijing, 2014, pp.1-4. served on the technical and executive committees of several major [12] S. H. Bae and D. H. Cha, “A new approach for 1 km urban conferences and workshops. He was the Conference Chair to propagation model of the recommendation IUT-R P.1546,” IEEE/ACIS ICIS 2011, Conference Co-Chair to SNPD 2010, London, presented at Radio Science Meeting, 2014 USNC-URSI, Memphis, Conference Chair to IEEE/ACISto SERA 2009, and Program Chair to Tennessee, July, 235, 2014. SNPD 2009, Daegu, Korea. [13] S. E. Shumate, “Longley-rice and ITU-P.1546 combined: A new international terrain-specific propagation model,” in Proc. IEEE 72nd VETECF Conf. Vehicular Technology Conference Fall, Xing Chen was born in Hainan Province, Ottawa, 2010, pp. 1-5. China, in 1991. She received her B.S. degree [14] E. Ostlin. H. J. Zepernick, and H. Suzuki, “Evaluation of the in communication engineering from Nanjing propagation recommendation ITU-R P.1546 for mobile services in University of Science and Technology, China, rural Australia,” IEEE Trans. on Vehicular Technology, vol. 57, in 2013. Now, She is a master student, major no. 1,pp. 38-50, Jan. 2008. in information and communication [15] A. S. Owadally, E. Montiel, and S. R. Saunders, “A comparison of engineering in Hainan University. Her the accuracy of propagation models using hit rate analysis,” in research interest is the marine white spectrum Proc. IEEE Vehicular Technology. Conf., Atlantic City, NJ, 2001, occupancy analysis. vol. 4, pp. 1979-1983.

Dong You Choi

#36 Naedok-dong, Sangdang-gu, Cheongju-city 360-764 Korea Division of Information Engineering and Telecommunication, Cheongju University [email protected]

Abstract. In order to estimate the signal parameters accurately for wireless multimedia services, it is necessary to estimate a system’s propagation characteristics through a medium. Propagation analysis provides a good initial estimate of the signal characteristics. The ability to accurately predict radio propagation behavior for wireless multimedia services is becoming crucial to system design. Since site measurements are costly, propagation models have been developed as a suitable, low cost, and convenient alternative [1]. A number of studies have been conducted to quantitatively predict the characteristics of propagation in inhabited areas on land having many wireless multimedia service users, resulting in a number propagation prediction models being proposed. However, since very few such studies have been conducted for the sea, which has a different physical layer structure from land, the propagation prediction model for free space has been commonly used. Thus, in this study, I measured the propagation path loss of a 1950 MHz band signal over the sea surface, and analyzed the results by comparing them with the path loss data of a propagation prediction model in free space, which is frequently used to predict the propagation path loss over the sea surface.

1 Introduction

The commercial success of wireless communication, since its initial implementation in the early 1980s, has led to there being an intense interest among wireless engineers in understanding and predicting the radio propagation characteristics in various urban and suburban areas, and even within buildings. Given that the explosive growth of wireless multimedia service is continuing unabated, it would be very useful to have the capability of determining the optimum base-station location, obtaining suitable data rates, and estimating their coverage, without having to conduct extensive propagation measurements, which are very expensive and time consuming [1]. Whereas many studies have been conducted to predict the characteristics of propagation quantitatively land, including the development of many propagation prediction models, few such efforts have been conducted for the sea. In fact, there are many difficulties involved in providing wireless multimedia services over the sea, viz. the lack of economic viability associated with long and short distance services, the absence of good locations for new base-stations, and the difficulties associated with these locations. To solve these problems, facility investment and maintenance ex- penses need to be reduced by optimizing the service area per base-station the precise prediction of the propagation path loss over the sea surface. Accordingly, in this study, I measured the propagation path loss of a 1950 MHz band signal over the sea surface, and analyzed the results by comparing them with the predicted propagation path loss in free space, which is frequently used to predict the propagation path loss over the sea surface.

2 Propagation environment and propagation path loss

The radio propagation over the sea surface is different from the land propagation prediction model. In other words, the total received power of a mobile unit situated over the sea is the sum of the direct wave, the reflected wave from the sea surface, and the reflected wave from the ground. As a result, it gives more intense interference to other base-stations and mobile units, as compared with land propagation, and so special attention and care is needed. The received power over the sea surface is given by Eq. (1) [2].

2 ⎛ λ ⎞ jd jd 2 P = P × ⎜ ⎟ 1 − e θ 1 − e θ 2 r t 4πd ⎝ ⎠ (1) 2 ⎛ λ ⎞ 2 = Pt × ⎜ ⎟ 1 − (cos d θ 1 + cos d θ 2 ) − j (sin d θ 1 + sin d θ 2 ) ⎝ 4π d ⎠

where d is the path length, λ is the wavelength, dθ1 is the difference in the propagation path between the direct wave and the reflected wave from the ground, dθ 2 is the difference in the propagation path between the direct wave and the reflected wave from the sea surface, pt is the transmitting power, and pr is the received power in free space. However, in Eq. (1), the difference in the propagation path between the two reflected waves, dθ1 and dθ 2 , is sufficiently small for propagation path loss over the sea surface to be replaced by the predicted value of the propagation path loss in free space, if a limiting value, 0, is adopted in both dθ1 and dθ 2 , as shown in Eq. (2).

2 ⎡ ⎧ ⎫ 2 ⎤ ⎪ ⎛ λ ⎞ ⎪ jd θ 1 jd θ 2 Pr ≈ lim ⎢ ⎨ Pt × ⎜ ⎟ ⎬ 1 − e − e ⎥ d θ 1 ,d θ 2 → 0 ⎢ ⎪ ⎝ 4π d ⎠ ⎪ ⎥ ⎣ ⎩ ⎭ ⎦ (2) 2 ⎛ λ ⎞ ≈ Pt × ⎜ ⎟ ⎝ 4π d ⎠

In general, the propagation path loss in a downtown environment is known to be about 20 ~ 50dB/dec, based on empirical measurements, depending on the environment, and an approximate value of 20dB/dec is generally use for the propagation path loss over the sea surface [1, 3, 4].

3 Experimental setup and measurement procedure

The propagation path loss over the sea surface was measured in the vicinity of the islands situated off the coast of in Latin America.

(b) (c)

(a)

(d) (e)

Fig. 1. Sample areal photograph. (a) Test antenna (b) East (c) West (d) South (e) North

Table 1. Location information

Latitude Longitude

Test site 17-55-9.7 (-)87-57-40.6

UTM coordinate 1981526.786 398186.5393 Table 2. Measured parameters

Tx Cable Tx ant. Tx ant. Rx ant. power loss gain height height 22m 20W 2dB 6dB (Ant. 2m + 2.5m (43dBm) Building 20m）

The transmitting signal (1950 MHz) was generated using a signal generator installed in the steel tower of an existing base-station in Latin America. The signal sent from the base-station was processed on a real-time basis using HP RF Coverage Measurement equipment manufactured by Agilent. The location information was obtained using a GPS (Global Positioning System) embedded in the receiving set and Mercator projection [4, 5]. To measure the propagation path loss over the sea surface, a boat with a speed of 40 ~ 60km/h was used.

4 Results

Fig. 2 shows the path loss slope from the measured data. The measured data in Fig. 2 represent the mean values of tens ~ hundreds of data collected when the mobile unit’s location was changed by 0.001 degrees in latitude or longitude.

Measured data -80 Linear fit of measured data

-90

-100

-110

-120

-130 Signal level [dB]

-140

-150

-160 0 2000 4000 6000 8000 10000 12000 14000 16000 Path distance [m]

Fig. 2. Path loss slope from measured data

Fig. 3 shows the results calculated using Eq. (2) for the predicted data of the propagation path loss in free space, which is frequently used to predict the propagation path loss over the sea. The propagation path loss and regression analysis according to distance are also presented using the measured data.

-80 Measured data Regression curve -90 Prediction model

-100

-110

-120

-130 Signal level [dB] level Signal

-140

-150

-160 0 2000 4000 6000 8000 10000 12000 14000 16000 Path distance [m]

Fig. 3. Propagation path loss according to distance

Fig. 4 shows the difference of the regression of measured data against predicted data of propagation path loss in free space.

-5

-10

-15

Path loss difference [dB] -20

-25

0 2000 4000 6000 8000 10000 12000 14000 16000 Path length [m]

Fig. 4. Path loss difference according to distance

The results in Fig. 2, 3 and 4 suggest that the measured data of the propagation path loss over the sea surface were smaller than the predicted data of the propagation path loss in free space up to 2,200m, but bigger at a distance above 2,200m. As the path length was increased, the measured data were greatly increased compared to the path loss of predicted data. The smaller of the measured propagation path loss up to 2,200m, as compared to the predicted ones, may result from the absence of obstacles in the area, leading to a strong radio strength and large radio filed strength of the reflected radio signal. More specifically, the propagation path loss over the sea surface was about 40dB/dec, which is 20dB/dec bigger than the propagation path loss in free space of 20dB/dec. Table 3 and 4 show the difference and the standard deviations of the predicted data of the propagation path loss in free space.

Table 3. Difference between predicted and measured data according to distance

Path length [m] Difference range [dB] Min. difference [dB] Max. difference [dB]

0 ~ 2000 -10.7 ~ +11.6 0.7 11.6 2001 ~ 4000 -10.6 ~ +2.3 0.1 10.6 4001 ~ 6000 -15.6 ~ +0.8 0.2 15.6 6001 ~ 8000 -21.5 ~ -4.4 4.4 21.5 8001 ~ 10000 -25.1 ~ -13.9 13.9 25.1 10001 ~ 12000 -30.6 ~ -13.6 13.6 30.6 12001 ~ 14000 -24.9 ~ -16.4 16.4 24.9 14001 ~ 14700 -29.1 ~ -21.2 21.2 29.1

Table 4. Standard deviation of predicted and measured data according to distance

Measured data Regression data Path length [m] & predicted data & predicted data 0 ~ 2000 6.3 3.0 2001 ~ 4000 3.9 1.8 4001 ~ 6000 4.5 1.9 6001 ~ 8000 4.8 1.5 8001 ~ 10000 3.9 1.2 10001 ~ 12000 4.6 0.9 12001 ~ 14000 2.0 0.6 14001 ~ 14700 3.0 0.2

Total 10.3 9.3 From the experimental data, we know that one standard deviation of data spread on any radio path length is about 8dB. This spread is due to the various terrain conditions from which the data are collected at the same radio path length [7, 8, 9]. How- ever, the standard deviation of the measured data for the propagation path loss over the sea surface is 10.3dB and the regression is 9.3dB, as compared to the predicted data for the propagation path loss in free space.

5 Conclusions

It is common practice to use the predicted model of the propagation path loss in free space to predict the propagation path loss over the sea. Thus, in this study, we measured the propagation path loss of a 1950 MHz band signal over the sea, and compared the results to the predicted data of the propagation path loss in free space. The principal results of this comparison are as follows.

- The propagation path loss over the sea surface was about 40dB/dec, which was 20dB/dec bigger than the propagation path loss in free space (20dB/dec). - The Standard deviation of the predicted and measured data for the propagation path loss over the sea surface is 10.3dB, which is bigger than the standard deviation of the propagation loss land (8dB). - As path length was increased, the differences were greatly increased.

The measured results reported in this paper are very valuable in that they provided a means of determining the optimum base-station locations, suitable data rates and estimating their coverage, without having to conduct extensive propagation measurements, which are very expensive and time consuming. Further studies are needed to develop the propagation prediction model for above the sea, by measuring the propagation path loss over the sea surface for various frequency bands.

Acknowledgment: I would like thank Dea-sick Choi, Jin-man Kim, Kyung- Jae Kim (RF Engineering Dept. R&D Center LGE) for useful discussions and valuable data.

References

1. Tapan K. Sarkar, Zhong Ji, Kyungjung Kim, Abdellatif Medouri and Magdalena Salazar- Palma, “A Survey of Various Propagation Models for Mobile Communication,” IEEE An- tennas and Propagation Magazine, Vol. 45, No. 3 (June 2003) 51-82 2. Ki-sun Kim et al., Mobile Cellular Telecommunication (Analog and Digital Systems-2nd), Sigmapress (1996) 158-164 3. Dea-sick Choi, Jin-man Kim, Kyung-Jae Kim, “An Analysis of Radio Propagation Path Loss in the Sea,” KEES proceeding, Vol. 10, No 1 (November, 2000) 255-258 4. Dr. Kamilo Feher, Wireless Digital Communications: Modulation & Spread Spectrum Ap- plications, Prentice Hall Inc., (1995) 66-69 5. Seung-min Wee, Si-hwa Kim and Il-dong Chang, “On the Implementation of Route Plan- ning Algorithms on the Electronic Chart system,” Journal of Korea Institute of Navigation, Vol. 24, No. 3 (2000) 167-176 6. Weon-jae Yang, Seung-hwan Jun and Gei-kak Park, “Development of GPS simulation Tool Kit for personal computer,” Journal of Korea Institute of Navigation, Vol. 24, No. 4 (2000) 219-226 7. William C. Y. Lee, Mobile Communications Design Fundamentals, John Wiley & Sons Inc.(1993) 51-53 8. William C. Y. Lee, Mobile Communications Engineering, McGraw Hill Book Co. (1982) 107 9. K. K. Kelly Ⅱ, “Flat Suburban Area Propagation of 820 MHz,” IEEE Transactions on Vehicular Technology, Vol. Vt-27 (November 1978) 198-204 824 Y. H. LEE, F. DONG, Y. S. MENG, NEAR SEA-SURFACE MOBILE RADIOWAVE PROPAGATION AT 5 GHZ . . .

Near Sea-Surface Mobile Radiowave Propagation at 5 GHz: Measurements and Modeling

Yee Hui LEE1, Feng DONG1, Yu Song MENG2

1School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore 2 National Metrology Centre, Agency for Science, Technology and Research (A*STAR) 1 Science Park Drive, Singapore 118221, Singapore

[email protected], [email protected], meng [email protected]

Abstract. Near sea-surface line-of-sight (LoS) radiowave be operated in a near sea-surface line-of-sight (LoS) mobile propagation at 5 GHz was investigated through narrowband environment (e.g., between a BS and a moving vessel). The measurements in this paper. Results of the received sig- radio channel for this scenario differs from those reported nal strength with a transmission distance of up to 10 km in [1–4], [8], [9]. Moreover, our previous investigations [9], were examined against free space loss model and 2-ray path [12] indicate that ducting is significant for radiowave propa- loss model. The experimental results have good agreement gation over the tropical ocean of Singapore in 5 GHz band. with the predicted values using the 2-ray model. However, Therefore, although similar scenarios [5–7], [13] were re- the prediction ability of 2-ray model becomes poor when ported, their maritime conditions and frequencies are differ- the propagation distance increases. Our results and anal- ent and may not be applicable to the 5 GHz radio link over ysis show that an evaporation duct layer exists and there- a tropical ocean where the occurrence of an evaporation duct fore, a 3-ray path loss model, taking into consideration both is known to be of a higher probability and more predomi- the reflection from sea surface and the refraction caused by nant [5]. evaporation duct could predict well the trend of LoS signal From the literature, it is well-recognized that radiowave strength variations at relatively large propagation distances propagation over a sea surface is affected by the ducts such in a tropical maritime environment. as surface ducts, elevated ducts, and evaporation ducts [14], [15]. For near sea-surface LoS propagation, the effects of evaporation duct are obvious and dominant amongst all the Keywords ducts. This is because the evaporation duct exists over the ocean almost all of the time [14], since it is a result of the Evaporation duct, maritime mobile, modeling, path loss rapid decrease in vapor pressure from a saturation condition prediction, sea reflection. at the sea surface to an ambient vapor pressure at levels several tens of meters above the sea surface. This decrease in vapor pressure generally results in a decrease in the modified 1. Introduction refractivity and thus, creates a ducting layer adjacent to the sea surface. Research work in [2] indicated that an evapo- Radiowave propagation in maritime environments has ration duct above the sea surface can result in a substantial been of great interest to many researchers over the years increase in the received signal strength at frequencies above [1–9]. The understanding of over-sea radiowave propaga- 3 GHz. Measurement results in [3] also showed an enhance- tion is very important for system designers when planning to ment in signal strength of more than 10 dB, observed 48% establish a reliable radio link between a sea vessel and an on- of the time along a 27.7 km over-sea path at 5.6 GHz. shore base station (BS). Over the years, different application Therefore, it is important to study the effect of evapora- scenarios have been investigated; in [2–4], over-the-horizon tion duct in characterizing and modeling of near sea-surface fixed links were examined; in [6], [7], mobile channels were radiowave propagation at the frequency of 5 GHz. As a con- studied; and in [8], [9], slant-path propagation for aeronauti- tinuation of our previous work [16] where channel character- cal applications was analyzed. istics of a non-line-of-sight (NLoS) sea-surface link at 5 GHz Recently, there has been growing interest in implemen- for Singapore maritime environment was reported, the main tations of wireless systems in 5 GHz band for maritime en- objective of this paper is to perform a detailed investigation vironments, mainly in the applications of WiMAX at sea- on near sea-surface LoS propagation through path loss mod- ports [10] and for military off-shore anti-terrorist surveil- eling with the considerations the evaporation duct and the lance [11]. These applications require the 5 GHz systems to sea-surface reflection. RADIOENGINEERING, VOL. 23, NO. 3, SEPTEMBER 2014 825

In the following, measurement campaign is described antenna installed on the roof of a building as shown in Fig. 1b in Section 2. In Section 3, free space loss model and 2-ray with different heights above sea level (in order to study the path loss model are examined against the experimental re- effect of antenna height), and connected to a spectrum ana- sults ﬁrst. Based on the observed discrepancy, a 3-ray path lyzer. The span of the spectrum analyzer was set to 20 kHz loss model is introduced to take into consideration the re- around its center frequency to reduce the noise bandwidth. fracted wave by evaporation duct. Section 4 then gives a dis- Peak marker readings of the received signal were recorded at cussion on empirical estimations of the effective duct height. intervals of 1 second using a Labview program. All the data Finally, conclusions of this paper are given in Section 5. recorded was time-stamped for synchronization with the location, heading and orientation of the mobile transmitter. The whole system was carefully calibrated on-site before the sea trials, and checked again after the measurements. The system effect was minimized through the removal of the antenna gains and the measurement of a back-to-back connection between the transmitter and the receiver. Weather conditions were also recorded using a weather meter at the transmitter side and a weather station at the receiver side respectively. However due to the lack of more suitable meteorological instruments such as a weather balloon, the weather information was restricted at the transmitter and receiver al- (a) Transmitter on a speed boat titude levels.

(b) Receiver on shore Fig. 1. Setup of transmission and data-logging systems for near sea-surface LoS measurements. Fig. 2. Measurement route and receiver location, taken from Google Earth. 2. Measurement Campaign Parameter Value Carrier frequency 5.15 GHz Transmitted power 30 dBm 2.1 Measurement System Transmitter height 3 to 4 m Receiver height 20 m, 10 m and 7.6 m Fig. 1 shows the main equipment and setup used Maximum route distance 10 km for narrowband measurements. It is noted that 5.15 GHz Antennas type Omni-directional was chosen and approved for continuous-wave transmission where no interfering signal was detected. As shown in Tab. 1. Summary of measurement parameters. Fig. 1a, an omni-directional antenna was mounted on top of a speed boat at a height of approximately 3.5 m above 2.2 Measurement Routes sea level. It was connected to a signal generator and a high- Measurements were carried out over an open sea area power ampliﬁer housed within the boat cabin, forming a mo- off the southeast coast of Singapore. BS was located at bile transmitter with an output power of 30 dBm. During the a yacht club (N 01◦1900900, E 104◦102200). The mobile trans- measurements, GPS data was continuously logged so as to mitter was on board a speed boat with a maximum speed of obtain the instantaneous time, altitude, longitude and lati- 30 knots. During the trials, the boat traveled along a 10 km tude coordinates. The pitch and roll of the moving boat was route as shown in Fig. 2 with a speed of approximately logged using a ﬂuxgate compass. 6 knots. LoS was maintained throughout the measurements The BS was on shore, close to the sea. In order to except for some occasional passing-by ships for a short pe- ensure a LoS condition between the transmitter and the re- riod of time. For each receiver height, multiple trials on dif- ceiver, the receiver consisted of an identical omni-directional ferent days have been carried out in order to compare and 826 Y. H. LEE, F. DONG, Y. S. MENG, NEAR SEA-SURFACE MOBILE RADIOWAVE PROPAGATION AT 5 GHZ . . .

verify the results. It is noted that all the measurement cam- FSL model and the 2-ray model will be used to predict the paign was conducted under similar sea status; calm or near received signal strength under ideal conditions for near sea- calm. The main measurement parameters are summarized in surface radiowave propagation at 5.15 GHz. Tab. 1. −40 Raw 2−ray 3. Near Sea-Surface Path Loss Model −60 FSL

ITU-R P.1546 model [13] provides a set of curves for −80 prediction of propagation loss over a sea path at a frequency from 30 MHz to 3 GHz, which limits its application in our −100 study at 5.15 GHz. However, a general trend can be observed from those ﬁeld strength versus distance curves in [13] that −120 radiowave propagation over sea paths (path distance less Received signal strength(dBm) than 10 km) approaches free space propagation at 50% of 0 2000 4000 6000 8000 time when the frequency increases. Since the frequency of Distance(m) (a) Results at h = 20 m conducted on 23 November 2011 operation in this study is 5.15 GHz, free space loss (FSL) r model could therefore be used. Moreover, works in [5] also −40 Raw reveals that propagation loss of a 5-GHz over sea-water radio 2−ray link can be predicted using the FSL model when the trans- −60 FSL mission range is less than 10 nautical miles (18.52 km) but with some interference nulls. −80 These nulls could be due to the interference between −100 the direct ray and the sea-surface reﬂected ray. Our previous study [9] also found that around 86% of all the trials −120

for over-sea radiowave propagation can be represented by Received signal strength(dBm) a 2-ray multipath model (i.e., a direct ray with a sea-surface 0 2000 4000 6000 8000 reﬂected ray) over the tropical ocean. Although the infor- Distance(m) mation in [9] is for aeronautical applications, there is a very (b) Results at hr = 10 m conducted on 28 September 2011 high probability for the existence of a sea-surface reﬂected −40 wave for near sea-surface LoS propagation. Therefore in this Raw 2−ray section, we will examine the received signal strength against −60 FSL the predicted results using the FSL model and the 2-ray path loss model. −80

3.1 Propagation Loss Models −100

For the radiowave propagation in free space, the prop- −120 agation loss can be predicted by the FSL model [17], Received signal strength(dBm) 0 2000 4000 6000 8000 Distance(m) LFSL = −27.56 + 20log10( f ) + 20log10(d) (1) (c) Results at hr = 7.6 m conducted on 3 April 2012 where LFSL is the free space loss in dB, f is the frequency in Fig. 3. Received signal strengths versus distance at different re- MHz, and d is the propagation distance in meters. ceiver heights; measured and predicted. When a reflected ray from the sea surface exists besides the LoS path (direct ray), the propagation loss could be 3.2 Comparison with the Predicted Results predicted by a 2-ray path loss model. Since there is a near- Typical measurement results at the receiver height hr grazing incidence on the sea surface in our measurements, of 7.6 m, 10 m, and 20 m are shown in Fig. 3, together the reflection coefficient for a vertically polarized wave ap- with the predicted signal strengths using the FSL model and proaches -1. Therefore, the 2-ray path loss model can be the 2-ray model (assuming a perfect sea-surface reflection). simplified as [17] From Fig. 3, it can be observed that the FSL model is able to ( 2 2) predict the exponential decreasing strength trend for all the λ 2πht hr L2−ray = −10log 2sin (2) 3 receiver heights. The FSL model is well-suited for predic- 10 4πd λd tion of the local mean (large scale) propagation loss. These where L2−ray is the 2-ray propagation loss in dB, λ is the observations are consistent with those reported in [5], [13]. wavelength in meters, and ht , hr are the heights of a trans- More interestingly, when the propagation distance is less mitter and a receiver in meters. In the following, both the than about 2000 m to 3000 m (from Fig. 3), the measured RADIOENGINEERING, VOL. 23, NO. 3, SEPTEMBER 2014 827

results show a similar trend as the predicted results using the 4ht hr 2-ray model for all the measurement scenarios. That is, there dbreak = . (3) are some interference nulls which are due to the destructive λ summation of the radiowaves (the direct wave and the sea- In our measurement campaign, dbreak is around 4000 m, surface reflected wave from modeling process of the 2-ray 2000 m and 1500 m when hr is 20 m, 10 m and 7.6 m re- model) arriving at the receiver. The slight misalignments spectively. For path loss modeling beyond dbreak, a multi-ray of the predicted nulls with respect to the measured ones in path loss model is considered. In the following, a 3-ray path Fig. 3 are due to the sea-surface roughness [18], [19] and the loss model, taking into consideration the refraction due to refraction of the propagating waves mainly. Both of them the evaporation duct as shown in Fig. 4 is proposed. can weaken the applicability of 2-ray path loss model which is assumed for a perfect reflection and straight rays. 3.3 Modeling with the Ducting Effect However, it is observed from Fig. 3 that as the propa- As discussed above, 2-ray path loss model will lose gation distance increases beyond about 2000 m to 3000 m its prediction ability when the propagation distance exceeds (known hereby as the break point dbreak), the prediction abil- dbreak roughly. For example, interference nulls observed in ities of both the FSL model and the 2-ray model become Fig. 3a at around 900 m, 1100 m and 2100 m (< d , poor for near sea-surface LoS environments. There are some break dbreak ≈ 4000 m) can be predicted using the 2-ray model. interference nulls beyond dbreak that cannot be predicted us- While the nulls at 4100 m, 5300 m and 6200 m (> dbreak) ing both the models. As seen in Fig. 3, both the models tend cannot be predicted by the 2-ray model. The interference to approach a stable signal level with an exponential decay. nulls could be caused by the trapped wave within the evaporation duct as illustrated in Fig. 4. Although a distance of about 5000 m is reported as the one after which the ducting should be accounted for in [21], there is a higher probability for a shorter ducting distance (e.g., < 4500 m) in the tropical ocean where our measurements were carried out. This is because the evaporation duct heights in tropical waters are typically larger than those re- (a) Main propagation mechanisms in an evaporation duct ported for temperate cooler waters [22]. It therefore could start to trap the radiowaves at a shorter distance. The trapped wave in the evaporation duct is more likely due to the radiowave refraction as shown in Fig. 4a. This is because the upper boundary of the evaporation duct layer is not homoge- nous [20], the radiowaves incident onto the upper boundary would be diffused immediately and make the reflection from (b) Approximate representation of the refracted wave the boundary insignificantly in the received field. Moreover Fig. 4. Near sea-surface radiowave propagation where a re- for the refraction, the refracted ray (3rd ray) will not appear at fracted wave is approximately represented as a reflected short distances, but would appear as the distance increases. wave. Especially when the distance increases much further, more additional rays could also appear although there is minor These additional interference nulls beyond d could break probability for them to happen in our measurements. be due to an additional ray which appears as the propagation distance increases. It can be a refracted ray (as shown Therefore in this study, a 3-ray path loss model (in- in Fig. 4) caused by the evaporation duct which exists over cluding a direct LoS ray, a reflected ray from sea surface, the tropical sea surface almost all of the time [14]. From the and also a refracted ray by evaporation duct) is used for reported works, the evaporation duct can trap the over-water modeling and predicting near sea-surface LoS propagation propagating signals [20]. Therefore, a multi-ray path loss preliminarily. Although there are other methods which may model which takes into consideration the signals trapped in be better for modeling the radiowave propagation in a duct an evaporation duct should be considered beyond dbreak. (e.g., parabolic equation approximation method [19]), ray- tracing is preferred in this study due to the simplicity of its In order to estimate this break point d , we reviewed break final mathematical expression that describes the scenario and the trends of the FSL model and the 2-ray model. As the hence, its straightforward application in radio planning. As distance increases beyond the last null predicted by 2-ray shown in Fig. 4b, the refracted ray is approximately repre- model, both the models tend to level out slowly. Therefore, sented by a near-grazing reflected wave to simplify the pro- it could be concluded that the 2-ray path loss model will process of ray-tracing modeling since the antenna heights are vide accurate predictions roughly up to a distance of its last much smaller than the propagation distance. predictable null (or more precisely the first maximum after the last predictable null). This break distance dbreak can be For 3-ray path loss modeling, the evaporation duct roughly estimated as layer is assumed to be horizontally homogeneous. Similar to 828 Y. H. LEE, F. DONG, Y. S. MENG, NEAR SEA-SURFACE MOBILE RADIOWAVE PROPAGATION AT 5 GHZ . . .

the 2-ray path loss model described earlier, a near-grazing in- regular vertical intervals above the sea surface up to a height cidence on the sea surface is assumed and finally, the reflec- of 40 m or more. However, due to the limitation of meteo- tion coefficient for a vertically polarized wave approaches to rological instruments (e.g., lack of a weather balloon as we -1. With these assumptions, the 3-ray path loss model [23] mentioned above), it is difficult to get the vertical weather can be simplified into (4), information on site which restricts an accurate estimation of ( 2 ) h. Therefore, the single point weather information collected λ 2 L3−ray = −10log10 [2(1 + ∆)] , (4) at both the transmitter and receiver has to be used to estimate 4πd the duct height h for all the 21 measurement campaign, using a modified P-J formulation given in [24]. The P-J formula- with tion was developed for an open ocean and therefore it may 2πh h 2π(h − h )(h − h ) lose the prediction accuracy when applied to this coastal en- ∆ = 2sin t r sin e t e r (5) λd λd vironment. It is also noted that the P-J formulation is very sensitive to the weather information. where ht and hr are the heights of the transmitter and the receiver in meters, and h is the effective duct height as shown e in Fig. 4b. he is approximately equal to (or slightly less −40 Raw than) the height of evaporation duct layer which is used as 2−ray FSL a reference in 3-ray path loss modeling. A preliminary in- −60 3−ray vestigation of 3-ray path loss modeling is then performed through evaluating the measured signal strengths against the −80 estimated ones using (4) with different he assumed. Fig. 5 −100 shows an example of results with he = 25 m and 35 m respectively. −120

Received signal strength(dBm) 0 2000 4000 6000 8000 Raw Distance(m) −40 h =35 e Fig. 6. The received signal strength versus distance with the pre- h =25 dicted results using the FSL, 2-ray and 3-ray models: an −60 e example with hr = 10 m.

−80 An empirical method of determining the effective duct height he is therefore performed through the curve ﬁt- −100 ting/regression technique on the measured data in this study. The method is similar to the one reported in [25] where he −120

Received signal strength(dBm) was obtained based on a split-step Fourier solution of the

0 1000 2000 3000 4000 5000 6000 7000 parabolic equation approximation to the wave equation. Em- Distance(m) pirical values of he in (5) are then estimated by performing Fig. 5. Example of the received signal strength versus distance curve fitting onto our measurement data to align the pre- with the estimated values using 3-ray model with he = dicted nulls with the measured ones at larger distances for 25 m and 35 m. each trial. An example of the curve fitting of 3-ray path From Fig. 5, it can be observed that the signal nulls be- loss model to the experimental results for determining he is yond dbreak of 1500 m in the measured results which cannot shown in Fig. 6. For completeness, both the FSL model and be predicted by the 2-ray model previously could be esti- the 2-ray model are also shown in Fig. 6. mated using the 3-ray model. The misalignments of the mea- From Fig. 6, it can be observed that the fitted 3-ray sured nulls with the predicted ones are due to the improper model shows a good prediction ability when the propaga- values assigned to he. Therefore for modeling of radiowave tion distance is beyond dbreak, especially for the sudden drop propagation over a tropical sea surface, 3-ray path loss model of the received signal level at distances between 4 km and that not only considers the direct ray and the sea-surface re- 5 km. The sudden signal drop is due to the destructive sum- flected ray, but also the refracted ray by an evaporation duct, mation of additional refracted wave by evaporation duct with should be considered. the LoS ray and the sea-surface reflected wave. The effective evaporation duct height he obtained from the curve fittings of (4) to the measurement data is able to predict the signal vari- 4. Estimation of Effective Duct Height ations very well. The observations hold for all the 21 measurement campaign performed over the tropical sea environ- In order to know the effective duct height h as shown e ment, and correspondingly h are estimated and summarized in Fig. 4b, the height h of a typical evaporation duct which is e in Fig. 7 with the calculated h using the P-J formulation [24]. approximately equal to (or slightly higher than) he is used as a reference. For deriving h, weather information such as hu- The observations from Fig. 6 also indicate that the 3- midity, temperature, wind speed and pressure is required at ray path loss model taking into consideration the contribu- RADIOENGINEERING, VOL. 23, NO. 3, SEPTEMBER 2014 829

tion of the refracted wave by evaporation duct is more ap- height in this coastal environment. A possible reason for propriate for near sea-surface propagation loss prediction as this observed discrepancy may be because: the measure- compared to the popular 2-ray path loss model particularly ments were made in a coastal environment where stable at- at a distance beyond dbreak. This is because the ducting ef- mospheric conditions could exist, while the P-J formulation fect usually becomes signiﬁcant for long-range over-sea ra- is for an open ocean which may lead to underestimation of diowave propagation. When the propagation distance is be- the evaporation duct height as discussed [4]. low dbreak, the ducting effect is almost negligible where the reﬂection from the sea surface and the direct LoS ray dom- inate. Thus, the 2-ray model should be only used for short- 5. Conclusions range (< dbreak) near sea-surface propagation. This paper reported an experimental investigation of Moreover from the literature, it is found that the oc- near sea-surface LoS radiowave propagation at 5 GHz currence probability of an evaporation duct around nearby through narrowband measurements. Good agreement has marine environments such as the South China Sea is around been observed between the measured results and the pre- h 80%, and the annual average height of the evaporation duct dicted values using a 2-ray path loss model when the propa- is around 7 m to 15 m [26]. These values are similar to those gation distance is less than d . However when the propa- calculated h using the P-J formulation as shown in Fig. 7a. break gation distance increases beyond dbreak, its prediction ability becomes poor. 8

6 Our results and analysis indicated that a 3-ray path

4 loss model taking into consideration the refracted wave by an evaporation duct and the reﬂection from the sea surface Histogram 2 could well predict the trend of the signal strength variations 0 0 10 20 30 40 50 in the tropical marine environments when the propagation Ducting Height (m) distance increases beyond d . The results also show that (a) Calculated ducting height h break most (around 86%) of the estimated effective duct height h 8 e falls into the range of 20 m to 40 m with a median value of 6 30.5 m, which is close to the reported average height h of 4 evaporation duct at a nearby marine environment.

Histogram 2 Therefore for radio planning, near sea-surface LoS ra- 0 0 10 20 30 40 50 diowave propagation loss L in dB could be estimated gener- Ducting Height (m) ally with the following, (b) Effective ducting height he

Fig. 7. Histograms of the calculated duct height h and the em-  2 2 h 2πh h i pirical estimated effective duct height he.  −10log λ 2sin t r , d ≤ d ,  10 4πd λd break L = 2 However, the results in Fig. 7b show that most (around  λ 2  −10log10 4πd [2(1 + ∆)] , d > dbreak. 86%) of the estimated he (supposed to be approximately  equal to/slightly less than h) in this study falls within the range of 20 m to 40 m, with a median value of 30.5 m. This Here, ∆ = 2sin 2πht hr sin 2π(he−ht )(he−hr) as introduced statistic is found to be close to those reported in [26] where λd λd previously. Furthermore, although the break point d = the averaged evaporation duct height h at nearby marine en- break 4ht hr has been deﬁned based on the antenna heights and the vironment is between 25 m to 40 m during similar months λ signal wavelength, the accuracy of the path loss model above of March, April, September and November. Another exam- could vary around this point depending on the sea status. ple for the empirical h at a similar environment is reported in [22], which was based on the data measured between the Palm Islands and the Australian mainland in Northern Queensland which is also with a tropical climate. The duct height h is found to increase as the wind speed increases, Acknowledgements and can be up to 25 m at a wind speed of 10 knots roughly. This work was supported in part by the Defence Sci- Referring to the reported information in [22], the estimated ence and Technology Agency, Singapore. effective duct height he in Fig. 7b which has a measured wind speed within the range of 10 knots to 15 knots looks reasonable. References The values of he in Fig. 7b are then more reliable than the calculated h using the P-J formulation which mainly falls [1] INOUE, T., AKIYAMA, T. Propagation characteristics on line-of- into the range of 8 m to 18 m as shown in Fig. 7a. That is, sight over-sea paths in Japan. IEEE Transactions on Antennas and the P-J formulation tends to underestimate the actual duct Propagation, 1974, vol. AP-22, no. 4, p. 557 - 565. 830 Y. H. LEE, F. DONG, Y. S. MENG, NEAR SEA-SURFACE MOBILE RADIOWAVE PROPAGATION AT 5 GHZ . . .

[2] HITNEY, H. V., HITNEY, L. R. Frequency diversity effects of evap- [20] GUNASHEKAR, S. D., SIDDLE, D. R., WARRINGTON, E. M. oration duct propagation. IEEE Transactions on Antennas and Prop- Transhorizon radiowave propagation due to evaporation ducting. agation, 1990, vol. 38, no. 10, p. 1694 - 1700. Resonance, 2006, vol. 11, no. 1, p. 51 - 62. [3] HEEMSKERK, H. J. M., BOEKEMA, R. B. The influence of evap- [21] ITU-R P.452-14. Prediction Procedure for the Evaluation of Inter- oration duct on the propagation of electromagnetic waves low above ference between Stations on the Surface of the Earth at Frequencies the sea surface at 3-94 GHz. In Proceedings of the Eighth Interna- above About 0.1 GHz. Geneva (Switzerland): International Telecom- tional Conference on Antennas and Propagation. Edinburgh (UK), munication Union, 2009. 1993, p. 348 - 351. [22] KERANS, A., KULESSA, A. S., LENSSON, E., FRENCH, G., [4] GUNASHEKAR, S. D., WARRINGTON, E. M., SIDDLE, D. R., WOODS, G. S. Implications of the evaporation duct for microwave VALTR, P. Signal strength variations at 2 GHz for three sea paths radio path design over tropical oceans in Northern Australia. In Pro- in the British Channel Islands: Detailed discussion and propagation ceedings of the 2002 Workshop on the Applications of Radio Science. modeling. Radio Science, 2007, vol. 42, p. 1 - 13. Leura (Australia), 2002. [5] HITNEY, H. V., RICHTER, J. H., PAPPERT, R. A., ANDER- [23] LEE, W. C. Y. Mobile Communications Engineering: Theory and SON, K. D., BAUMGARTNER, G. B. Tropospheric radio propa- Applications. 2nd ed. McGraw-Hill, 1997. gation assessment. Proceedings of the IEEE, 1985, vol. 73, no. 2, [24] PAULUS, R. A. Practical application of an evaporation duct model. p. 265 - 283. Radio Science, 1985, vol. 20, p. 887 - 896. [6] MALIATSOS, K., CONSTANTINOU, P., DALLAS, P., [25] LEVADNYI, I., IVANOV, V., SHALYAPIN, V. Assessment of IKONOMOU, M. Measuring and modeling the wideband mo- evaporation duct propagation simulation. In Proceedings of the bile channel for above the sea propagation paths. In Proceedings of XXXth URSI General Assembly and Scientific Symposium. Istanbul the 2006 First European Conference on Antennas and Propagation. (Turkey), 2011. Nice (France), 2006. [26] ZHAO, X. L., HUANG, J. Y., GONG, S. H. Statistical analysis of [7] YANG, K., ROSTE, T., BEKKADAL, F., HUSBY, K., TRANDEM, an over-the-sea experimental transhorizon communication at X-band O. Long-distance propagation measurements of mobile radio chan- in China. Journal of Electromagnetic Waves and Applications, 2008, nel over sea at 2 GHz. In Proceedings of the 2011 IEEE Vehicular vol. 22, no. 10, p. 1430 - 1439. Technology Conference. San Francisco (USA), 2011. [8] LEI, Q., RICE, M. Multipath channel model for over-water aeronautical telemetry. IEEE Transactions on Aerospace and Electronic About Authors . . . Systems, 2009, vol. 45, no. 2, p. 735 - 742. [9] MENG, Y. S., LEE, Y. H. Measurements and characterizations of Yee Hui LEE received the B.Eng. (Hons.) and M.Eng. air-to-ground channel over sea surface at C-band with low airborne degrees in Electrical and Electronics Engineering from altitudes. IEEE Transactions on Vehicular Technology, 2011, vol. 60, Nanyang Technological University, Singapore, in 1996 and no. 4, p. 1943 - 1948. 1998, respectively, and the Ph.D. degree from the University [10] JOE, J., HAZRA, S. K., TOH, S. H., TAN, W. M., SHANKAR, J., of York, York, U.K., in 2002. Since July 2002, she has been HOANG, V. D., FUJISE, M. Path loss measurements in sea port for WiMAX. In Proceedings of the 2007 IEEE Wireless Communica- with the School of Electrical and Electronic Engineering, tions and Networking Conference. Kowloon, 2007, p. 1871 - 1876. Nanyang Technological University where she is currently an [11] DONG, F., CHAN, C. W., LEE, Y. H. Channel modeling in maritime Associate Professor. Concurrently, she is also appointed as environment for USV. In Defence Technology Asia 2011. Singapore, Assistant Chair (Student) for the School of Electrical and 2011. Electronic Engineering. Her interest is in channel charac- [12] LEE, Y. H., MENG, Y. S. Empirical modeling of ducting effects on terization, rain propagation, antenna design, electromagnetic a mobile microwave link over a sea surface. Radioengineering, 2012, bandgap structures, and evolutionary techniques. vol. 21, no. 4, p. 1054 - 1059. [13] ITU-R P.1546-4. Method for Point-To-Area Predictions for Terres- Feng DONG received the B.Eng. (Hons.) degree in Electri- trial Services in the Frequency Range 30 MHz to 3000 MHz. Geneva cal and Electronics Engineering from Nanyang Technolog- (Switzerland): International Telecommunication Union, 2009. ical University, Singapore, in 2010, where he is currently [14] TETI, Jr. J. G. Wide-band airborne radar operating considerations for working toward the M.Eng degree. His research interest is low altitude surveillance in the presence of specular multipath. IEEE in wireless channel characterizations and modeling. Transactions on Antennas and Propagation, 2000, vol. 48, no. 2, p. 176 - 191. Yu Song MENG received the B.Eng. (Hons.) and Ph.D. de- [15] ITU-R P.453-10. The Radio Refractive Index: Its Formula and Re- grees in Electrical and Electronic Engineering from Nanyang fractivity Data. Geneva (Switzerland): International Telecommuni- Technological University, Singapore, in June 2005 and cation Union, 2012. February 2010, respectively. From May 2008 to June 2009, [16] LEE, Y. H., DONG, F., MENG, Y. S. Stand-off distances for non- he was a Research Engineer with the School of Electrical and line-of-sight maritime mobile applications in 5 GHz band. Progress Electronic Engineering, Nanyang Technological University, in Electromagnetics Research B, 2013, vol. 54, p. 321 - 336. Singapore. Since July 2009, he has been with the Agency nd [17] PARSONS, J. D. The Mobile Radio Propagation Channel. 2 ed. for Science, Technology and Research (A*STAR), Singa- Chichester (UK): Wiley, 2000. pore. He was firstly with A*STAR’s Institute for Infocomm [18] TIMMINS, I. J., O’YOUNG, S. Marine communications chan- Research as a Research Fellow, and then a Scientist I. In nel modeling using the finite-difference time domain method. IEEE Transactions on Vehicular Technology, 2009, vol. 58, no. 6, September 2011, he was transferred to A*STAR’s National p. 2626 - 2637. Metrology Centre where he is currently a Scientist II. His re- [19] ZHAO, X., HUANG, S. Influence of sea surface roughness on the search interests include electromagnetic metrology, electro- electromagnetic wave propagation in the duct environment. Radio- magnetic measurements and standards, and radiowave prop- engineering, 2010, vol. 19, no. 4, p. 601 - 605. agation.

RF PROPAGATION MEASUREMENT AND MODEL VALIDATION DURING RF/IR SYNERGY TRIAL VAMPIRA

Eric Heemskerk TNO Defence, Security and Safety, Department of Observation Systems P.O. Box 96864, 2509 JG The Hague, The Netherlands

ABSTRACT

The member nations of AC/323 SET-RTG056/RTG32 on Integration of Radar and Infrared for Ship Self Defence have performed the Validation Measurements for Propagation in the Infrared and Radar (VAMPIRA). The objective was to get insight into the radar and infrared synergy concentrated on propagation in a coastal environment including horizontal inhomogeneity and to validate radar and infrared propagation models. The trial was held in the period 25 March-5 April 2004 near Surendorf Germany. As part of the trial TNO made RF 1-way transmission measurements, 24 hours/day during the whole trial period. The transmission path over the Eckernförder Bucht in Northern Germany had a length of 8.2 km. The transmitted signal was a sweep consisting of 6 frequencies i.e. 3.36, 5.32, 8.015, 9.7, 13.45, and 15.71 GHz. The transmitter height was 11.5 m, the receiver height 6.4 m above ‘normal null’. At each end of the path a meteorological station was installed measuring every 30s the air temperature, relative humidity, air pressure, wind speed and wind direction. About halfway the path the TNO meteo buoy was anchored measuring air temperature and relative humidity at 5 heights between 0.65 and 5.15m above the sea surface. Also the sea water temperature was measured by the buoy on a depth of 1m below the sea surface. The effects of evaporation ducting at the propagation at the various frequencies were clearly demonstrated. Some times very deep fadings were present at 13.45 and 15.71 GHz where at the same time almost no effect at 3.36 and 5.32 GHz was observed. The measured propagation at 15.71 GHz was more enhanced than at 13.45 GHz due to the ducting conditions and the elevation angle of the transmitter and receiver antenna. In several sample cases the 1-way propagation factors are computed for every 5 minutes using the propagation model TERPEM (Signal Science) and the vertical refractivity profiles computed by the TNO model TARMOS. The 1- way computed propagation factors compared very well to the measured data at all frequencies, although the computed fadings were not always as deep as the measured ones. A first promising result has been obtained computing the observed height of the RF source under various atmospheric conditions using the transmission phases computed by TERPEM.

Keywords: electromagnetic propagation, model validation, measurement, radar, atmospheric refraction, ducting

1. INTRODUCTION

In the last decade many operations of the navies have been moved from open ocean towards littoral/warm waters. Operating in littoral environment includes new threats. Also the background and propagation under atmospheric conditions with a strong horizontal inhomogeneity have a severe impact on the radar and infrared sensor performance, on the detection / classification / recognition of small targets in the coastal region and on the ship self defence. Individual onboard sensors are all differently affected by the environment, depending on height, frequency, bearing, range etc. Simultaneous exploitation of radar and infrared sensors and multi-sensor fusion can overcome the difficulties with respect to target detection/recognition/classification by the individual sensors in a coastal environment. Within this context the so-called radar-infrared synergy with respect to propagation through the marine boundary layer plays an important role. The simultaneous occurrence of RF and IR sub- and superrefractive conditions is a part of the synergy and complementarity of the RF and IR wavelengths for detection of targets.

The member nations of NATO AC/323 SET-056/RTG32 are investigating this synergy. The first part of the RTG study included a review of existing data. It revealed that there was only one case where three from the four possible combinations of radar and infrared sub- and superrefractive conditions occurred. But the trial only contained radar and

Optics in Atmospheric Propagation and Adaptive Systems VIII, edited by Karin Stein, Anton Kohnle Proceedings of SPIE Vol. 5981 (SPIE, Bellingham, WA, 2005) 0277-786X/05/$15 · doi: 10.1117/12.637616

Proc. of SPIE Vol. 5981 598107-1

environmental measurements. Therefore the RTG members organised the Validation Measurements for Propagation in the Infrared and Radar (VAMPIRA) to obtain sufficient and appropriate experimental data on simultaneous radar and infrared propagation [1].

2. VAMPIRA TRIAL

VAMPIRA was held from 25 March-5 April 2004 near Surendorf (Schwedeneck) Germany. The location and time period were selected from the results of a study on the refractive conditions during previous trials and using a 5-year meteorological database of the Surendorf area. The trial area with an overview of instrumentation and assets is given in figure 1.

The objectives of VAMPIRA were to obtain a complete data set of simultaneously measured environmental parameters, and radar and infrared propagation factors to get insight into the radar and infrared synergy concentrated on propagation in a coastal environment including horizontal inhomogeneity and to validate radar and infrared propagation models.

Denmark

Baltic OsiceeSea

Flensburg

EokeFnfdrde - NdISd- •• •Schwedeneck kQJ,) FWG drift buoy Borgstedl • AsehaL? KieI-FIiedFichsorl S Kiel-ElleLbek WTD71 Bookniseck North Sea Plan •MS1LL[f / Elpersbu••ltel P

Lbbeck propagation path ship 8.2 km trajectory ~40 km Hamburg

FGAN IR black bodies corner reflectors TNO IR-source Surendorf

TNO German Navy WTD71 Pier WTD71 Kreienberg meteo buoy small targets Figure 1: VAMPIRA trial area, meteo buoys and assets.

The measurements included - simultaneous 1-way radar and infrared transmission measurements over a 8.2 km path between Surendorf and Bookniseck 24 hours/day during the whole trial period, - simultaneous radar and EO measurements on radar corner reflectors respectively black bodies and visible lights on the ship Stollergrund during inbound and outbound runs (path length max 40 km) at various times of the day, - simultaneous radar and EO measurements on an IR source suspended under a helicopter at various times of the day, - simultaneous radar and EO measurements on small targets, - meteorological measurements on the pier and Kreienberg in Surendorf (24 hours/day), - meteorological measurements at Bookniseck (24 hours/day), - meteorological measurements by a buoy about halfway Surendorf-Bookniseck (24 hours/day), - environmental measurements by 2 free drifting buoys near the ship trajectory during the ship runs (>8 hours/day), - radiosoundings from the pier and from the ship Helmsand (once per hour during the ship runs), - 1-way RF propagation measurement with a receiver on the Stollergrund.

Proc. of SPIE Vol. 5981 598107-2

This paper only deals with the RF 1-way propagation measurements between Surendorf and Bookniseck including the corresponding meteo. First results have been presented in [2].

For the 1-way RF propagation measurements the transmitter was positioned at Bookniseck, the receiver on the pier on the WTD71 (Wehrtechnische Dienststelle71) facility at Surendorf (Schwedeneck) Germany. The height of the transmitter and receiver antenna was 11.5 and 6.4 m above ‘normal null’ respectively. The 1-way RF propagation measurements were made by TNO once per minute during the trial period. The path length was 8.2 km. The transmitted signal included a sweep over 6 frequencies i.e. 3.36, 5.32, 8.015, 9.7, 13.45, and 15.71 GHz. Each frequency was transmitted during 9 s. The sweep repetition time was 60 s. At each end of the path a meteorological station was installed measuring every 30 s the air temperature, relative humidity, air pressure, wind speed and wind direction. About halfway the path the TNO meteo buoy was anchored. This buoy measured the air temperature and relative humidity at 5 heights from 0.5–5 m above the sea level once per minute, but the data was averaged over 5 minutes. Also the sea water temperature was measured by the buoy at a depth of 1 m.

3. METEOROLOGICAL AND DUCTING CONDITIONS ALONG THE PATH

3.1 Meteorological conditions The meteo buoy was positioned about halfway the path, which is a good position for obtaining representative data of the meteorological conditions along the transmission path. Land effects did affect temperature and humidity data that were collected at each end of the path. For instance the air temperature measured at the pier was on average about 2 °C higher than the one measured by the buoy. Of course this depends on the wind direction. The data from the buoy sensor at 5.15 m was selected for use in the analysis. Since the buoy did not contain an anemometer the meteo data set was completed by the wind speed measured at the pier. This was allowed because a comparison of the wind speeds measured at the Kreienberg (in Surendorf), the pier and Bookniseck showed only small differences. Figure 2 gives the measured air and sea temperature and the relative humidity from the buoy and the wind speed and wind direction from the pier during the whole trial period. The sea temperature showed only small variations during the trial period. The air temperature however varied largely between 0° and 8°, resulting in positive as well as negative ASTD values during the trial. These correspond to stable and unstable conditions respectively. Also the relative humidity showed large variations. In the first half of the trial a few rain events took place after a few days with moderate values of humidity. At the end of the trial a few time periods occurred with a sudden drop of humidity and a simultaneous rise respectively drop of the air temperature that might have given rise to radar subrefractive conditions as will be discussed later on. The wind was blowing from the open sea during about 24% of the time. During about 20% of the trial the wind was coming from directions between 90° and 120° i.e. partly from open sea but affected by the land. The remaining of the time the wind was blowing from the land. Most of the time the wind direction was eastern. During 37% of the time the wind direction varied between 180° and 240°. Only during 8% of the time there was western to northern wind. The wind speed varied strongly during the trial between 1 and 10 m/s. It means that based upon the observations made near Surendorf and Bookniseck the atmospheric conditions changed quite a lot during the trial including land effects as well during approximately half the trial period. A detailed description of the atmospheric conditions based upon buoy measurements in open sea at a range beyond 10 km from Surendorf is given in [3]. The latter study indicates that there might have been rather strong horizontally inhomogeneous conditions during the trial period all over the trial area towards northern directions i.e. along the ship runs.

At the beginning of the trial period the ASTD was negative, at the end of the trial positive ASTD values were met. Especially in the first part of the trial the relative humidity showed large variations over the day (60-100%). Noteworthy is the 3rd April where around 06:00 UTC the air temperature increased within a few hours from 4 to 8°C, while the relative humidity dropped from 80% to 60% within an hour and increased within the next 5 hours till about 90%. A comparable phenomenon occurred to a less extent also on 4th April around 11:00 UTC. Both phenomena have been observed also by the meteo station on the pier, at the Kreienberg and in Bookniseck. On the 4th April the sudden temperature and humidity decreases coincide with an also sudden change in wind direction from south (coming from the land) towards 240° (coming almost along the Eckernförder Bucht).

Proc. of SPIE Vol. 5981 598107-3

Vampire, NL mneteo station at Surermdorf pier OZ 2 2 H z 0 a 2

—mind speed —mind direction HO -, 2 - 360 or 14 - 315

00-S 31 12 - 270 0 (% ( 10 - 225 0 - 180 or

humidity - 135 rd.

temperature -90 0 - 45 r, 0 O-MO00J 00:00 0000 25/03/2004 UTC

Figure 2: Meteo conditions during the trial period. Left: air temperature, sea temperature and relative humidity measured by the buoy sensor at a height of 5.15 m. Right: wind speed and wind direction from NL meteo station at the pier.

3.2 Refractive conditions In order to obtain insight into the refractive conditions during the trial two methods have been applied. The Canadian method [4] plots the air-to-sea water vapour pressure difference (ASVPD) as a function of the air-to-sea temperature difference (ASTD). Four regions are distinguished: - region 1: simultaneous RF superrefractive and IR subrefractive conditions. This is the most common situation. - region 2: simultaneous RF superrefractive and IR superrefractive conditions. - region 3: simultaneous RF subrefractive and IR superrefractive conditions. This region is the most interested one, because radar may experience a reduced detection range, while IR may experience at the same time an increased detection range. - region 4: simultaneous RF subrefractive and IR subrefractive conditions. Figure 3 shows the ASTD-ASVPD relation for the whole trial period using the meteo data from the buoy sensor at 5.15 m. The colours indicate the conditions on the various days. On several days the refractive conditions significantly changed during the day. Most of the time RF superrefractive conditions occurred, while simultaneously IR experienced sub- or superrefraction. But the plot indicates also that there might have been a few occasions along the propagation path with simultaneous RF subrefractive and IR superrefractive conditions.

The French method [5] uses the M-gradient to compute the duct height when dM/dh=0. When no zero M-gradient is found the N-gradient is used to compute the height where dN/dh=0. This latter height is called refractive height and is given a negative sign. The method can be applied for both radar and IR/EO. A negative refractive height is not a physical height but an indication for a possible existence of subrefractive conditions.. For the present study the evaporation duct /refractive heights for radar only have been computed by TARMOS [6] and plotted in figure 4. It can be seen that there were time periods on 3 and 4 April 2005 with possibly radar subrefractive conditions. Correlating these with the temperature and relative humidity in figure 2, it is clear that these radar subrefractive conditions occurred when there was simultaneously a high positive ASTD and a moderate relative humidity. Also it became evident that there is a very good match between these periods and the data points in region 3 in the ASTD-ASVPD plot in figure 3.

Both the Canadian and the French method give information on the refractive conditions, but the results are dependent on the measurement height of the meteorological parameters within the duct. The Canadian method can be applied rather easily on board navy vessels and gives almost instantaneously information on the refractive conditions for both radar and IR without using a bulk model. The French method can also be applied on board, but one needs a bulk model to transfer the meteorological parameters to evaporation duct or refractive height.

A third method developed by TNO based upon refractivity gradients is presented in [6]. A more detailed analysis showed that the radar subrefractive conditions might have been met on 3 days covering in total about 4.5 hours out of 225 hours RF measurements. The most interesting day in this respect is 4 April 2005

Proc. of SPIE Vol. 5981 598107-4

S mctn. Ft nfl S Varrrpira, TARMOS coerpeted evaporation duct/refractive height effect of 00000r height. roeteo NL-buoy. 25 Mar - 4 April 2005 25-Mar-04 26-Mar-04 27-Mar-04 28-Mar-04 29-Mar-04 30-Mar-04 31-Mar-04 01-Apr-04 02-Apr-04 03-Apr-04 04-Apr-04 05-Apr-04 80 region3 60regi radar F rsubrefractrve 4.0 EO superrefractive 20 0.0 -20 -4.0 -6 0 -8 0 -10regwn2-8 -6 -4 -2 0 2 4 6 8 10 ASTD ( UTC

Figure 3: Air-Sea Water Vapour Pressure Difference vs Air- Figure 4: TARMOS computed evaporation duct/refractive Sea Temperature Difference for radar and EO height for radar during the Vampira trial as suggested propagation during the Vampira trial as suggested by by [10] from the sensor on the buoy at a height of [9] from the sensor on the buoy at a height of 5.15 m. 5.15 m.

3.3 Effect of height of meteo measurements On the buoy meteo sensors were mounted at different heights above the sea surface. These sensors are measuring the air temperature and relative humidity at different positions on the respective vertical profiles. This of course may affect the value of the computed duct height by the TNO model TARMOS [4]. As an example the data from 4 April 2004 have been selected to investigate this.

In figure 5 the air temperature (upper left), relative humidity (upper right) and the resulting duct heights (lower left) obtained from meteo data measured at 2.15, 3.15 and 5.15m above the sea surface are compared. Also the effect on the 1-way propagation factors has been computed in the lower right part of this figure. For comparison the air temperature and relative humidity measured at about 7.5 m height above ‘normal null’ at the pier are also given. As mentioned before there is a sudden decrease in air temperature around 11:00 UTC. All buoy sensors and the pier sensor have measured this. This temperature drop coincides with a very quick turn of the wind direction from south towards west- south-west, meaning that the wind is blowing over the Eckernförder Bucht. Clearly one can see the increase in temperature (on this day) with height, until about 11:00 UTC where after there is hardly any difference between the temperatures as measured by the considered buoy sensors. In the same time period a decrease of relative humidity is observed with increasing height. The effect on the (computed) duct height is small except for the period between 7:00 ad 11:00 UTC where the computations indicate a ‘negative’ refractive height. The computed effect on the 1-way path losses as shown in the lower left part of figure 5 is negligible.

Proc. of SPIE Vol. 5981 598107-5

Vanpira, 4 April 2004, NL buoy-air tenperature and NL pier-wind direction Vampira, 4 April 2004, NL buoy-relative humidity and NL pier-wind direction

—Tair 2.15n —Tair 3.15n Tair 5.15n Tair pier WD pier —RH 2.15m —RH 3.15m RH 5.15m RH pier WD pier 330

300 z 270

70 240 60 I 210 50 0! 180 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 UTC UTC

Vampira, 4 April 2004, NL-buoy, TARMOS refractive height Vanpira, 4 April 2004, NL-buoy, conputed i-way propagation factors 10— -I—Tarmos 2.15m —Tarmos 3.15m Tarmos 5.15m — i-way pf2.15n — i-way pf3.15n i-way pf5.15n

______

_10 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 UTC UTC

Figure 5: Effect of height of meteo sensors on the computed duct height on 4 April 2004.

4. RF MEASUREMENTS

The RF measurements were made from 25 March until 5 April 2004. The received powers at 6 frequencies were averaged over 9 s and the average was recorded once per minute. The transmitter and receiver antenna covered the whole frequency band from 2-18 GHz and were optically aligned to each other (line of sight). In a later stage however it proved that the transmitter antenna had a vertical offset of 1.7°. Calibration of the link was done by measuring the received power at the 6 frequencies after connecting the output of the transmitter antenna cable to the input of the receiver antenna cable via an appropriate calibrated attenuator. The attenuation value was such that the received powers during calibration were in the same order as the received powers during the actual measurements. By correcting the received powers with the calibration values, the antenna directivity and the free space path loss the RF propagation factors have been obtained. The measured 1-way propagation factors at the 6 frequencies during the entire trial period and the computed evaporation duct/refractive height are given in figure 6. No measured data is available from 3 April 16:25 till 4 April 09:08 UTC due to a electrical power failure.

The dependency of the RF propagation on the evaporation duct height is clearly observed in figure 6. For the lower frequencies (3.36 and 5.32 GHz) the measured propagation factor shows about the same behaviour as a function of time with propagation factors up to about 5 dB. The 8.015 and 9.7 GHz are more affected by the atmospheric conditions. Some signal reductions in the order of 5 dB have been observed during short time periods. However at the same time the signals at 13.45 and 15.71 GHz showed very strong propagation effects. Note that in figure 6 the measured propagation factor for 15.71 GHz is larger than for 13.45 GHz. The explanation is the elevation of the transmitter and receiver antenna resulting in a frequency dependent antenna pattern loss. I.e. depending on the frequency, the RF signal is transmitted and received in the mainlobe (lower frequencies) or one of the sidelobes (higher frequencies).

Proc. of SPIE Vol. 5981 598107-6

meteorological observations of wind atmospheric refraction profiles of radar and optical radiation in the marine surface layer. Based on standard (micro)-meteorological parameters. The results are used to TARMOS [7] and the Signal effects propagation ofthe and study validation the For 5.1 Model description vapour water, the scaling water,the scaling parameters vapour computes the vertical profiles of the wind speed, temperature and humidity, turbulent fluxes of momentum, heat and Left:Measured 1-way propagation factors duringVampira 3.36, at 5.32, 8.015, 9.7, 13.45and15.71 GHz, Figure 6: radio linksand radar systems. Itis ba The TERPEM propagation package is atool for the forecasting and analysis of refraction, ducting and terrain effects on mechanical mixing. or to alack due convection of absent islimited or mixing where vertical to condition the atmospheric inherent but model the is problem This not of a mathematical duct height. a the predict mayproblemreliably models to have bulk conditions stable the During has model used. been the Davidson-Kondo analysis present the For Charnock). or methods using different stabilityfuncti refractivity radar and IR/EO have experienced superre parameter and computed duct height on this day are presented in figure 7. From the ASTD-ASVPD computations both about 95%. This caused the ductheight to increase to fr decreased humidity relative the while 2° till about increasing about between existed conditions m. Stable of 0-1 height duct Casestudy 28 March 2004 UTCwith acomputed 08:00 tillabout atmospheric conditions almost neutral a day with concerns The first casestudy 5.2 programs can be used independently of each other. techniques. Analysis ofthe output produced by the TERPEM calculations is done by the TGRAPH program. Thetwo

1-way propagation factor(dB

26/03/040 0 0 0 0 0 0 00:00 TARMOS. path length 8.2 km, transmitter height 11.5m, receive N (z) and modified refractivity I0 '" < 0 I w

Science model TERPEM [8] have been used. model[8] havebeen Science TERPEM 0

5. 3

speed, air temperature and humidity, an sed onstate-of-the-arth u 0 *, ons (Businger-Dyer and Kondo) and di t * and * and M fractionthis day on(see figure 3). RF PROPAGATION PREDICTION PREDICTION RF PROPAGATION (z). Currently, TARMOS provides a Proc. of SPIE Vol.5981 598107-7 I0 q *, the*, structure function parameter a maximuma computed value of 17 m.meteorologicalThe in dependence of the refractive conditions the TNOmodel conditions therefractive of in dependence I0 estimate the vertical profiles of the index of refraction index refraction of ofthe profiles vertical estimate the ybrid models combining para 08:00 and 14:00 UTC. In the latter period the ASTD UTC. the was In and latterperiod 14:00 the 08:00 duct om almost 100% to 75% increasedom to75% and at 12:00 100% almost next to (n) r height 6.4m. Right:Duct height computed by TARMOS is a research code to predict predict to code a research is TARMOS d the sea surface temperature, TARMOS fferent solution methods (based on selection of six different solution solution different six of selection C T 2 , C bolic equation and ray-trace Q 2 and C TQ and other C n (z), DN ,

iaa was .n ia (Afl nS( Vanpira, 28 March 2004, duct height, TARMOS —WS astd —RH — tarmos-m4 astd 80 -, 100 7.0 .r 60jr ¼ 5.0 !!i 4.0 t ¾ 'J 4.0 30 t4 3.0 20 rI, 20 <0) 1.0 r 30 e 1.0 0.0 .X 0.0 -1.0 -1.0 -2.0 TEL -2 0 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 UTC UTC

Figure 7: Left: measured relative humidity (buoy) and ASTD (buoy) and wind speed (pier) on 28 March 2004. Right: duct height computed by TARMOS and ASTD (buoy).

Vanpira, 28 March 2004, i-way propagation factor Vanpira, 28 March 2004, i-way propagation factor TX height ii .5n, Rx height 6.4n, range 8200n, Rx elev 1 .1, TX elev 1 .7 TX height ii .5n, Rx height 6.4n, range 8200n, Rx elev 1 .1, TX elev 1 .7

3.36 GHz-M —3.36 GHz-C-M4 5.32 GHz-M —5.32 GHz-C-M4

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 UTC UTC

8015 GHz-M —8015 Gf-C-M4 9.7 GH-M —9.7 GI-lz-C-M4 0.0 0.0 -5.0 -5.0 -10.0 -10.0 -15.0 -15.0 -20.0 -20.0 -25.0 -25.0 -30.0 -30.0 -35.0 -35.0 -40.0 -40.0 -45.0 -45.0 -50.0 -500 00:00 0200 0400 06:00 0800 10:00 1200 1400 1500 1500 20:00 2200 00:00 0000 0200 0400 06:00 0800 10:00 1200 1400 1600 1800 20:00 22-00 0000 rrc rrc

13.45 G1z-M—13.45 Gf-C-M4 15.71 Glz-M —15.71 Gf-C-M4 0.0 0.0 -5.0 -5.0 -10.0 -10.0 -15.0 -15.0 -20.0 -20.0 -25.0 -25.0 -30.0 -300 -35.0 -35.0 -40 0 -40 0 -45.0 -45 0 -50.0 -50 0 00:00 0200 0400 06:00 0800 10:00 1200 1400 1600 1800 20:00 2200 00:00 0000 0200 0400 06:00 0800 10:00 1200 1400 1600 1800 20:00 2200 0000 rrc rrc

Figure 8: Computed (C-M4) and measured (M) 1-way propagation factors on 28 March 2004 at 3.36, 5.32, 8.015, 9.7, 13.45 and 15.71 GHz, horizontal polarisation.

Proc. of SPIE Vol. 5981 598107-8

The M-profiles computed by TARMOS have been used in TERPEM to compute the 1-way propagation factors for every 5 minutes during the day.

In the computations the actual geometry of the measurements is taken into account i.e. including the elevation of both the transmitter and receiver antenna. The computed and measured 1-way propagation factors for the 6 frequencies are shown in figure 9.

From this it is clear that under unstable and neutral conditions the predictions are in good agreement with the measurements as a function of time. During the period of stable conditions the model computed only a slightly reduced propagation factor at 9.7 GHz, while the measurements showed a reduction of about 10 dB. At 8.01 GHz the computations did not show a ‘dip’ at all. At the higher frequencies the model is in good agreement with the measurements during neutral and unstable atmospheric conditions. The measured decrease of the propagation factor at 13.45 GHz under stable conditions is predicted by the models. At 15.71 GHz the models predict several deep fadings in the period with stable conditions that do not show up in the measurements. This is still to be investigated in more detail.

5.3 Case study 3 April 2004 A second case study concerns 3 April 2004. The ASTD-ASVPD diagram computed from the buoy data at 5.15 m indicates that during this day RF subrefractive conditions were likely to exist between 17:00 and 18:00 UTC. During almost the entire day the ASTD was between 0° and 4.7°. The meteo data and the computed evaporation duct/refractive height are given in figure 9. Note that only positive values of the refractive height are corresponding to evaporation duct heights. The computed and measured 1-way propagation factors are given in figure 10. From 06:00 till 13:00 UTC there were stable atmospheric conditions. The wind speed was about 3m/s in that period.

All M-profiles computed by TARMOS have been used to compute the propagation factor by TERPEM. Despite the problem that bulk models have to predict reliable duct heights under stable conditions there is a good to very good match between computed and measured propagation factors as a function of time (figure 10), even under stable conditions.

Varpira, 3 April 2004, NL rneteo buoy (ASTD, RH) and pier )WS) C Ipdv 7009 ;onp 016!qSOV8VL

—WS astd —RH — 7w-souje; p;se 80 09 0 7.0 go 09 0 E 60 80 5.0 70 007 2 4.0 60 000 3.0 E 2.0 40 = 009 o 1.0 30 0.0 F- 00 20 -1.0 10 0•0 -2 0 0.0v 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 0000 0000 0970 00:90 0890 OOO OO0 OO4 OO9 00:9 0800 0800 00:00 UTC oJ-n

Figure 9: Left: Measured relative humidity (buoy), ASTD (buoy) and wind speed (pier) on 3 April 2004. Right: refractive height computed by TARMOS and ASTD (buoy).

Proc. of SPIE Vol. 5981 598107-9

Varnpira, 3 April 2004, 1-way propagation factor Varnpira, 3 April 2004, 1-way propagation factor TX height 11 .5rn, Rx height 6.4rn, range 8200rn, Rx elev 1 .1, TX elev 1 .7 TX height 11 .5rn, Rx height 6.4rn, range 8200rn, Rx elev 1 .1, TX elev 1 .7

3.36 GHz-M —3.36 GHz-C-M4 5.32 GHz-M —5.32 GHz-C-M4

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 UTC UTC

8.015 GHz-M —8.015 GHz-C-M4 9.7 GHz-M —9.7 GHz-C-M4-nea 0.0 0.0 -5.0 -5.0 -100 -10.0 -15.0 -15.0 -20.0 -20 0 -25.0 -25.0 -30 0 -30 0 -35 0 -35.0 -40.0 -40.0 -45.0 -45.0 -50.0 -50.0 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 UTC UTC

13.45 GHz-M —13.45 GHz-C-M4-017 15.71 GHz-M —15.71 GHz-C-M4

0.0 0.0 -5.0 -10.0 1_ -15.0 -15.0 -20.0 -20.0 -25.0 -25.0 -30 0 o-300 -35 0 -35 0 -40.0 -40.0 -45.0 -45.0 -50.0 -50.0 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 UTC UTC

Figure 10: Computed (C-M4) and measured (M) 1-way propagation factors on 3 April 2004 at 3.36, 5.32, 8.015, 9.7, 13.45 and 15.71 GHz, horizontal polarisation.

6. INCIDENCE ANGLE

The study on radar and infrared/EO synergy with respect to propagation is done using parameters that are common to both radar and EO propagation: transmission loss and angle of arrival or observed elevation position of the RF respectively IR/EO source can be the proper parameters. So far the RF part of the study concentrated on the transmission loss or propagation factor because unfortunately the RF measurements did not include angle-of-arrival measurements. But models can be used to simulate what, in our case, a phase interferometer would have seen under the trial conditions. A study has been started to investigate the phase behaviour of the received radar signal under the various ducting conditions. A special version of TERPEM is used that computes and outputs the complex propagation

Proc. of SPIE Vol. 5981 598107-10

factor. The complex propagation factor is defined as the ratio of the calculated (complex) electromagnetic field E to the (complex) free space field E0 and contains the transmission loss and phase.

The phase at two different heights has been used to compute the angle-of-arrival (AOA) of the received RF signal simulating a two element phase interferometer by using ⎛ λ.()ϕ ± k2π ⎞ AOA = ASIN⎜ ⎟ k=1, 2, 3, ..... ⎝ 2πd ⎠ where λ = radar wavelength in metres ϕ = phase difference between the received RF signals at the heights of the interferometer antennas in radians d = height difference between the two interferometer antennas in metres

Note that the antennas are considered to be a point receiver. A baseline of 20 cm has been chosen. Using the computed AOA the height at which the RF source would have been observed under the various atmospheric conditions have been simply computed (free space conditions) by

H observed = R*TAN(AOA) where R = range from RF source to receiver in metres.

VimpIr., 0410412004, slmul.ted Th height, Interferometer height 6.4m effect of interferometer baseline

—20cm —40cm —100cm 200cm Tarmos refr. height — 1-way pf

E = 0, -D =ci, 0 0 Ct ci, = 0 0, a =ci'

0:00:00 4:00:00 8:00:00 12:00:00 16:00:00 20:00:00 0:00:00 UTC

Figure 11: Computed ‘observed’ height of the RF source, TARMOS refractive height and 1-way propagation factor assuming a two element interferometer at 6.4m height, 9.7 GHz, 8.2 km range and interferometer baseline of 20, 40, 100 and 200 cm.

A first result of the computed ‘observed’ height of the RF source on 4 April 2004 is given in figure 11 for 9.7 GHz, a range of 8.2 km and baselines of 20, 40, 100 and 200 cm. The figure also shows the TARMOS duct height and the computed 1-way propagation factor.

At first it can be seen that the observed height of the RF source varies under the considered conditions on the 4th April. There is a good correlation with the computed duct height. When the duct height gets lower the observed RF source height increases due to the fact that the AOA becomes more positive at the considered range and interferometer height. Also there are very small differences between the results for the considered baselines. However is must be noted that there may occur substantial differences when the interferometer height is changed and the two antennas are in different propagation lobes. A more detailed study is necessary to fully investigate the behaviour of the AOA under various atmospheric conditions and dependency on interferometer height, range and frequency. These first obtained results are promising.

Proc. of SPIE Vol. 5981 598107-11

7. SUMMARY AND CONCLUSIONS

The member nations of NATO AC/323 SET-056/RTG32 have held the propagation experiment VAMPIRA to demonstrate the radar-infrared synergy and/or complementarity. TNO has made amongst other things 1-way RF propagation measurements simultaneously at 6 frequencies from 3-16 GHz once per minute from 25 March-5 April 2004. Simultaneously, meteorological measurements were made by meteo stations ashore at both ends of the path and about halfway by sensors at 5 different heights on a buoy. A complete set of RF and meteo data covering a period of 225 hours has been obtained. A Canadian and French method to obtain insight into the refractive conditions i.e. RF and IR/EO synergy have been applied to the collected meteorological data. The results of both methods agreed very well and showed that there were most of the time RF superrefractive conditions and simultaneously IR sub- and superrefractive conditions along the propagation path. Only during about 4.5 hours RF subrefractive conditions existed.

The effects of evaporation ducting at the propagation at the various frequencies were clearly demonstrated. Some times very deep fadings were present at 13.45 and 15.71 GHz where at the same time almost no effect at 3.36 and 5.32 GHz was observed. The measured propagation at 15.71 GHz was more enhanced than at 13.45 GHz due to the ducting conditions and the elevation of the transmitter and receiver antenna.

The TNO TARMOS model 4 (Davidson and Kondo) was used to compute the M-profiles, based upon temperature and humidity data from the buoy and wind speed data measured at the pier. The 1-way propagation factors, computed by a combination of TARMOS and TERPEM compared very well to the measured data at all frequencies, although the computed fadings not always were as deep as the measured ones.

First results of computing the angle of arrival of the received RF signal by using the transmission phase are promising and showed an elevation variation of the position of the transmitter antenna due to varying ducting conditions.

Future work will include completing of the computations for the entire measurement period including ducting effects on the angle-of-arrival, correlating the predictions and measurements, comparing the RF data with the IR transmission data that were simultaneously measured along the same path and classifying the results to combined RF-IR refractive conditions.

ACKNOWLEDGEMENT

I like to express my appreciation to the members of AC/323 SET-056/RTG32 that have participated to the VAMPIRA trial, the WTD71 for hosting the trial and providing the facilities, ships and support during the trial.

REFERENCES

[1] H.J.M. Heemskerk, “Vampira Radar and Infrared Propagation Synergism Trial”, Eur. Conference on Propagation and Systems, 15-18 March 2005, Brest France. [2] H.J.M. Heemskerk, “VAMPIRA RF Propagation measurement and model validation”, Eur. Conference on Propagation and Systems, 15-18 March 2005, Brest France. [3] J. Foerster and J. Riechen, “In-situ measurements of marine boundary layer refractive variability during the Vampira experiment”, Eur. Conference on Propagation and Systems, 15-18 March 2005, Brest France. [4] L. Forand “A Study into Infrared Search and Track & Radar Synergy», DRDC Valcartier TR 2003-044, June 2003. [5 J. Claverie, and Y. Hurtaud, “Refractive height”, presented to the AC/323 SET056/RTG32, Nov 2004 [6] G.J. Kunz, H.J.M. Heemskerk, L. van Eijk, “Comparison of atmospheric refraction at radar and optical wavelengths”, SPIE Europe International Symposium on Remote Sensing 2005, 19-22 Sep 2005, Bruges Belgium. [7] G. J. Kunz, “A bulk model to predict optical turbulence in the marine surface layer”, TNO report FEL-96-A053, April 1996. [8] M. F. Levy and K. H Craig, “Millimetre-wave propagation in the evaporation duct”, Agard Conference proceedings no. 454, pp. 26-1; 26-10, Copenhagen, Denmark, 9-13 October 1993.

Proc. of SPIE Vol. 5981 598107-12

Measurements of Vertically Polarized Electromagnetic Surface Waves Over a Calm Sea in HF Band. Comparison to Planar Earth Theories Mathilde Bellec, Stéphane Avrillon, Pierre Yves Jezequel, Sébastien Palud, Franck Colombel, Philippe Pouliguen

To cite this version:

Mathilde Bellec, Stéphane Avrillon, Pierre Yves Jezequel, Sébastien Palud, Franck Colombel, et al.. Measurements of Vertically Polarized Electromagnetic Surface Waves Over a Calm Sea in HF Band. Comparison to Planar Earth Theories. IEEE Transactions on Antennas and Propagation, Institute of Electrical and Electronics Engineers, 2014, 62, pp.3823 - 3828. 10.1109/TAP.2014.2317493. hal- 01114356

HAL Id: hal-01114356 https://hal.archives-ouvertes.fr/hal-01114356 Submitted on 9 Feb 2015

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Measurements of Vertically Polarized This propagation phenomenon should present attractive and useful Electromagnetic Surface Waves Over a Calm Sea features for industrial applications because the surface waves propagate along the surface of Earth and beyond the radio-electric in HF Band. Comparison to Planar Earth Theories. horizon. Few examples already exist mainly operating in VLF, LF

and in the HF band. Thanks to these properties, the surface waves M. Bellec, S. Avrillon, P.Y. Jezequel, S. Palud, F. Colombel, Ph. Pouliguen allow communication in hard environments (forest…) or target Abstract —Radio communication over Earth along mixed-paths in the HF detection at very low elevation and at large distances. band is a relevant subject today. In this paper, we present measurements of electric field propagating over sea water in HF Band compared to K.A. This paper presents the measurements of surface waves over the sea Norton, R.W.P. King and G. Millington’s theories, thanks to a reliable in HF band and gives the comparison between the experimental measurement setup. The transmitting antennas are located on the coast studies and the theoretical models provided in the literature, while the receiver antenna is installed on a boat steering a constant course. The electric field measurements are carried out with a loop especially in order to validate the surface wave decrease as 1/d and antenna and we measured the field strength attenuation versus distance 1/d² along the path, and the characteristic distances introduced by between the transmitter and the boat along a sea water path. In order to King. take into account the media change (the coast and the sea water), Millington’s solution has been added to King’s and Norton’s theories Section II presents the surface-wave propagation theories proposed with the planar Earth model. The measurements performed at three by Norton, King, and Millington where the radiating element is a frequencies (10 MHz, 20 MHz and 30 MHz) and the calculations are in a vertically polarized antenna located above the ground. Section III good agreement. At 10 MHz, the “smoothly” attenuation is shown and is presents measurements realized over the sea. The measurement very well correlated with the theory. The EM field decrease as 1/d² has been clearly observed at 20 MHz and 30 MHz. process is carefully described including the design of the antennas used to radiate the surface waves. Then, section IV provides a Keywords— HF band measurements, surface-wave propagation, comparison between the measurements and the theoretical results. ground-wave field, vertical electric dipole, planar Earth model. II. PROPAGATION THEORIES I. INTRODUCTION This section describes theoretical approaches and then provides an Introduced at the beginning of the last century, surface-wave interpretation of the surface wave propagation theories on a planar propagation has been largely investigated starting by Sommerfeld [1] Earth thanks to the research of Sommerfeld [1], K.A. Norton [2], [3], and followed by Norton [2], [3], King [4]-[6], Wait [7], [8]. These R.W.P. King [4]-[6], and G. Millington [9], [10]. These authors have pioneer researchers provided mainly theoretical studies and analytical provided electromagnetic field formulas radiated by an infinitesimal solutions to this problem. Sommerfeld started by calculating the EM vertical electric dipole located at a specified height h e, over an field radiated by an infinitesimal vertical electric dipole located on imperfectly conducting half-space. In this section, we have the surface of the planar Earth. Then, Norton introduced the summarized these theories with a standardized notation system. We -iωt attenuation function, the ground effect, and the frequency dependence have employed a harmonic time factor e throughout. The of the surface wave radiated by a vertical dipole. Based on Maxwell’s infinitesimal dipole is fed by a unit electric moment Idl=1 A.m equations, R.W.P. King established the EM field expression (current I, infinitesimal length dl ). generated by a vertical electric dipole located on or in the vicinity of Fig. 2 describes the set of coordinates and the geometry parameters. the surface of a planar Earth and described the surface-wave Since the propagation characteristics are dependent on ground propagation behavior. King also introduced characteristic distances to properties, we use the wave numbers k0 and kg, respectively in the air explain the attenuation factor variation along the path. In order to and in the ground, where εrg and σg are respectively the relative model mixed-path surface wave propagation effects, Millington [9], permittivity and the conductivity of the ground. These media are [10] developed an analytical method which takes into account the assumed to have the same permeability µ 0 as that of free space. ground characteristics changes along the path. Recently, L. Sevgi [11]-[13] has developed significant contributions which integrate surface-waves propagation along mixed-path. All these studies are ( 1 ) mainly theoretical and the measurements are unusual. ( 2 )

( 3 )

Where ε is the complex refractive index of the ground. c This work was supported in part by TDF and the “Direction Générale de l’Armement”. A. Norton’s Model M. Bellec, S. Avrillon and F. Colombel are with the institute of Electronics and Telecommunication of Rennes (IETR), UMR CNRS 6164, University of Rennes 1, Campus de Beaulieu, Rennes Cedex 35042, France. The EM field radiated by an infinitesimal vertical electric dipole on (e-mail: [email protected]; [email protected]; the surface of the planar Earth was firstly analyzed by A. Sommerfeld [email protected]) [1]. Then, K.A. Norton simplified the calculation, the formalism and S. Palud and P-Y Jezequel are with TDF, La Haute the interpretation [2], [3] by introducing the attenuation function. Galesnais, Centre Mesure d'Antennes, 35340 Liffré, France. (e-mail : [email protected]; [email protected]) Norton’s formalism starts from the Hertz vector and Norton’s model Ph. Pouliguen is with the « Direction Générale de l’armement » (DGA), is valid for any transmitter or receiver height (h e or h r). The Hertz DGA-DS/MRIS, 7-9, rue des Mathurins, Bagneux Cedex 92221, France vector expression Πz contains 3 terms: a direct wave, a reflected wave (e-mail : philippe.pouliguen@ intradef.gouv.fr) and the surface wave. When the transmitter and the receiver are both located on the ground (h e=h r=0 m), the direct and reflected

Norton’s model King’s model

Fig. 2. Set of coordinates and location of an infinitesimal vertical electric dipole (D) at a height he in the air (wave number k 0) over the ground (wave number kg). components cancel each other out, and only the surface wave is propagated (third component in (4)). ( 10 )

( 4 )

Where Z0=120 π is the free space impedance, Rv the Fresnel’s reflection coefficient and F the attenuation function of the surface wave. Rv and F(p 0) are defined with the following formula:

Fig. 1. Path loss of the field radiated by a vertically polarized dipole at 10 MHz, 20 MHz, and 30 MHz over the sea water (SW) and over a dry ground (DG) versus distance [11]. Based on Norton’s theory, we have also investigated the ground type and the frequency dependence of surface waves. Fig. 1 depicts the path loss versus the distance for sea water ( εrg =80 and σg=4 S/m ), and dry ground ( εrg =8 and σg=0.04 S/m ) [5], each at three frequencies (10 MHz, 20 MHz, and 30 MHz). These results were calculated with L. Sevgi’s tool [11]. Whatever the frequency, we notice that the path ( 5 ) losses over the sea are lower than over dry ground. As a result, at 30 MHz and for a distance of 50 km, the path loss over dry ground is 44 dB higher than over sea water. Likewise, over any grounds, the 1- path losses of the surface wave become more important when the frequency increases. As a result, over a dry ground at 50 km, the path j ( 6 ) loss at 30 MHz is 30 dB higher than at 10 MHz. B. King’s Model ( 7 )

R.W.P. King established the EM field expression generated by a Where p is the Sommerfeld numerical distance. 0 vertical electric dipole located on or in the vicinity of Earth surface

from Maxwell’s equations. According to [6], the transverse magnetic The parameters b and p are respectively a numerical distance and a g g induction B and associated electric field E are governed by the numerical velocity: φ formulas:

( 8 ) ( 11 )

( 9 )

Where X is the loss tangent of the ground. ( 12 )

The relation between the magnetic field and the Hertz vector is:

( 13 )

Earth boundary. As well as the antenna dimensions, the electrical 3 2 Where P0=k 0 d/2k g is the Sommerfeld numerical distance and F(P 0) horizon depends on the wavelength. Fig. 3 sketches the propagation is defined by the following formula: behavior and we can distinguish two main areas:

• Up to the intermediate distance di, the EM field decreases as 1/d. • ( 14 ) Starting from di, the EM field decreases as 1/d².

Where, C2(P 0)+iS 2(P 0) is the Fresnel integral and ω is the pulsation. Physically, the transition between the two areas is smooth. We call this transition area the “smoothly” attenuation. As a result, Fig. 3 These formulas have been proposed with the following conditions sketches three cases of surface-wave propagation: issued from [4] and [6] respectively: • At 10 MHz over sea water, the intermediate distance di is close to dc. Consequently, the 1/d² field strength attenuation is not achieved, but the “smoothly” attenuation is expected. ( 15 ) • At 20 MHz over sea water, the intermediate distance di is equal to 17 km and the critical distance dc is equal to ( 16 ) 58 km. Consequently, the three behaviors (1/d, smoothly, and 1/d² attenuation) are expected. Where r 0 is defined in Fig. 2. • At 30 MHz over sea water, the intermediate distance di is equal to 8 km and the critical distance dc is equal to 50 km. Consequently, the three behaviors (1/d, smoothly, and 1/d² attenuation) are expected. In table I, we calculate |k g|/k 0 versus ground characteristics and frequencies. As a result, we notice that the condition | kg|≥3k 0 is verified in the HF band. But over a dry ground (poor conductivity and relative permittivity) in the UHF band, the initial condition (15) TABLE II. CHARACTERISTIC DISTANCES D I AND (DC) VERSUS SEVERAL is not verified. GROUND TYPES AND FREQUENCIES .

di TABLE I. CALCULATION OF |K |/ K VERSUS GROUND CHARACTERISTICS G 0 (d ) 10 MHz 20 MHz 30 MHz (SEA WATER AND DRY GROUND ) AND FREQUENCIES IN ORDER TO VERIFY THE c

CONDITION |KG|≥3K Sea 0 69 km 17 km 7.6 km ε =80 ; rg (73 km) (58 km) (50 km) σ =4 S/m |k |/k 10 MHz 20 MHz 30 MHz g g 0 Dry Ground 690 m 176 m 80 m Sea Water εrg =8 ; σ (73 km) (58 km) (50 km) εrg =80 ; 85 27 10 g=0.04 S/m σg=4 S/m Dry Ground εrg =8 ; 8 3.3 2.83 σg=0.04 S/m

King’s theory describes physically surface-wave propagation behavior by defining characteristic distances: the critical distance dc, and the intermediate distance di. The critical distance dc represents the boundary of the planar Earth model and the intermediate distance di is a baseline which defines the modification of attenuation law of the surface waves.

( 17 )

( 18 )

Where a is the Earth radius.

Table II contains several characteristic distances d c and d i depending on the environment and frequency. The critical distance dc varies Fig. 3. Behavior of the surface wave propagation along a path: according to only with frequency, while the intermediate distance di varies both the frequency and the ground type, the surface wave attenuation decreases as with frequency and the ground characteristics. 1/d to reach smoothly 1/d². (a) – At 10 MHz over a sea water path, the surface There are two notions of horizon. First, the optical horizon is close to field strength attenuation decreases as 1/d up to di=69 km ~ d c. (b) – At 20 MHz over a sea water path, the surface field strength attenuation decreases 37 km at sea level. Secondly, in the field of surface-wave as 1/d over up to d i=17 km, then decreases as 1/d² up to d c=58 km. (c) – At propagation, the radio-electrical horizon dc is the electrical planar

30 MHz over a sea water path, the surface field strength attenuation decreases as 1/d over up to d i=8 km, then decreases as 1/d² up to d c=50 km. The experimentation took place at the Mediterranean Sea in June 2013, at distances varying from the transmitting antennas. Fig. 5 depicts the environmental topography of the measurement area. The C. Millington’s Model attenuation of the electric-field strength versus distance from the transmitting antennas was measured at 10, 20, and 30 MHz. Each The Millington’s model is applied as soon as the environment measurement has been geo-localized and stored with an acquisition contains several ground types along the propagation path. A simple software developed by TDF. HF antennas installed on the coast (T x) case is sketched in Fig. 4 with two transitions through three media. have the capability to emit HF signal at each chosen frequency (10, 20, and 30 MHz) and the received signals are carried out with a loop antenna installed on a boat (R x). The boat followed a southwestward path across the Mediterranean Sea, steering a constant course.

Fig. 4. Sketch of the Millington’s model: The transmitting antenna (T x) is The three transmitting antennas operating respectively at 10 MHz, located over the medium 1 while the receiver antenna (R x) is located over the 20 MHz, and 30 MHz are located over salt ponds. A sand zone is medium 3. The medium 1 represents a segment range of length d 1, the located between the sea water and the transmitters. The relative medium 2 represents a segment range of length d and the medium 3 2 permittivity εrg1 of the salt ponds is 80, and the conductivity σg1 is represents a segment of a variable length d. 8.8 S/m. For the sand transition, ε =8 and σ =0.038 S/m . For the rg2 g2 sea water, εrg2 =80 and σg2 =5 S/m . The path over the salt ponds and According to Fig. 4, the semi empirical method can be explained with the sand is respectively 1 km and 6 km. These conductivities have the following equations (19)-(21) coming from [11] and [14]: been measured thanks to the following commercial devices:

The total field ET along a multi-mixed propagation path at the • HANNA HI 993310 with HI 76305 probe for ground receiver is defined by: • HANNA HI 9033 with HI 76302 probe for liquid

The measurements have been carried out in Continuous Waves (CW). No sky waves could be received, due to: ( 19 )

Where ED and ER are respectively the fields along the direct and • A low Sunspot Number of 52.5 observed for June 2013 reverse paths: • A maximal path of 50 km

• A monopole shape radiation pattern for the transmitting antennas ( 20 ) This medium characteristic change allows observing the Millington’s ( 21 ) effect.

Where E , E and E are respectively the field over the medium 1, the 1 2 3 • At 10 MHz, we measure the electric field strength from medium 2 and the medium 3. 9 km until 35 km. Consequently, the “smoothly”

attenuation can be observed. At this frequency, the The Millington’s method shows that the EM surface wave field Millington’s effect is notable because of the sand transition. strength is subject to the medium change. According to the • modification of electrical ground characteristics, sea-ground or At 20 MHz, we measure the electric field strength from ground-sea, the EM field could respectively increase or decrease 6.5 km until 60 km. Consequently, the 1/d² attenuation can from each transition. be observed. At this frequency, the Millington’s effect is more significant. III. MEASUREMENT SETUP • At 30 MHz, we measure the electric field strength from 6.5 km until 50 km. Consequently, the 1/d² attenuation can 1) Global description be observed. At this frequency, the Millington’s effect

increases. This section presents the objectives of the measurements and describes the setup used to measure the attenuation of the electric field over the sea. The goals of our measurements are:

• To validate the 1/d and 1/d² attenuation of the surface wave propagation in the HF band calculated with the planar model. • To validate Millington’s model.

In order to check the different attenuation behaviors over the sea, we have selected three frequencies (10 MHz, 20 MHz, and 30 MHz). Fig. 5. The measurement path over the sea by boat from the vicinity of the The sea has been constantly calm (Sea State 0) all over the transmitter (Tx) until the receiver (Rx). experimentation, so no sea roughness parameter has been considered in the theories. 2) Antennas used for the measurements

field used to normalize the theoretical models with the transmitting The measurement setup used 2 types of antennas (see Fig. 6 and Fig. antenna characteristics. Thus, the reference field E(d 0=1 km) at 7). The transmitting antenna, a patented surface-wave antenna, called 10 MHz, 20 MHz, and 30 MHz is respectively 90.48 dBµV/m, DAR antenna [19], has been manufactured by TDF (see Fig. 6). The 92.79 dBµV/m, and 88.24 dBµV/m. antenna is manufactured with a steel galvanized wire with a diameter of 2.7 mm. Its dimensions have been adjusted according to the TABLE V. MEASURED ANTENNA GAINS (DBI) AND THE ASSOCIATED selected frequency. The horizontal radiation pattern is ELECTRIC FIELD STRENGTH AT A DISTANCE D 0=1 KM . omnidirectional. Table III summarizes the horizontal length L , the t Antenna Gain Input power E(d ) vertical height h, and the gap Z of the DAR-antennas for each f (MHz) 0 e (dBi) (dBm) (dBµV/m) frequency. The receiving antenna, a loop (see Fig. 7) installed on a 3.3 45 90.48 boat, operates across a broad frequency band. The table IV presents 10 the performance of the loop at 10, 20 and 30 MHz. The K-Factor is 20 5 45 92.79 inversely proportional to the gain. So, the lower the K-factor, the 30 -1.3 47 88.24 higher the efficiency. The receiver antenna is installed 1 meter above the surface of the sea. A Rohde & Schwarz EB200 is used as the receiver. IV. COMPARISON BETWEEN MEASUREMENTS AND THEORETICAL CALCULATION

1) The “smoothly” attenuation at 10 MHz

Fig. 8 presents the theoretical and the measured electric field along a 35 km mixed-path (salt ponds, sand, and sea water) at 10 MHz. 46 points per kilometers have been recorded. The theoretical models (King and Norton) and the measurement results are in good agreement besides a maximal deviation of 1.13 dB. The EM field attenuation slope is between the 1/d and 1/d² slopes. This behavior

Fig. 6. DAR-antenna design and tuning parameters: horizontal corresponds to the “smoothly” attenuation because the maximum measured distance (35 km) is lower than King’s intermediate distance length (L t) and vertical height (h). di which is equal to 69 km. The decrease of electric field level is due to the sand transition because the sand is a medium which is not suitable to the HF propagation.

2) The 1/d² attenuation and Millington’s effect at 20 MHz and 30 MHz

Fig. 9 presents the theoretical and the measured electric field along a 60 km mixed-path (salt ponds, sand, and sea water) at 20 MHz. 40 points per kilometers have been recorded. The theoretical models (King and Norton) and measurement results are in good agreement. The discrepancies between theories and measurements are less than Fig. 7. Receiver-antenna design: copper tube (16/14 mm) loop of 420 mm 2 dB between 6 km and 10 km. From 30 km, the 1/d² attenuation is diameter observed.

TABLE III. DIMENSIONS OF THE DAR-ANTENNAS .

10 MHz 20 MHz 30 MHz

Lt 6.37 m 2.52 m 1.83 m h 1.8 m 1.8 m 1.5 m

Ze 0.5 m 0.5 m 0.5 m

TABLE IV. PERFORMANCES OF LOOP ANTENNA VERSUS FREQUENCY .

10 MHz 20 MHz 30 MHz K – Factor 42 41 43 (dB/m)

Fig. 8. Theoretical and measured electric field attenuation at 10 MHz over a The antenna gain has been measured in the relevant azimuth in order mixed path: salt ponds, sand, and sea water. The theoretical results are to calculate the electric field strength. The antenna gains at 10 MHz, calculated from Norton’s and King’s models. The red curve represents the 20 MHz, and 30 MHz are respectively 3.3 dBi, 5 dBi, and -1.3 dBi measurements over the sea water. (see table V). The electric field strength E(d 0) at 1 km is the reference

we have described briefly the theoretical results proposed by Norton and King. Then, we have presented measurement results at 3 frequencies (10 MHz, 20 MHz and 30 MHz) and compared it to the theories including Millington’s modification in order to take into account the interface between the coast and water sea. At 10 MHz, the “smoothly” attenuation is shown and is very well correlated with the theory. The EM field decrease as 1/d² has been clearly observed at 20 MHz and 30 MHz.

A future work is scheduled to measure the electric field strength of the surface wave at larger distances in order to consider the roundness of Earth.

AKNOWLEDGEMENTS

The Authors thank warmly J. Y. Laurent from TDF for his technical Fig. 9. Theoretical and measured electric field attenuation at 20 MHz over a support. They also thank TDF and Direction Générale de l’Armement mixed path: salt ponds, sand, and sea water. The theoretical results are for their funding. calculated from Norton’s and King’s models. The red curve represents the measurements over the sea water.

REFERENCES At 20 MHz, the effect of the sand transition is more significant than [1] A. Sommerfeld, “Propagation of waves in wireless telegraphy,” Ann. at 10 MHz. At the interface between the sand and the sea water Phys., vol 28, pp. 665-736, 1909 (6 km), we notice small perturbations on the measured electrical [2] K.A. Norton, “The propagation of radio waves over the surface of the fields. The sand transition includes a sand dune. This relief could earth and upper atmosphere - PART 1,” Proceeding of the institute of induce diffraction and it could be the reason of this perturbation. The radio engineers, 1936 phenomenon has not been taken into account theoretically; a bigger [3] K.A. Norton, “The propagation of radio waves over the surface of the change of the slope has been predicted. earth and upper atmosphere - PART 2,” Proceeding of the institute of radio engineers, 1937 Fig. 10 exhibits the theoretical and the measured electric fields along [4] R. W. P. King, “On the radiation efﬁciency and the electromagnetic ﬁeld a 50 km mixed-path (salt ponds and sea water) at 30 MHz. 46 points of a vertical electric dipole in the air above a dielectric or conducting half-space,” in Progress in Electromagnetic Research, J. A. Kong, Ed. per kilometers have been recorded. The theoretical models (King and New York: Elsevier, vol. 4, ch. 1, 1990 Norton) and the measurement results are in good agreement. The [5] R. W. P. King, S. S. Sandler, “The electromagnetic field of a vertical maximal deviation between theories and measurements is close to electric dipole over the Earth or sea,” Antennas and Propagation, IEEE 4 dB between 6 km and 10 km. Starting from 15 km, the 1/d² Transactions on , vol. 42, no. 3, pp. 382-389, Mar. 1994 attenuation behavior is observed. [6] R. W. P. King, C. W. Harrison, “Electromagnetic ground-wave field of vertical antennas for communication at 1 to 30 MHz,” Electromagnetic The perturbation, underlined at 20 MHz at the interface between the Compatibility, IEEE Transactions on , vol. 40, no. 4, pp. 337-342, Nov. sand and the sea water, is amplified at 30 MHz because the relief is 1998 higher related to the wavelength. [7] J. R. Wait, Electromagnetic Waves in Stratified Media, New York, Pergamon Press, first edition 1 962; second enlarged edition 1970; first edition reprinted by IEEE Press 1996 [8] J. R. Wait, “The ancient and modern history of EM ground-wave propagation,” Antennas and Propagation Magazine, IEEE , vol. 40, no. 5, pp. 7-24, Oct. 1998 [9] G. Millington, “Ground-wave propagation over an inhomogeneous smooth earth,” Proceedings of the IEE - Part III: Radio and Communication Engineering , vol. 96, no. 39, pp. 53-64, Jan. 1949 [10] G. Millington, G.A. Isted, “Ground-wave propagation over an inhomogeneous smooth earth. Part 2: Experimental evidence and practical implications,” Electrical Engineers, Journal of the Institution of , vol. 1950, no. 7, pp. 190-191, July 1950 [11] L. Sevgi, “A mixed-path groundwave field-strength prediction virtual tool for digital radio broadcast systems in medium and short wave bands,” Antennas and Propagation Magazine, IEEE , vol. 48, no. 4, pp. 19-27, 4, Aug. 2006

[12] L. Sevgi, F. Akleman; L. B. Felsen, “Groundwave propagation Fig. 10. Theoretical and measured electric field attenuation at 30 MHz over a modeling: problem-matched analytical formulations and direct mixed path: salt ponds, sand, and sea water. The theoretical results are numerical techniques,” Antennas and Propagation Magazine, IEEE , calculated from Norton’s and King’s models. The red curve represents the vol. 44, no. 1, pp. 55-75, Feb. 2002 measurements over the sea water. [13] L. Sevgi, “Groundwave Modeling and Simulation Strategies and Path Loss Prediction Virtual Tools,” Antennas and Propagation, IEEE V. CONCLUSION Transactions on , vol. 55, no. 6, pp. 1591-1598, June 2007 In this paper, the measurements of surface waves propagating along [14] L. Boithias, "Propagation des ondes radioélectriques dans sea water path in the HF band are compared with theories. First of all, l’environnement terrestre", Dunod, 1983

[15] IUT-R, “Ground-wave propagation curves for frequencies between 10 [17] S. Rotheram, “Ground-wave propagation. Part 1: Theory for short kHz and 30 MHz,” http://www.itu.int/pub/R-REC/fr distances,” Communications, Radar and Signal Processing, IEE [16] IUT-R,“Calculation of free space attenuation,” http://www.itu.int/pub/R- Proceedings F , vol. 128, no. 5, pp. 275,284, Oct. 1981 REC/fr [18] http://www.ips.gov.au/Products_and_Services/1/4 [19] S. Palud, P. Piole, P.Y. Jezequel, J.Y. Laurent, L. Prioul, “ Large-area broadband surface-wave antenna,” Patent WO/2012/045847

Wireless Communications for marine sensor networks

J.C. Reyes-Guerrero1, M. Bruno Mejías2, A. Medouri3 and Luis A. Mariscal1 1 Dept. of Computer Languages and Systems, University of Cadiz, Nautical Science Faculty, Puerto Real (Cádiz), Spain, {josecarlos.reyes, luis.mariscal}@uca.es 2 Dept. of Applied Physics, University of Cadiz, Marine Science and Environmet Faculty, Puerto Real (Cádiz), Spain, [email protected] 3 Dept. of Statistics and Informatics, University of Abdelmalek Essaâdi, Polydisciplinary Faculty, Route de Martil, Tetuan, Morocco, [email protected]

Abstract – Current marine wireless communication systems used for monitoring tion systems) and the large size and weight of antennas and hardware transceiv- applications based on buoys suffer from lots of weakness. Our research project con- ers (VHF systems). These limitations have motivated a new research activity. The cerns the design and development of new technological applications to improve general goal is to design and develop a novel broadband wireless communica- marine communications. Particularly, a novel wireless sensor network based on tion system to perform applications like the one mentioned above. WiMAX standard operating at the 5.8 GHz band (license-exempt band) is proposed. A wireless sensor network based on WiMAX standard ([2], [3]) could be a good As an initial task, a propagation channel measurement campaign in maritime en- candidate to accomplish this task. WiMAX is an evolving technology that is op- vironments was carried out to investigate the impact of the wireless channel in dif- timized for operating on land environments where its good performance has ferent situations. This work provides radio measurements over sea around urban been extensively demonstrated. Several frequency bands can be used for de- environments. In particular, a radio link between a buoy and a ship at 5.8 GHz is ploying this system. The license-exempt 5 GHz band is of interest to WiMAX studied. LOS (Line-Of-Sight) and NLOS (Non-Line-Of-Sight) paths are investigated. because this is generally available worldwide and it is free for anyone to use, The designed measurement system is described and the experimental measure- i.e. it could enable deployments in underserved markets, like the maritime one. ments are shown. This investigation is useful, among others, for planning World- In particular it is the upper 5.725 GHz-5.850 GHz band that is most attractive wide Interoperability for Microwave Access (WiMAX) networks offshore around due to the fact that many countries allow higher power output compared to these challenge environments. other bands. This facilitates less costly deployments. Regarding range and peak Keywords - propagation channel measurements, WiMAX, maritime environment, data rates, field tests on land have shown tens of kilometers and Mbps, respec- wireless sensor networks. tively. These potential characteristics overcome the weakness described above. However, the performance of WiMAX networks in marine environments is not EXTENDED ABSTRACT. optimum due to the different radio propagation conditions. The main goal of a Recently, many studies have identified an emerging demand for telecommuni- research project between the universities of Cadiz and Abdelmalek Essaâdi is to cation services in several applications over sea. Some of them are getting great optimize the WiMAX standard for maritime applications. interest for the scientific community, e.g. those related to real-time monitoring An initial and crucial task for the optimization of this standard over sea is to through sensing multiple physical parameters from the sea. Although the num- study the radio propagation channel in these scenarios in the 5 GHz band. Par- ber and kind of parameters depend on the specific application, monitoring sys- ticularly, buoy-to-ship propagation measurements were performed over sea. tems are quite similar. Basically, these systems are based on a set of buoys and Fig. 1 shows the routes where the measurements were carried out. each one is equipped with two main subsystems. Firstly, a subsystem including Propagation models and measurements for land, both large-scale path loss and a lot of sensor devices that measure locally the data. Secondly, a radio system small scale multipath, have been discussed extensively. Further works in this which is in charge of transmitting them to a central base station for process- field have been done in urban and suburban environments. Besides, although ing and monitoring purposes. The base station could be installed on shore or some works present experimental measurements of propagation characteristics aboard a ship. This last case is particularly interesting for some applications, e.g. for maritime radio links, they do not apply to conditions covered by our study. those related to oceanography campaigns. To the best of the authors’ knowledge, buoy-to-ship links characteristics over Current wireless technologies used in this kind of applications are mainly based sea at 5.8 GHz have not been investigated. In this work, we focus on LOS (Line- on VHF, cellular mobile telecommunication systems (GSM, UMTS, etc.) and sat- Of-Sight) and NLOS (Non-Line-Of-Sight) paths. We discuss them analyzing the ellite communications systems (INMARSAT, VSAT, etc.). However, these systems measurements performed in a real marine scenario. This work is helpful, among suffer from lots of weakness [1], like low bandwidth or capacity (GSM, Satellite others, to deploy WiMAX systems in these challenges scenarios. and VHF systems), short range (cellular mobile telecommunication systems), high cost for certain applications (satellite and cellular mobile telecommunica-

Fig. 1. Route followed by the ship (red) and fixed location of the buoy.

Instrumentation Viewpoint/11/ MARTECH11 44 Rep. ITU-R M.2123 1

REPORT ITU-R M.2123*

Long range detection of automatic identification system (AIS) messages under various tropospheric propagation conditions

(2007)

1 Introduction In the 1990s, the International Maritime Organization (IMO), the International Telecommunications Union (ITU) and the International Electrotechnical Commission (IEC) adopted a new navigation aid known as the Automatic Identification System (AIS) to help improve safety-of-navigation, maritime traffic control and efficiency of maritime commerce. The primary purpose of the AIS is to facilitate the efficient exchange of navigation and voyage data between ships, and between ships and shore stations. The technical characteristics of the AIS using time division multiple access (TDMA) techniques in the VHF maritime mobile band are described in Recommendation ITU-R M.1371. Like most VHF terrestrial systems, the maximum range of AIS communications is normally governed by line-of-sight and diffraction mode propagation mechanisms. Assuming typical technical parameters of AIS equipment, maximum reliable ship-to-ship radiocommunications over sea water is in the range of 20-25 NM. Shore stations, with high antennas, can reliably receive AIS messages from ships at distances of up to 20 to 35 NM, depending on antenna heights above sea level. The safety-of-navigation and traffic control functions provided by the AIS dictate a requirement for high communications reliability in which a high percentage of the AIS messages are detected and corrected decoded. Because of the continued growing importance of the AIS traffic, a need has arisen to monitor shipping at distances from shore greater than can be achieved via these conventional propagation mechanisms. Recommendation ITU-R M.1371 introduces the concept of long range detection of AIS data but does not define a specific communications mechanism to accomplish long range AIS detection. As contrasted with normal AIS safety-of-navigation functions, this long range AIS capability does not necessarily need the same high degree of communications reliability. This lower reliability requirement follows from the fact that it is necessary to only detect a fraction of the AIS messages sent from a given ship to accomplish the goal of updating ship locations on a regular basis. This Report addresses long range monitoring of ship locations at sea through detection of AIS messages. Section 2 focuses on concepts that may enhance long range detection of ships at sea by AIS coast stations. Section 3 assesses whether long range detection has similar spectrum sharing characteristics with other co-channel mobile systems as that of normal AIS safety-of-navigation functions. Section 4 provides an overall summary.

* This Report should be brought to the attention of Radiocommunication Study Group 3, the International Maritime Organization (IMO), the International Association of Marine Aids to Navigation and Lighthouse Authorities (IALA) and the Comité International Radio Maritime (CIRM). 2 Rep. ITU-R M.2123

2 Concepts to enhance the long range detection capabilities of AIS coast stations

2.1 Off-shore platforms One mechanism to effectively supplement the shore-based reception range of AIS ship messages is through the installation of AIS receivers on off-shore platforms. Two cases are considered herein – off-shore oil platforms and off-shore weather buoys.

2.2 Oil platforms In some parts of the world, off-shore oil platforms are extensively deployed along coastal areas. These platforms are typically very large, as illustrated in Fig. 1, and have extensive radiocommunications capability with the nearby mainland. Consequently, in such areas, oil platforms can provide a very useful base on which to locate AIS receiving equipment to enhance shore-based coverage.

FIGURE 1 Typical offshore oil platform

The first step in evaluating the range enhancements possible with the use of offshore oil platforms is the development of a baseline capability for conventional shore-based reception. Table 1 gives representative technical parameters applicable for the AIS ship to shore link. Using these parameters and a radio propagation model as described in Appendix 3, Fig. 2 was developed which shows the median predicted received power for AIS ship to shore radiocommunications as a function of shore station antenna height. The results from Fig. 2 indicate a median radiocommunication range of about 38 to 60 km (20 to 32 NM) for the antenna heights shown. Offshore oil platforms have tower heights considerable higher than the coast station heights used above and consequently have larger communications range with ships at sea. Rep. ITU-R M.2123 3

TABLE 1 Representative technical parameters for AIS ship-to-shore communications

Technical parameter Typical value Ship Transmit power 41 dBm Antenna line losses 3 dB Antenna gain 2 dB Antenna height ~10 M Shore station Antenna gain 5 dBi (omnidirectional) Antenna line losses 3 dB Receiver sensitivity –107 dBm for 20% per (minimum) –109 dBm for 20% per (typical) Median received signal level from edge of service –104 dBm range Antenna height 10 to 50 m

FIGURE 2 Predicted AIS median received power at shore station

4 Rep. ITU-R M.2123

In those cases where there are sufficient numbers of such oil platforms, a line of AIS-equipped oil platforms appropriately located surrounding a key port can provide reliable AIS reception coverage at up to triple the distance from coast stations alone. Because of the extensive communications equipment already installed on most oil platforms, backhaul of the received AIS data to shore, either via satellite or undersea cable, generally presents no problems. One such location is the Gulf of Mexico on the Southern Border of the United States of America. Figure 3 gives an overview of the oil platforms deployed in this area. Also shown as dark circles on the figure is one possible arrangement of AIS-equipped platforms. This arrangement of platforms extends reliable AIS reception coverage for ships to at least 120 NM from two major port areas, Houston, Texas and New Orleans, Louisiana in the United States of America.

FIGURE 3 Example of off shore oil platform deployment near New Orleans LA, United States of America

The communication range enhancement as a result of anomalous propagation modes, primarily atmospheric ducting, discussed later in this section would also be applicable for AIS receivers installed on off-shore platforms. Consequently, on an intermittent basis, AIS reception coverage would extend further out to sea. For the Gulf of Mexico example described here, climatic conditions are often conducive to atmospheric ducting. It is concluded that in areas where off-shore oil drilling occurs, installation of AIS receivers on these platforms can increase the coverage range of shore-based AIS detection by a factor of up to three. A line of well-placed platforms can provide complete extended-range coverage around key port facilities.

2.3 Weather buoys As in the case of oil platforms, off-shore weather buoys are also used along coastal areas. Figure 4 illustrates a typical 10 m buoy. While the density of deployment is typically much lower than the example described above for oil platforms, their use is much more widespread in many coastal areas. Rep. ITU-R M.2123 5

FIGURE 4 Typical offshore weather buoy

These platforms offer an additional opportunity for the installation of AIS receivers to supplement shore-based AIS detection. However, use of off-shore weather buoys presents several additional challenges. The much smaller size of the weather buoys significantly limits the possible height for placement of the antennas; consequently the detection range is smaller than normal shore based reception. Table 2 presents representative technical parameters for a ship-to-weather buoy link.

TABLE 2 Representative link parameters for AIS weather buoy communications

Technical parameter Typical value Ship parameters Transmit power 41 dBm Antenna line losses 3 dB Antenna gain 2 dB Antenna height 10 m Weather buoy Antenna gain 2 dBi Antenna line losses 1 dB Antenna height ~3 m Median signal level from edge of service area –104 dBm Receiver sensitivity –107 dBm for 20% per (minimum) –109 dBm for 20% per (typical)

Using analysis methods as described earlier, the reliable communication range of weather buoy AIS detection is estimated to be approximately 33 km (18 NM). 6 Rep. ITU-R M.2123

FIGURE 5 AIS communication range from typical weather buoys

Figure 6 shows one example of the distribution of off-shore weather buoys along the East Coast of the United States of America. As contrasted with the use of oil platforms, the lower density and shorter communication range of AIS-equipped weather buoys typically could not provide enhanced AIS coverage in all directions from a key port area. For this example, installation of AIS receivers on several weather buoys could provide limited enhanced AIS coverage along major shipping lanes approaching the port of New York but not full umbrella coverage.

FIGURE 6 Example of off shore weather buoy deployment

Rep. ITU-R M.2123 7

One significant limitation on the use of weather buoys to enhance AIS range is the mechanism to backhaul the data to shore. While the use of undersea cables is impractical, communication to shore via satellite communications is possible. However, the reliance on solar energy as the source of power limits satellite communications to intermittent, low data rate communications. This limitation, in turn, would require the use of on-board signal processing with the AIS receiver so that only significant new or updated ship information would be sent and all unnecessary repetitive AIS data eliminated. One option to add the backhaul AIS data to the existing weather data channel via LEO and geostationary earth orbiting (GEO) weather satellite links was investigated. The weather data transmitters installed on the weather buoys typically operate on an intermittent, low-duty-cycle, basis at a data rate of 300 to 1 200 bit/s on a channel shared by many other weather buoys. The sharing of a single uplink channel among many buoys generally limits the transmissions from a given buoy to about once per hour. In order to not significantly impact the primary weather functions, any added AIS data must be limited to some small fraction of the existing weather traffic. With these combined limitations, the use of the existing satellite weather communication channel for AIS derived data on an operational basis appears impractical. The alternative is to include a separate transmitter on the weather buoy capable of communicating with an existing LEO satellite communications network. The communications requirements of such a link can be estimated as follows: a) Assume that the on-board AIS signal processing limits the data forwarded via the satellite link to only a single message when a ship enters its AIS communications zone and one message when a ship leaves the zone. b) Assume that the average ship traffic along the associated shipping lane is X inbound ships and Y outbound ships per hour. Under these assumptions, the net number of AIS messages forwarded via the satellite link would then be 2·X + 2·Y per hour. Under any reasonable estimate of ship traffic, the transmitted AIS message rate via the satellite link would be quite low. In summary, the installation of AIS receivers on off-shore weather buoys can provide useful extended range coverage along key shipping routes, although full umbrella coverage will typically not be possible.

2.4 Anomalous propagation Another concept to take advantage of the lower reliability requirement for long range AIS detection is to supplement normal coverage via line-of-sight and diffraction propagation modes by placing additional reliance on anomalous propagation mechanisms. Several ITU-R Recommendations address the characteristics of a number of these mechanisms including: Tropospheric scatter Atmospheric ducting Meteor burst Rain scatter Sporadic-E Two of these mechanisms, tropospheric scatter and atmospheric ducting, are investigated in the following paragraphs. 2.4.1 Tropospheric scatter Tropospheric scatter (hereinafter called troposcatter) is a mode of transhorizon radio wave propagation that results from the random reflections and scattering from irregularities in the dielectric gradient density of the troposphere. This propagation mode is applicable from below 100 MHz to above 8 000 MHz and may extend for distances of several hundred kilometres. 8 Rep. ITU-R M.2123

Although it is included herein under the general category of anomalous propagation, the propagation loss resulting from this effect, although quite large, can be sufficiently steady and predictable to support reliable long range communications. Because of the typically large propagation losses involved, it is clear at the outset of this study that reliable reception of AIS messages from ships at sea via troposcatter propagation will not be possible using current AIS coast station designs. The discussion herein focuses on design factors that may allow reliance on tropospheric scatter for long range AIS detection. Since the characteristics of ship-borne AIS equipment is fixed and cannot be modified in the near term, the key factors are the propagation loss, coast station receiver sensitivity, receiver antenna gain, receiver signal processing, and radio noise. a) Propagation loss Appendix 1 describes in detail two ITU-R Recommendations that address the characteristics of troposcatter propagation. Recommendation ITU-R P.617 addresses propagation loss from the standpoint of the design of transhorizon communications systems in the frequency range 200 MHz to 5 GHz. Recommendation ITU-R P.452 addresses the evaluation of interference via various propagation mechanisms, including tropospheric scatter, for the frequency range 100 MHz to 50 GHz.1 Although the latter recommendation is not directly applicable to the present subject, consideration of the recommendation is useful here in that it demonstrates that propagation trends continue smoothly, free from unexpected results, as low as 100 MHz. A third propagation model was examined which similarly showed consistent trends as low as 20 MHz. Consequently, for purposes of this report, the propagation methodologies described in Recommendation ITU-R P.617 are assumed applicable when extrapolated to 162 MHz. Figure 7, drawn from Appendix 1, describes troposcatter loss as a function of distance and reliability statistics. The curves were developed based on a ship height of 10 m and a receiving antenna height above average terrain of 50 m. However, since troposcatter propagation losses over water are relatively insensitive to antenna heights, the curves will be generally applicable at most practical shore station heights.

FIGURE 7 Tropospheric propagation loss at 162 MHz

1 The subject recommendation is focused on interference considerations above 700 MHz. However, the recommendation states that the propagation calculation methods described therein are believed to be reliable at frequencies down to 100 MHz, except for the ducting model. Rep. ITU-R M.2123 9 b) Receiver sensitivity The specifications in Recommendation ITU-R M.1371 for receiver sensitivity of ship-borne AIS equipment call for a sensitivity of –107 dBm for a 20% PER or better. Commercial equipment currently on the market typically have a sensitivity of –109 dBm. To successfully communicate via troposcatter, custom designed receivers would be necessary having a significantly higher sensitivity. Textbooks describing optimum receiver performance of differential GMSK modulation indicate that performance at a carrier-to-noise ratio of 13 dB for a bit error ratio (BER) of 10–5 is possible, including reasonable implementation losses. This BER approximately corresponds to a PER of 1%. With an assumed receiver noise figure of 3 dB and optimum receiver design, a sensitivity for reception of 9 600 bit/s GMSK modulation of –118 dBm for a 1% PER appears feasible. Relaxing the criteria to 20% PER would further increase sensitivity to approximately –120 dBm. c) Receiving antenna Similar to the above discussion, current AIS antenna designs with gains of 2 to 5 dBi would be impractical for effective AIS reception via troposcatter. Antennas using a four element collinear vertical array are commercially available for operation at 162 MHz with a peak gain of 8 dBi. The gain can be further increased using offset dipole elements resulting in a cardioid pattern with a peak gain of 11 dBi. Higher gain will likely require custom designs – for example constructing a horizontal array of 2 to 4 elements, where each element is as described above. Using this approach, an antenna main beam gain of up to 17 dBi may be feasible. d) Receiver correlation processing A much more uncertain improvement factor would take advantage of the repetitive nature of AIS messages. For example, during a 10 min period, a given ship will transmit about 86 AIS messages. During this period, approximately 60% of the bits in each of these AIS ship messages are repeated identically. The MMSI ship identification code is, in particular, repeated with each message. Given the moderately low data rates of AIS transmissions, use of parallel correlation processing techniques may permit an effective increase in receiver sensitivity in a manner somewhat analogous to spread spectrum. Further study would be required to determine the degree of correlation gain, if any, that may be achievable using this technique. For purposes herein, correlation gains of 0, 5 and 10 dB are considered. e) Radio noise environment In the VHF band, the environment radio noise can be a significant factor that limits the achievable receiver sensitivity. This would be especially true for the very high receiver sensitivity that would be required for AIS reception via troposcatter propagation. The ambient noise environment results from a combination of natural and manmade sources. At 162 MHz, galactic noise is the predominate natural radio source. Manmade sources of radio noise include vehicular ignitions, power lines and industrial machinery. Recommendation ITU-R P.372 presents typical values of ambient radio noise for galactic radio noise and three generic environments; business, suburban, and rural. The noise density values as a function of frequency are described relative to –204 dB(W/Hz) based on an ambient temperature of 290 K. For 162 MHz, the values are as follows: Quiet rural ~0 dB (limited by galactic noise) Rural + 2 dB Residential +5 dB Business +12 dB. 10 Rep. ITU-R M.2123

To realize the very high receiver sensitivity values described earlier, it is clear that the location of the receiving antenna in a very quiet rural environment is necessary to avoid degrading the achievable sensitivity. f) Results Combining the factors described above, a simple link calculation shows the potentially achievable AIS detection range via troposcatter propagation as follows:

Prec = EIRP − Lp + Gr + Gcorr − Lmisc > Sens where:

Prec: received power (dBm) EIRP: ship-borne AIS equivalent isotropic radiated power (dBm)

Lp: troposcatter propagation loss (dB)

Gr: receiver antenna gain (dBi)

Gcorr: receiver correction gain (dB)

Lmisc: miscellaneous cable losses (assumed to be ≤ 1 dB) Sens: receiver sensitivity (dBm). Rearranging terms yields:

Lp < EIRP + Gr + Gcorr – Lmisc – Sens Substitution of values for various assumed parameters yields the results shown in Table 3.

TABLE 3 Calculated AIS detection range via troposcatter (Based on EIRP = 40 dBm; Sensitivity = –120 dBm; 90% Reliability))

Antenna gain Correlation gain Maximum propagation Detection distance (dBi) (dB) loss (km) (dB) 11 0 170 93 11 5 175 180 11 10 180 157 17 0 176 124 17 5 181 168 17 10 186 230

These results show that with optimized receiver and antenna design, enhanced detection range would be possible using the troposcatter mode of propagation to distances on the order of 100 to 200 km. 2.4.2 Atmospheric ducting Atmospheric ducting is a propagation phenomenon that occurs predominately when the atmosphere stratifies into layers of differing temperatures and humidity. This layering results in varying index of refraction at different heights above ground. This situation, in turn, can lead to a condition where radio waves are effectively “trapped” between the layers somewhat similar to a waveguide. The trapping can occur between the Earth’s surface and a ducting layer, called a surface duct, or between two upper atmosphere ducting layers, called an elevated duct. Rep. ITU-R M.2123 11

The key features of this propagation mode are generally low propagation loss values but sporadic and intermittent occurrence. In some instances, the propagation loss values can approach free space values, but more typically is 5 to 30 dB higher than free space for distances in the range of about 50 to 200 km (see “Atmospheric ducting” in Appendix 5). For propagation over sea water, the rate of occurrence is very dependent on the general climatic and current weather conditions. Considerable diurnal and seasonal variations also exist. Conditions to the occurrence of atmospheric ducting may sometimes persist for hours or even days while at other times not occur for many consecutive days. On a long term average, conditions for atmospheric ducting occur less than 20% of the time and in most cases less than 5 to 10% of the time. Geographic areas where atmospheric ducting is most prevalent are humid climates and warm seas. One such area is the Gulf of Mexico and surrounding areas; conditions for atmospheric ducting with higher percentages have been recognized in other areas. Because of its intermittent nature, reliance on atmospheric ducting propagation conditions for long range AIS detection on a regular basis, such as every four hours, is generally not practical. However, the long range detection capability that does occur can be a useful supplement to normal shore-based AIS detection. Figure 8 is based on measurements in the Gulf of Mexico and describe the percent of time that AIS equipped ships were detected at an AIS receiver site on two days in April 2005. This represented normal non-ducting conditions. Figure 9 shows the same data collected on two other days when much greater detection distances were achieved from the same site. Only atmospheric ducting can account for these statistical results.

FIGURE 8 Measured AIS detection range under normal atmospheric conditions

12 Rep. ITU-R M.2123

FIGURE 9 Measured AIS detection range under atmospheric ducting conditions

These limited measurements show that AIS detection ranges can expand many-fold during periods of atmospheric ducting. As described in Appendix 2, Recommendation ITU-R P.452 examines the long term statistical characteristics of atmospheric ducting. However, as stated in the recommendation the prediction methods described have not been tested below the frequency of 700 MHz. Consequently, any extrapolation of the methods in that recommendation to 162 MHz must be viewed cautiously. Noting this limitation, statistic curves, shown here in Fig. 10, are developed in the Appendix in the same manner as done earlier for troposcatter propagation. As characterized in the Fig. 10 quite low values of propagation loss can occur over extended distances for very low percentages of time and conversely for high percentages of time, the predicted loss values are much higher – about 40 to 45 dB higher for 20% versus 1% occurrence. However, limited measurements suggest that in certain geographic areas, such as the Gulf of Mexico, the atmospheric ducting probability of occurrence predicted by Recommendation ITU-R P.452 may be lower than that actually experienced. As described earlier regarding troposcatter propagation, with significant upgrades to the AIS receiver and antenna, reliable communications may be achievable at distances exceeding 100 km. In contrast, with atmospheric ducting radio equipment upgrade will only provide marginally improved performance. In summary, atmospheric ducting propagation can provide useful enhanced AIS detection range, sometimes at distances of several hundred kilometres, on an intermittent basis but cannot approach the potential communication reliability described earlier for troposcatter propagation. Rep. ITU-R M.2123 13

FIGURE 10 Ducting propagation loss at 162 MHz using the ITU-R P.452 model (smooth terrain)

3 Spectrum sharing of AIS long range detection with other co-channel mobile service systems The two frequencies that have been designated as channels within the maritime mobile service for the terrestrial AIS function are not allocated on an exclusive basis. Rather, these channels and adjacent channels are allocated and used throughout various regions of the world for other mobile service applications including VHF public correspondence stations (VPCS) in the maritime mobile service and land mobile radio (LMR) systems. Long range AIS must be able to successfully operate in the interference environment resulting from existing services. Since long range detection of AIS is a new and evolving concept, it would be beneficial to examine the sharing of long range AIS operating with existing services, such as co-channel mobile systems and determine whether long range AIS presents sharing issues. Current spectrum sharing arrangements between existing AIS operation and co-channel mobiles systems have evolved and have been implemented in various ways by administrations. This section compares sharing between AIS operating in a long range detection mode (hereinafter long range AIS) and other co-channel mobile systems to that of normal AIS with the same mobile systems. Four steps are addressed in the following paragraphs: – characterization of technical parameters of systems involved, – designation of assumed sharing conditions – determining an appropriate radio propagation model, and – combination of these factors to determine whether long range AIS presents additional sharing issues. 14 Rep. ITU-R M.2123

3.1 Technical parameters The following paragraphs describe the basic technical parameters used in this study of sharing between long range AIS and the other co-channel mobile systems. 3.1.1 Mobile systems The first step in investigating long range AIS sharing with mobile systems is identification of technical parameters of LMR and VPCS systems. Table 4 lists representative technical parameters for VPCS and LMR systems. As seen in this table, both the VPCS and LMR systems typically employ an effective radiated power (ERP) 10 dB (or more) higher than the ship AIS transmitters sharing these frequencies.

TABLE 4 Typical VPCS and LMR technical parameters

Parameter Land mobile base Land mobile base VHF public station station correspondence coast (wideband) (narrow-band) station Transmit ERP 37-56 dBm 37-56 dBm 50-61 dBm (54 dBm typical) (54 dBm typical) Modulation 16F3E 11F3E 16F3E Channelling 25 kHz 12.5 kHz 25 kHz Typical antenna height 10-150 m 10-150 m 10-150 m (60 m typical) (60 m typical) Transmit duty cycle See Table 5 See Table 5 See Table 5

One of the important factors in evaluating sharing between AIS and other mobile systems is the mobile system transmit duty cycle. Most mobile communications systems operate at less than a 100% transmit duty cycle. A series of over-the-air spectrum measurements were completed in the United States in selected portions of the 138-174 MHz band that included the percent of time that a given channel was found to be in use.2 Based on these and other data sources, it is possible to broadly categorize mobile service transmitters into high (30-100%), medium (10-30%) and low (<10%) duty cycle categories. Although it is not practical to generate a complete list, examples for each category are given in Table 5. 3.1.2 AIS coast stations Recommendation ITU-R M.1371 describes in detail the characteristics of AIS systems as they are used for their primary purpose – improving safety-of-navigation. It is this safety aspect of AIS operation that dictates the requirement for a high level of reliability. For purposes herein, reliability refers to the ability to receive and correctly decode a high percentage of the transmitted ship AIS messages within its communications range. For the long range detection and tracking function described in this Report, a somewhat lower level of reliability is adequate. Table 6 summarizes key technical parameters for AIS coast station receivers used in this study for these two functions along with typical parameters for the associated ship transmitters.

2 See for example: SANDERS, F.H., HAND, G.R. and LAWRENCE, V.S. [September 1998] Land mobile radio channel usage measurements at the 1996 Summer Olympic Games. NTIA Report 98-357 (available at www.its.bldrdoc.gov/pub/ntia-rpt/98-357/. Rep. ITU-R M.2123 15

TABLE 5 Examples of mobile system transmit duty cycle

High duty cycle Medium duty cycle Low duty cycle (30-100%) (10-30%) (<10%) Paging systems Multiple user LMR Most single-user private LMR business/industrial repeaters systems (i.e. community repeaters) Trunking system control channel Public safety dispatch Most administrative government LMR systems Broadcast type systems (such as Trunking system communication Some types of LMR fixed weather broadcasts) channels control links Some transportable telemetry VHF maritime mobile working (such as seismic sensors) channels VHF public correspondence coast stations Some types of LMR fixed control links

TABLE 6 Typical parameters for an AIS ship-to-shore link

Parameter Safety-of navigation mission Long range tracking (via troposcatter) AIS ship transmitter Transmit power 41 dBm Transmit antenna gain 2 dBi Transmit miscellaneous losses 3 dB Transmit EIRP 40 dBm Antenna height ~ 10 M Modulation 9 600 bit/s GMSK Message rate 6 to 15 messages per minute (~8.5 average) AIS coast receiver Sensitivity –109 dBm for 1% per –120 dBm for 20% (Note 1)) Median signal level from edge of –104 dBm –115 dBm service area Antenna type Two element stacked dipole Multi-element array Antenna main beam gain 5 dBi (omnidirectional) 11-17 dBi (Note 2)) Antenna backlobe gain 5 dBi Estimated average ~ 0 dBi Antenna height 10-50 m (30 m typical) 10-50 m (30 m assumed) Miscellaneous losses 3 dB (assumed) 1 dB (assumed) NOTE 1 – The sensitivity specified for the long range function is considerable higher than the minimum value specified in Recommendation ITU-R M.1371 for ship-borne AIS receivers. In order to maximize the detection range for long range tracking, an optimized receiver is needed, with –120 dBm expected to be an achievable value. NOTE 2 – In order to maximize detection range, higher gain directional antennas are assumed.

16 Rep. ITU-R M.2123

3.1.3 Sharing factors The next step in evaluating sharing is designating sharing conditions and assumptions for this analysis. Within ITU-R Recommendations, a number of different methodologies are used to define sharing criteria for various radiocommunication service categories, most of which are performance based. One method is to define the sharing criteria in terms of a percent reduction in the performance of the victim system, with a 10% reduction in performance being a commonly used value. For AIS receivers, the principal measure of system performance is the packet error rate (PER). This is the percent of AIS message packets that are received and correctly decoded without error. Measurements completed on typical shipborne AIS receiver show that they are capable of achieving a PER of 20% with a desired-to undesired ratio (D/U) of 10 dB or more.3 This D/U ratio is assumed to be adequate for the long range tracking function. Performance levels have not been established for regular AIS coast station receivers. Because of their safety-of-navigation functions a higher D/U of 15 dB is assumed for this study. Taking the preceding factors into account, the following AIS sharing criteria is assumed for the analysis contained in this document: a) D/U < 10 dB for no more than 50% of the locations and 10% of the time for the long range tracking function. b) D/U < 15 dB for no more than 50% of the locations and 10% of the time for the primary safety-of-navigation function. These criteria only apply when the co-channel operating signal is present 100% of the time. When the type of mobile system is of the type where the transmit duty cycle is less than 100%, the D/U criteria may be relaxed. An assumed relaxation of 5 dB is proposed for a transmit duty cycle of 10% or less. 3.1.4 Radio propagation Appendices 1, 2 and 3 provide a detailed review of available ITU-R recommended models describing various radio propagation models appropriate for certain VHF communication in temperate climates. Sharing considerations using conventional diffraction mode propagation as described in Appendix 3 is used for the study described here. 3.1.5 Sharing analysis There is no existing ITU-R Recommendation establishing sharing criteria for AIS operation. It is common practice within many administrations to assign the frequencies used by AIS exclusively for that purpose only to land mobile stations far enough inland from navigable waterways and coastlines to ensure adequate performance. With the information presented in the preceding subsections, an analysis can be undertaken to compare distance separations for sharing between AIS coast stations and other co-channel mobile stations. The calculation is performed as follows:

D/U = Dmedian − [EIRP − Lp(d,%) + Gr − Lmisc] > D/UCriteria

3 Performance requirements established in IEC 61993-2 require AIS receivers to operate at a C/I ratio of 10 dB with a PER of 20% or less. Tests show that this co-channel interference test results vary little with the type of modulation used in the interfering signal. For example, results for GMSK and FM voice modulation are generally the same. Rep. ITU-R M.2123 17 where: D/U : desired to undesired ratio (dB)

Dmedian : median desired signal level from the edge of the service area (dBm) EIRP : mobile transmitter equivalent radiated power (dBm)

Lp(d,%) : propagation loss as a function of distance and percent occurrence (dB)

Gr : receiver antenna gain (dBi)

Lmisc : miscellaneous receiving system losses (dB) (assumed 3 dB)

D/UCriteria : applicable D/U protection criteria. Rearranging terms yields:

Lp(d,%) > D/UCriteria − Dmedian + EIRP + Gr − Lmisc

Using this equation and the propagation model described in Appendix 3, sharing conditions based on the nominal sharing criteria can be tabulated, shown in Table 7, and compared for the AIS safety-of-navigation and long range detection functions.

TABLE 7 Comparison of the calculated distance separation from co-channel LMR base stations to meet the assumed sharing AIS criteria

LMR Calculated LMR Calculated Propagation D antenna Propagation propaga- AIS function D/U (dB) median ERP distance mode criteria (dBm) height statistics tion loss (dBm) (km) (m) (dB) Safety-of 50% locations Diffraction 15 –104 50 15 171 117 navigation 10% time Safety-of 50% locations Diffraction 15 –104 56 60 177 127 navigation 10% time Long range 50% locations Diffraction 10 –115 50 15 169 115 tracking 10% time Long range 50% locations Diffraction 10 –115 56 60 175 125 tracking 10% time

As a result of varying site-specific considerations, it is not practical to identify a generic separation distance which could be appropriate for any sharing scenario. The calculated distances shown in Table 8 apply only for low lying inland coastal plain areas. No terrain features were considered. In many cases, coastal plain terrain does not extend inland for the distances calculated in this table; consequently additional factors need to be considered. Where mountainous areas are present additional diffraction losses significantly reduce these calculated distance values. The analysis results presented in this section indicate that, under conditions of diffraction mode propagation, similar distance separations would suffice irregardless of whether the AIS coast stations were operating in a safety-of-navigation function or long range detection function. In certain geographic and climatic regions additional propagation factors may be applicable. In Appendix 4, a comparison of the sharing for long range AIS and normal AIS is provided considering other propagation methods. 18 Rep. ITU-R M.2123

4 Summary This Report examined four options to enhance the detection range of shore-based AIS coast stations: – installation of AIS receivers on off-shore oil platforms; – installation of AIS receivers on off-shore weather buoys; – use of tropospheric scatter propagation (part-time enhancements); and – use of atmospheric ducting propagation (part-time enhancements). As summarized below, each option discussed entailed one or more limitations, as described in more detail in the text. However, each option was found to provide some degree of enhanced AIS detection range. Installation of AIS receivers on off-shore oil drilling platforms can greatly enhance shore AIS detection in an umbrella fashion around key ports areas but their use has very limited geographic applicability. Off-shore weather buoys are widely but sparsely distributed along many coast lines. Installation of AIS receivers on weather buoys offers limited range enhancement but may be useful in enhance detection capabilities along major shipping routes. Reliance on troposcatter propagation can increase AIS detection range from shore to 100 to 200 km but requires significant advances over current AIS receiver and antenna designs. Atmospheric ducting is also a propagation mode that can provide intermittent detection range enhancement to 100 to 200 km or more. Because long range detection of AIS is a new and evolving concept, the report also examined operations with other co-channel mobile systems whether long range AIS presents sharing issues beyond that for normal AIS functions. Based on conventional diffraction mode propagation, preliminary results developed herein show that similar distance separations would be sufficient irregardless of whether the AIS coast stations were operating in a safety-of-navigation function or long range detection function. This Report is an initial study that addresses possible implementation concepts and spectrum sharing issues relative to long range AIS detection. Further contributions are invited to provide measurements and/or analyses to supplement or update the results derived herein.

Appendix 1

Tropospheric scatter propagation

1 Introduction Tropospheric scatter is a mode of transhorizon radio wave propagation that results from the random reflections and scattering from irregularities in the dielectric gradient density of the troposphere. This propagation mode is applicable from below 100 MHz to above 8 000 MHz and extends for distances of several hundred kilometres. Although it is included herein under the general category of anomalous propagation, the propagation loss resulting from this effect, although quite large, can be sufficiently steady and predictable to support reliable long range communications. In contrast, somewhat lower propagation loss values occur on a more intermittent basis that sometimes leads to radio interference conditions. Figure 11 below illustrates this mode as compared with line-of-sight and diffraction knife-edge propagation modes. Rep. ITU-R M.2123 19

Two Recommendations of the ITU-R, Recommendations ITU-R P.452 and ITU-R P.617, provide long term statistical equations for predicting the tropospheric scatter propagation loss under various conditions. Because of some frequency limitations on these models, a third model was also investigated. The results of the investigation of these three models are discussed below.

1.1 Recommendation ITU-R P.617-1 model Recommendation ITU-R P.617-1 describes propagation prediction techniques and data required for the design of trans-horizon radio-relay systems. Since the propagation model described in this recommendation is for the design of radio systems, the focus is on moderate to high path reliability conditions where the propagation loss is not exceeded greater than about 20% of the time. Although the focus of this Recommendation is on the design of fixed radio-relay systems, the physics of the propagation mode is not limited to just the fixed service but is driven primarily by factors such as frequency, path geometry, terrain, and climate. The characteristics of this model are extrapolated to 162 MHz and summarized in the paragraphs below for purposes of this study for temperate climates. The applicable frequency limits stated for the propagation model described in Recommendation ITU-R P.617-1 is 200 MHz to 4 GHz. Since the frequency of interest is slightly outside of this range some uncertainty exists as to it’s validity at the lower frequency. However, other models described later in this Annex show consistent loss trends down to frequencies as low as 20 MHz. Consequently, it is assumed that any errors introduced by extrapolation of the methodology to 162 MHz are expected to be minor. The Recommendation divides the climatic regions of the world into nine zones: 1) Equatorial 2) Continental sub-tropical 3) Maritime sub-tropical 4) Desert 5) Mediterranean 6) Continental temperate 20 Rep. ITU-R M.2123

7a) Maritime temperate, overland 7b) Maritime temperate, oversea 8) Polar The focus of this study is on the temperate regions of the world defined as follows: Continental Temperate corresponds to regions between 30o and 60o latitude. Such a climate in a large land mass shows extremes of temperature and pronounced diurnal and seasonal changes in propagation conditions may be expected to occur. The western parts of continents are influenced strongly by oceans, so that temperatures here vary more moderately and rain may fall at any time during the year. In areas progressively towards the east, temperature variations increase and winter rain decreases. Propagation conditions are most favorable in the summer and there is a fairly high annual variation in these conditions. Maritime Temperate, Overland also corresponds to regions between latitudes of about 30o and 60o where prevailing winds, unobstructed by mountains, carry moist maritime air inland. Typical of such regions are the United Kingdom, the west coast of North America and of Europe and the north-western areas of Africa. Although the islands of Japan lie within this region, climate 6 is considered more appropriate. Maritime Temperate, Overseas corresponds to coastal and oversea areas in regions similar to those for climate 7a. The distinction is that a radio propagation path having both horizons on the sea is considered to be an oversea path (even though the terminals may be inland); otherwise climate 7a is considered to apply. From this Recommendation the average tropospheric scatter propagation loss can be described by a series of equations as follows:4

L(q) = M + 30 log f + 10 log d + 30 log θ + LN − Y(q)

θ = θe + θt + θr 3 θe = d · 10 /(k · a)

LN = 20 log(5 + γ · H) + 4.34 γ · h H = 10–3 θ · d/4 h = 10–6 θ2 · k · a/8 Y(q) = C(q) · Y(90) Y(90) = 2.2 − (8.1 − 2.3 · 10−4 · f) · e (−0.137 h) for climate regions 6 and 7a

= –9.5 − 3 · e (−0.137 h) for climate region 7b where: M : meteorological factor defined by the climatic region (dB) = 29.73, 33.20, 26.00 for climate regions 6, 7a, and 7b, respectively f : frequency (MHz) d : transmitter-to-receiver distance (km)

4 The equation defined in the recommendation is for transmission loss that includes the antenna gain factors. Since the present study addresses relatively low gain antennas, the aperture-to-medium coupling loss factor described in the recommendation is effectively zero. Consequently, the equation can be described herein for propagation loss without the antenna gain factors with insignificant error. Rep. ITU-R M.2123 21

θ : effective scatter angle (mrad)

θe : geocentric angle between transmitter and receiver (mrad)

θt,r : transmitter and receiver horizon angles (mrad) (assumed to be 0)

LN : propagation loss dependence on the height of the common volume Y(q) : conversion factor for percentage of time other than 50% (dB) k : effective Earth radius factor (assumed to be 4/3) a : mean Earth radius (6 370 km) γ : atmospheric structure parameter (0.27 for temperate climate regions 6, 7a, and 7b) H, h : height factors (km) C(q) : probability factor (dB) (defined in Table8) W : factor to convert average annual loss to worst month loss (see Fig. 11).

TABLE 8 Values of C(q) of interest q 20* 50 80* 90 99 99.9 99.99 C(q) –0.66 0 0.66 1.00 1.82 2.41 2.90 * Interpolated values added.

Using the equations as described above, Figs. 12 through 14 give example calculations for transmit and receive antenna heights of 10 m and 50 m, respectively.

FIGURE 12

22 Rep. ITU-R M.2123

FIGURE 13

FIGURE 14

1.2 Recommendation ITU-R P.452-12 model Recommendation ITU-R P.452-12 provides a prediction procedure for the evaluation of microwave interference between stations on the surface of the Earth at frequencies above 0.7 GHz. However, the recommendation states that the propagation models are believed valid down to 100 MHz. Since this recommendation addresses interference paths, the focus is on propagation statistics below 50% and is not appropriate for time percentages above 50%. The recommendation notes that for time percentages much below 50%, it is difficult to separate true tropospheric scatter mode propagation from other secondary phenomena. The model is an empirical generalization to include these other secondary effects as well. From this Recommendation the tropospheric scatter propagation loss for any defined percentage of time, p, can be described by a series of equations as follows: 0.7 L(p) = 190 + Lf + 20 log d + 0.573 θ – 0.15 N0 – 10.1 [–log (p / 50)] 2 Lf = 25 log(f) – 2.5 [log(f / 2)]

θ = θe + θt + θr 3 θe = d · 10 / (k · a) Rep. ITU-R M.2123 23 where:

Lf : frequency loss factor (dB)

N0 : sea-level surface refractivity (see Fig. 15) θ : path angular distance (mrad)

θe : defined as above

θt, θr : defined as above (assumed to be zero) f : frequency (GHz) p : time percentage (%).

FIGURE 15

Examples from this model are shown in Figs. 16 and 17. 24 Rep. ITU-R M.2123

FIGURE 16

FIGURE 17

2 Irregular terrain model The irregular terrain model (ITM) estimates radio propagation losses for frequencies between 20 MHz and 20 GHz as a function of distance and the variability of the signal in time and space. It is an improved version of the Longley-Rice Model, which gives an algorithm developed for computer applications.5 The model is based on electromagnetic theory and signal loss variability expressions derived from extensive sets of measurements. The ITM algorithm works in two modes: – point-to-area prediction mode – used when an exact terrain description is not available, and – point-to-point prediction mode – used when the terrain profile between the terminals is available.

5 See http://flattop.its.bldrdoc.gov/itm/itm_alg.pdf. Rep. ITU-R M.2123 25

The “point” is where a broadcast station or a base station for mobile service may be located and “area” refers to the service area locations of broadcast receivers or mobile stations. The model considers several propagation modes, one of which is tropospheric scatter, and is applicable for frequencies 20 MHz through 20 GHz. This model was developed using similar principles to those described in the two ITU-R Recommendations discussed above. In order to supplement the tropospheric scatter propagation data derived earlier from relevant ITU-R Recommendations, especially for frequencies below 200 MHz, the ITM model was exercised for the frequency 162 MHz for distances greater than 100 km where the tropospheric scatter mode predominates. Figures 18 and 19 plot the resulting tropospheric scatter propagation loss for continental temperate and maritime temperate over water for probability statistics ranging from 1 to 99%.

FIGURE 18 Tropospheric scatter loss using the ITM model

FIGURE 19

26 Rep. ITU-R M.2123

3 Summary Although the three tropospheric scatter models described above gave different values for median propagation loss, the differences in most cases were no more than 3 dB. The consistent results among the models suggest that extrapolating the model defined in Recommendation ITU-R P.617 beyond its stated limits may be suitable. Consequently, for purposes of this study only it is assumed that the two models defined by Recommendations ITU-R P.617 and ITU-R P.452 can be used for design and interference calculations of VHF AIS systems, respectively.

Appendix 2

Atmospheric ducting propagation

1 Introduction Atmospheric ducting is a propagation phenomena that been known and studied in some detail since before the 1980s.6 The condition occurs predominately when the atmosphere stratifies into layers of differing temperatures and humidity. This layering results in varying index of refraction at different heights above ground. This condition, in turn, can lead to a condition where the varying indices of fraction can effectively be “trapped” between the layers somewhat similar to a waveguide. The trapping can occur between the Earth’s surface and a ducting layer, called a surface duct, or between to upper atmosphere ducting layers, called an elevated duct. Figure 20 compares atmospheric ducting with other short term propagation phenomena. Recommendation ITU-R P.452 is the only ITU-R text that addresses atmospheric ducting. However the recommendation states that the ducting model has not been tested to frequencies lower than about 0.7 GHz. Noting this uncertainty, it is assumed for purposes of this study to be applicable at 162 MHz. The key features of this model are summarized in the following paragraphs based on an extrapolation to the VHF band. Other methodologies are also discussed.

6 See for example: DOUGHERTY, H.T. and DUTTON, E.J. [1980] The Role of Elevated Ducting for Radio Service and Interference Fields. National Telecommunications and Information Administration. Rep. ITU-R M.2123 27

2 Recommendation ITU-R P.452 This Recommendation is intended for the evaluation of microwave interference between stations on the surface of the Earth at frequencies above about 0.7 GHz. The recommendation states that surface ducting is the most important short-term interference mechanism over water and in flat coastal land areas, and can give rise to high signal levels over long distances. Signals received under these conditions can exceed the equivalent free space values under certain conditions. The calculated ducting propagation loss calculated by this model for a given percent of time is a complex function of frequency, latitude, path geometry and climatic region. The climatic regions are defined as: – coastal land, – other inland areas and – sea. A simplification of the basic equations describing the model is as follows:

Lba (p) = 102.45 + 20 log f + 20 log (dlt + dlr) + Ad (p)

Ad (p) = γd · θ′ + A (p) –5 1/3 γd = 5 × 10 ae · f θ′ = Same as θ defined in Appendix 1 where:

dlt, dlr : distances to the optical horizon from the transmitter and receiver antennas, respectively

ae : effective Earth radius f : frequency (GHz) A(p) : complex function of geometry, latitude and climatic region. 28 Rep. ITU-R M.2123

Figure 21 evaluates these equations extrapolated to the frequency 162 MHz over smooth terrain with no intervening obstacles with transmitter and receiver antenna heights of 10 m and 50 m, respectively.

FIGURE 21

Ducting propagation loss at 162 MHz using the ITU-R P.452 model (Smooth terrain)

3 Hepburn predictions The Hepburn predictions are a service provided to the radio amateur community for those interested in long distance communications via atmospheric ducting.7 These predictions take into account terrain, weather conditions and other factors to provide qualitative atmospheric ducting predictions. The qualitative predictions range from “marginal” to “extremely intense opening” for ducting communications paths. No propagation loss values are provided. Consequently, these predictions are of limited value when specific quantitative results are required. Copyright restrictions also constrain widespread use. However, these predictions do provide a useful visualization of the effects of terrain and local climate conditions on atmospheric ducting.

7 See http://www.dxinfocentre.com/tropo.html. Rep. ITU-R M.2123 29

Appendix 3

Diffraction propagation

1 Introduction Diffraction propagation loss is the principal propagation mode applicable for most desired and interfering paths among terrestrial radiocommunication systems. A number of factors influence diffraction mode propagation including frequency, distance, transmitter and receiver antenna heights, surface admittance and conductivity, and polarization. However, at 162 MHz, the latter three factors generally do not play a major role. Recommendation ITU-R P.526 describes in detail the methodology for calculation of diffraction mode propagation losses.

2 Recommendation ITU-R P.526 Recommendation ITU-R P.526 describes the calculation of propagation loss over diffraction paths including a spherical earth surface or over irregular terrain. For purposes herein only the smooth spherical earth is considered. At 162 MHz, the smooth earth diffraction propagation loss can be described by the following equations:

LDiff = LFS + F(X) + G(Y1) + G(Y2) F(X) = 11 + 10 log (X) − 17.6 ·X X = 0.029 ·D G(Y) = 20 log (Y + 0.1 · Y3) (NOTE – Different equations apply for very low and very high antennas) Y = 0.014 H

where:

LDiff : smooth Earth diffraction loss (dB)

LFS : free space propagation loss (dB) F(X) : distance factor (dB)

G(Y1,2) : height gain factors (dB) D : distance separation H : transmitter, receiver antenna height (m).

Implementation of the methods described above yields the following results for several example cases. 30 Rep. ITU-R M.2123

FIGURE 22 Calculated diffraction propagation loss over smooth terrain using ITU-R P.526 model (162 MHz; Receiver height = 30 m; 50% of locations; 50% of time)

FIGURE 23 Calculated diffraction propagation loss over smooth terrain using ITU-R P.526 model (162 MHz; Receiver height = 30 m; 50% of locations; 10% of time)

Rep. ITU-R M.2123 31

Appendix 4

Additional spectrum sharing considerations

Section 3 of the Report addresses spectrum sharing between AIS systems and other co-channel mobile systems. The results were derived assuming normal diffraction mode propagation mechanisms. This appendix addresses additional sharing factors taking into account anomalous propagation modes, specifically tropospheric scatter and atmospheric ducting. Following the procedures described in § 3, the nominal distance separation for long range AIS and normal AIS under conditions of anomalous propagation can be estimated and compared using the following:

D/U = Dmedian – [EIRP – Lp(d,%) + Gr – Lmisc] > D/UCriteria

where: D/U : desired to undesired ratio (dB)

Dmedian : median desired signal level from the edge of the service area (dBm) EIRP : mobile transmitter equivalent radiated power (dBm)

Lp(d,%) : propagation loss as a function of distance and percent occurrence (dB)

Gr : receiver antenna gain (dBi)

Lmisc : miscellaneous receiving system losses (dB) (assumed 3 dB)

D/UCriteria : applicable D/U protection criteria. Rearranging terms yields:

Lp(d,%) > D/UCriteria – Dmedian + EIRP + Gr – Lmisc

Using this equation and the extrapolated propagation models described in Appendices 1 and 2, the nominal distance separation can be tabulated, shown in Table9, and compared for the AIS safety-of-navigation and long range detection functions. The calculated distances shown in Table 9 apply only for low lying inland coastal plain areas. No terrain features were considered. In many cases, coastal plain terrain does not extend inland for the distances calculated in this table; consequently additional factors need to be considered. As a result of these varying site-specific considerations, it is not practical to identify a generic separation distance which could be appropriate for any sharing scenario. The nominal distance separations calculated above using anomalous propagation modes are significantly larger, as expected, than those calculated under normal diffraction mode propagation. The analysis results presented in this section indicates that, comparable to the case of diffraction mode propagation, similar distance separations would suffice irregardless of whether the AIS coast stations were operating in a safety-of-navigation function or long range detection function. 32 Rep. ITU-R M.2123

TABLE 9 Calculated distance separation from co-channel mobile systems to meet the assumed sharing AIS criteria

Mobile Calcula- Mobile antenna ted Calculated Propagation D/u D Propagation AIS function criteria median ERP height propaga- distance mode (dB) (dBm) statistics (dBm) (m) tion loss (km) (dB) Safety-of Tropo- 50% locations 15 –104 50 15 171 200 navigation scatter 10% time Safety-of Tropo- 50% locations 15 –104 56 60 177 220 navigation scatter 10% time Long range Tropo- 50% locations 10 –115 50 15 169 190 tracking scatter 10% time Long range Tropo- 50% locations 10 –115 56 60 175 210 tracking scatter 10% time Safety-of 50% locations Ducting 15 –104 50 15 171 375 navigation 10% time Safety-of 50% locations Ducting 15 –104 56 60 177 400 navigation 10% time Long range 50% locations Ducting 10 –115 50 15 169 370 tracking 10% time Long range 50% locations Ducting 10 –115 56 60 175 395 tracking 10% time

Appendix 5

AIS propagation observations – Ducting

1 Introduction Techniques and methods to improve AIS signal detection at shore facilities are the subject of field trials. Observations made during these trials provide practical examples that generally support the material presented in this paper.

2 Atmospheric ducting The impact of ducting can be viewed two ways. – Ducting enhances the performance of a system intended for long range tracking. – Ducting introduces interference sources to signals from AIS stations within the nominal AIS reception range. This can result in a reduction of the performance of AIS both ashore and on vessels. Rep. ITU-R M.2123 33

Figures 24 and 25 plot data collected during 24 h of a 48 h ducting event during 7-9 October 2006. The figures provide actual signal level information from a high performance receiver installation. The plots show all the AIS signals received during 8 October 2006. The signal measurements are displayed based upon the distance separation of the transmitting AIS unit from the AIS receiver installation. The “message color bar” at the right of the figures indicates the number of messages (by colour) that are involved with each calculation. Figure 24 plot shows the peak, average, and minimum signal measurements for each nautical mile from 1 to 200 nautical miles. The level indicated by the black triangles show the strength of the weakest received AIS messages. This level is directly affected by “local radio noise” conditions. Figure 25 shows the number of messages received at each dB level for each mile of the 1 to 200 nautical miles. The figure consists of color cells that show the number of received messages. The color cells dimensions are 1 nautical mile by 1 dB. The plots can be used to show how atmospheric ducting is useful for long range detection of ships. The plots can also be used to show that during ducting events, the strength of signals received from distant ships can be greater than signals from nearby ships. This is not a problem since the AIS technology was designed to operate properly under these conditions. The logic built into every AIS unit would enable the AIS units to organize their broadcasts in a manner to avoid causing co- channel AIS interference.

FIGURE 24

34 Rep. ITU-R M.2123

FIGURE 25 Observed AIS signal levels during 8 October 2006 “ducting event.”

The plots could also be used to illustrate how ducting could allow other types of emissions to affect local AIS signals on an intermittent basis. Under the ducting conditions shown in the plots, it is possible that a non-AIS continuous signal from 150 to 200 miles away could affect reception of an AIS signal from a AIS unit less than 10 miles away. The distances in the plots considered reception of signals from ships at sea; for land based transmissions, other factors such as terrain features or attenuation from manmade structures would need to be taken into account. These signals could be co-channel signals from non-AIS services using the AIS channels or spurious emissions from non- AIS transmitters. They could effectively raise the “local radio noise,” as viewed from the perspective of the AIS technology. To be reliably detected, an AIS signal needs to be 10 dB stronger (the AIS test standard for receiver co-channel rejection) than the “local radio noise”. The conditions shown in the plots could result in degraded AIS performance for the duration of the ducting event. However, given its sporadic and intermittent occurrence, ducting should not be used as a basis for determining compatibility between VHF systems. Ducting is an anomalous propagation occurrence that can occur for days or weeks at a time. It is dependant on geographical location and electromagnetic atmospheric effects.

Copy No. _____

Defence Research and Recherche et développement Development Canada pour la défense Canada

& DEFENCE DÉFENSE

VHF Propagation Study

D. Green, C. Fowler and D. Power C-CORE

J. K. E. Tunaley London Research and Development Corporation

Prepared by: C-CORE 1 Morrissey Road St. John's, NL A1B 3X5

Contractor's Document Number: R-11-020-868 Contract Project Manager: Chris Fowler, 709-864-8373 PWGSC Contract Number: W7707-115279 Contract Scientific Authority: Anna-Liesa Lapinski, Defence Scientist, 902-216-3100 ext. 180

The scientific or technical validity of this Contract Report is entirely the responsibility of the contractor and the contents do not necessarily have the approval or endorsement of Defence R&D Canada.

Defence R&D Canada – Atlantic Contract Report DRDC Atlantic CR 2011-152 September 2011

This page intentionally left blank.

VHF Propagation Study

D. Green C-CORE

J. K. E. Tunaley London Research and Development Corporation

C. Fowler C-CORE

D. Power C-CORE

Prepared By: C-CORE 1 Morrissey Road St. John's, NL A1B 3X5

Contractor's Document Number: R-11-020-868 Contract Project Manager: Chris Fowler, 709-864-8373 PWGSC Contract Number: W7707-115279 CSA: Anna-Liesa Lapinski, Defence Scientist, 902-216-3100 ext. 180

The scientific or technical validity of this Contract Report is entirely the responsibility of the Contractor and the contents do not necessarily have the approval or endorsement of Defence R&D Canada.

Defence R&D Canada – Atlantic Contract Report DRDC Atlantic CR 2011-152 September 2011

Principal Author

Original signed by David Green David Green C-CORE

Approved by

Original signed by Francine Desharnais Francine Desharnais Head, Maritime Information and Combat Systems Section

Approved for release by

Original signed by Ron Kuwahara for Calvin Hyatt Chair, Document Review and Library Committee

© Her Majesty the Queen in Right of Canada, as represented by the Minister of National Defence, 2011 © Sa Majesté la Reine (en droit du Canada), telle que représentée par le ministre de la Défense nationale, 2011

Abstract ……..

This literature review provides DRDC researchers with information related to the propagation of VHF signals, particularly as it relates to the fluctuating limits of Automatic Identification System (AIS) message reception. The review focuses on the AIS frequency range and the factors that influence signal propagation in Maritime environments. The approach taken was consultation with classical textbooks on propagation to capture fundamental equations, followed by a search of the literature for papers involving VHF propagation of AIS signals. Three effects were determined to most likely extend the range of AIS transmission: diffraction over the sea around the curvature of the earth, ducting resulting from the varying refractivity of air, and multipath effects.

Résumé ….....

La présente revue de littérature fournit aux chercheurs de RDDC des renseignements sur la propagation des signaux VHF, particulièrement en ce qui concerne les limites fluctuantes de réception de messages du Système d'identification automatique (AIS). Elle porte surtout sur la gamme de fréquences de l’AIS et les facteurs qui influencent la propagation des signaux en milieu maritime. L’approche utilisée a consisté à d’abord consulter des manuels classiques sur la propagation afin de relever les équations fondamentales, puis à effectuer des recherches dans les publications scientifiques pour trouver des articles portant sur la propagation des signaux AIS. Ce travail a permis de déterminer que trois effets étendent très vraisemblablement la portée de transmission AIS : la diffraction sur la surface de la mer en raison de la courbure terrestre, la propagation guidée résultant de la variation de l’indice de réfraction de l’air, et les effets des trajets multiples.

DRDC Atlantic CR 2011-152 i

This page intentionally left blank.

ii DRDC Atlantic CR 2011-152

Executive summary

VHF Propagation Study D. Green; J. K. E. Tunaley; C. Fowler; D. Power; DRDC Atlantic CR 2011-152; Defence R&D Canada – Atlantic; September 2011. Introduction: This literature review provides DRDC researchers with information related to the propagation of VHF signals, particularly as it relates to the fluctuating limits of Automatic Identification System (AIS) message reception. The review focuses on the AIS frequency range and the factors that influence signal propagation in Maritime environments.

The approach followed is designed to meet the requirements laid out in the Request for Proposal (RFP) to provide DRDC with information regarding the propagation of VHF signals in Canada’s coastal areas. This approach ensures that the literature review is well substantiated and that all supporting information is traceable to reliable sources. The approach was, for the most part, consultation with classical textbooks on propagation to capture fundamental equations, followed by a search of the literature for papers involving VHF propagation of AIS signals (modeling, effects, issues, etc.).

Results: Three primary phenomena were determined to impact AIS transmission. Diffraction over the sea around the curvature of the earth, as well as ducting resulting from a variable refractivity of air, would both potentially extend the transmission range of AIS signals. Multipath causes significant variability in the signal strength. These are the major factors identified in the software implementation of ITU-R (International Telecommunication Union Recommendation) P.452-14, Prediction procedure for the evaluation of interference between stations on the surface of the Earth at frequencies above about 0.1 GHz, and hence it is recommended that this implementation be considered for further modelling of VHF propagation effects.

Significance: Using knowledge of the VHF propagation, it is likely possible to predict areas of reliable AIS coverage based on current conditions or forecasted conditions. Reliable AIS reception can be used to indicate areas where a surveillance officer can be assured that broadcast AIS messages are being received reliably. In areas identified as having unlikely coverage, surveillance officers can recognize that not all broadcasting ships in the area will be identified using AIS.

Future plans: Future work could include using the identified factors that influence the temporal variations in AIS coverage to predict AIS coverage.

DRDC Atlantic CR 2011-152 iii

Sommaire .....

VHF Propagation Study D. Green; J. K. E. Tunaley; C. Fowler; D. Power; DRDC Atlantic CR 2011-152; R & D pour la défense Canada – Atlantique; Septembre 2011. Introduction : La présente revue de littérature fournit aux chercheurs de RDDC des renseignements sur la propagation des signaux VHF, particulièrement en ce qui concerne les limites fluctuantes de réception de messages du Système d'identification automatique (AIS). Elle porte sur la gamme de fréquences de l’AIS et les facteurs qui influencent la propagation des signaux en milieu maritime.

L’approche suivie a été choisie pour répondre aux exigences de la demande de propositions (DP), en vue de fournir à RDDC des renseignements sur la propagation des signaux VHF dans les zones côtières du Canada. Cette approche permet de veiller à ce que la revue de littérature soit bien justifiée et que l’origine de toute l’information donnée en appui puisse être attribuée à des sources fiables. En somme, l’approche a consisté à d’abord consulter des manuels classiques sur la propagation afin de relever les équations fondamentales, puis à effectuer des recherches dans les publications scientifiques pour trouver des articles portant sur la propagation des signaux AIS (modélisation, effets, questions, etc.)

Résultats : Les travaux effectués ont permis de déterminer que trois principaux phénomènes influent sur les transmissions AIS. La diffraction sur la surface de la mer en raison de la courbure terrestre ainsi que la propagation guidée résultant de la variation de l’indice de réfraction de l’air pourraient toutes deux étendre la portée des signaux AIS. Les trajets multiples créent une variabilité appréciable de l’intensité des signaux. Il s’agit des principaux facteurs distingués dans l’implémentation logicielle de la recommandation de l’Union internationale des télécommunications P.452-14, intitulée Prediction procedure for the evaluation of interference between stations on the surface of the Earth at frequencies above about 0.1 GHz [Procédure de prédiction pour l’évaluation du brouillage entre des stations sur la surface de la Terre à des fréquences au-dessus d’environ 0,1 GHz]; il est donc recommandé d’envisager l’utilisation de cette implémentation pour la modélisation future des effets de propagation VHF.

Portée : L’utilisation de connaissances sur la propagation VHF pourrait permettre de prédire les zones où la couverture AIS est fiable en fonction des conditions actuelles ou des conditions prévues. Une réception AIS fiable peut servir à indiquer les zones où un officier de surveillance peut être sûr que les messages AIS émis sont reçus de manière fiable. Dans les zones pour lesquelles la couverture est peu probable, les officiers de surveillance peuvent tenir compte du fait que les navires émettant dans la zone ne seront pas tous identifiés par AIS.

Recherches futures : Les recherches futures pourraient utiliser les facteurs qui influencent la variation de la couverture AIS dans le temps en vue de la prédire.

iv DRDC Atlantic CR 2011-152

Table of contents

Abstract ……...... i Résumé …...... i Executive summary ...... iii Sommaire ...... iv Table of contents ...... v List of figures ...... vii List of tables ...... viii 1. Introduction ...... 1 2. Literature Review Methodology ...... 2 3. AIS Technology ...... 3 3.1. SOLAS ...... 4 3.2. Signal Specifications ...... 5 3.3. Signal Modulation ...... 6 3.4. Hardware ...... 10 3.5. Link Budget ...... 11 4. Troposphere Effects on AIS Technology ...... 12 4.1. Tropospheric Scatter ...... 12 4.2. Refractive Index of Air ...... 14 4.2.1. Atmospheric Profile ...... 18 4.2.2. Refraction ...... 19 4.2.3. Software ...... 20 4.2.4. Ray Methods ...... 21 4.2.4.1. Ducting ...... 22 4.2.5. Parabolic Equation Methods ...... 23 4.3. Lightning ...... 25 4.4. Hydrometeors ...... 25 5. Ionosphere Effects on AIS Technology ...... 28 5.1. D Region ...... 28 5.2. Sporadic E ...... 29 5.3. Meteor Trails ...... 30 6. Other Propagation Effects ...... 33 6.1. Multipath ...... 33 6.2. Propagation Over Land ...... 36 6.3. Diffraction Over Sea ...... 38 7. ITU Model Applied to AIS ...... 41

DRDC Atlantic CR 2011-152 v

8. Conclusions ...... 44 References ...... 45 List of symbols/abbreviations/acronyms/initialisms ...... 47 Distribution list ...... 49

vi DRDC Atlantic CR 2011-152

List of figures

Figure 1: Spectral densities as a function of normalized frequency ...... 8 Figure 2: GMSK degradation...... 9 Figure 3: Bit Error Rates for GMSK maximum likelihood coherent demodulation () and non-coherent demodulation () ...... 10 Figure 4: Geometry of tropospheric scattering ...... 13 Figure 5: Refractivity of air as a function of temperature for relative humidities of 0% (), 50% (), 80% () and 100% ()...... 18 Figure 6: Loss due to refraction around the earth ...... 40

DRDC Atlantic CR 2011-152 vii

List of tables

Table 1: Class A Shipborne Mobile Equipment Reporting Intervals ...... 5 Table 2: Reporting Intervals for Equipment Other than Class A Shipborne Mobile Equipment .... 5 Table 3: AIS Transmission Parameters ...... 6 Table 4: Message Report Types (IDs) ...... 6 Table 5: Class A Link Budget Parameters ...... 11 Table 6: AIS Parameters ...... 21 Table 7: Parameters for Diffraction Loss ...... 39 Table 8: Input Data ...... 41 Table 9: Radio-climatic zones ...... 42

viii DRDC Atlantic CR 2011-152

Introduction

The primary objective of this study is to conduct a literature review to provide a basis of understanding on the principles of VHF propagation. This was done in the context of AIS signal reception in a Maritime environment. The review provides the scientific literature that forms the basis of understanding VHF signal propagation and highlights information sources that are highly relevant.

A secondary objective of this project is to identify sources that can provide suitable data for use in calculating VHF propagation limits for various geographic regions around the country. Emphasis was given to coastal regions of Canada’s Atlantic, Pacific and Arctic oceans.

DRDC Atlantic CR 2011-152 1

Literature Review Methodology

The initial component of the literature review was to establish a parameter list for a literature search. The parameter list includes a list of keywords parameterized according to subject, paper title and author. The parameter list was initially established in consultation with DRDC; however, it was appended to as the literature search proceeded. For instance, if a relevant article was found due to a keyword search, then the associated author was added to the list of keywords. The parameter list will be maintained in a checklist format to ensure a comprehensive search was completed. From this, key documents will be identified for subsequent analysis and reporting.

Personnel from C-CORE have access to the Queen Elizabeth II Library belonging to Memorial University of Newfoundland. Being within the university’s firewall allows access to library resources and online databases, allowing the project team to view full text articles. In particular, personnel are allowed access to the Institute of Electrical and Electronics Engineers (IEEE) Xplore service, which provides the full text of all articles published within the institute. Personnel from London Research and Development Corporation (LRDC) have access to the library system at the University of Western Ontario for the purposes of research, subject to library regulations and copyright.

2 DRDC Atlantic CR 2011-152

AIS Technology

In order to better understand the impacts of the various VHF propagation effects on AIS transmission, it is necessary to review the developments and the fundamentals of AIS technology. The transmission frequencies, multiple access schemes, and signal modulation techniques are all important considerations when evaluating the propagation phenomena outlined in this report. In 2000, as a part of the Safety Of Life At Sea (SOLAS) regulations, the International Marine Organization (IMO)1 added Automatic Identification of Ships (AIS) to the shipboard navigational carriage requirement for a number of ship categories. These categories are ships of 300 tons (gross) or greater that travel internationally, cargo ships of 500 tons gross or greater, and all passenger ships. The requirement came into full force for these ships on December 31st 2004 and the system is known as “Class A” AIS. After this date all ships in service in the said categories are mandated to operate their AIS equipment continuously except where international agreements allow navigational data to be protected. In 2007, Class B was introduced for small craft, including pleasure vessels.

AIS was conceived mainly as a collision avoidance system and is based on regular VHF transmission and reception of short binary messages containing information about the ship’s identity and includes its position, speed and course. These are “dynamic data”. “Static data,” such as the ship’s name, IMO number, cargo type and estimated time of arrival (ETA) are also transmitted but less frequently. The AIS system is specified in an International Telecommunication Union (ITU) document ITU-R M1371 [1]. The latest version (#4, April 2010) can be found at the ITU site2 and it is also possible to download the first version, which was published in 1998. AIS systems can also be used for other types of safety related messaging as well as base station interrogations and commands. Another useful document has been published by the International Association of Lighthouse Authorities (IALA); this is IALA Technical Clarifications on Recommendation ITU-R M.1371-13.

Reporting rates are specified in ITU-R M1371-4 in Annex 1, Tables 1 and 2. The dynamic data rate varies with ship speed from a minimum of two seconds for Class A (five seconds for Class B) at speeds greater than 23 knots to three minutes for ships at anchor. For the static data a report is transmitted every six minutes. The two VHF bands are also specified in this annex as well as the transmitted power and the modulation type (Gaussian Minimum Shift Keying (GMSK)) and its parameters. Other useful documents specifying GPS and control performances are IMO MSC.74(69) and IMO NAV 48(18).

The AIS systems are based on Time Domain Multiple Access (TDMA). This means that short messages are sent during specific time slots. To avoid confusion when the signal traffic is high, schemes are adopted to ensure that signals are not transmitted simultaneously by different ships into the same time slot. For Class A this is a self-organizing method called Self-Organized Time Domain Multiple Access (SOTDMA). In this method, a transceiver actively searches for an appropriate empty slot before transmitting. For Class B, a transceiver first listens to a slot to

1 http://www.imo.org/OurWork/Safety/Navigation/Pages/AIS.aspx 2 http://www.itu.int/rec/R-REC-M.1371/en 3 e.g. http://www.ialathree.org/iala/pages/AIS/IALATech1.5.pdf

DRDC Atlantic CR 2011-152 3

determine if anyone is using it and, if it is free, proceeds to transmit. This is known as Carrier- Sense TDMA (CSTDMA).

Class A messages are divided into various types. The position reports are Types 1, 2 and 3, which contain dynamic data; the static reports are of Type 5. The position reports for Class B messages are Types 18 and 19, while the static report is Type 24. There is also a report designed for long range transmissions to a satellite with an altitude less than 1000 km; this is Type 27. The formats of the reports are defined in terms of bit patterns4. These are compact in the sense that the length of a message in bits is minimal.

National governments can add carriage requirements on to those specified by the IMO. The situation in the United States is described on the US Coast Guard Navigation Center website5.

When AIS is operated as a terrestrial system, the SOTDMA protocols ensure that signals from different ships usually do not interfere with one another. However the number of time slots is limited to 2250 on each of two VHF channels and these slots are reassigned every 60 seconds. Therefore in a very high shipping density, some signals may be dropped. The system is configured so that the weaker signals in the far range are omitted; it results in a reduction of the size of a communications cell and so has little effect on the collision avoidance aspect of the system.

As a collision avoidance system, AIS needs only a line of sight communications system. Therefore the VHF band of frequencies is ideal for terrestrial use. This often limits the range to about 20 nmi but it can be somewhat greater under favourable propagation conditions, such as ducting. The limitation in range implies that the number of ships in the field of view of the antenna that can be accommodated by the AIS scheme is sufficiently small to avoid interference. The typical range can be estimated from the heights of the antennas, h1,2, by assuming that the troposphere has its average properties. This latter condition corresponds to setting the radius of the earth to 4/3 of its true value. Therefore, according to simple geometry the typical line of sight range, R, is given in kilometres by (e.g. [2]):

1/ 2 1/ 2 R 4.131u h1 h2 (1) Unfortunately the implementation of AIS is subject to error. Some of the errors made in setting up the ship-board systems are discussed in [3]. For example in various studies, the Maritime Mobile Service Identity (MMSI) number was incorrect 2% of the time and some authors claim that up to 80% of the AIS setups contained at least one error, though this is usually minor.

SOLAS The Class A and Class B reporting intervals are provided in Table 1 and Table 2. The transmission parameters in section 3.2 are a result of these specifications.

4 http://www.navcen.uscg.gov/?pageName=AISMessagesBA 5 http://www.navcen.uscg.gov/

4 DRDC Atlantic CR 2011-152

Table 1: Class A Shipborne Mobile Equipment Reporting Intervals

Ship's dynamic conditions Nominal reporting interval Ship at anchor or moored and not moving faster than 3 knots 3 min Ship at anchor or moored and moving faster than 3 knots 10 s Ship 0-14 knots 10 s Ship 0-14 knots and changing course 3 1/3 s Ship 14-23 knots 6 s Ship 14-23 knots and changing course 2 s Ship > 23 knots 2 s Ship > 23 knots and changing course 2 s

Table 2: Reporting Intervals for Equipment Other than Class A Shipborne Mobile Equipment

Platform’s condition Nominal reporting interval Class B “SO*” shipborne mobile equipment not moving faster than 2 knots 3 min Class B “SO” shipborne mobile equipment moving 2-14 knots 30 s Class B “SO” shipborne mobile equipment moving 14-23 knots 15 s Class B “SO” shipborne mobile equipment moving > 23 knots 5 s Class B “CS**” shipborne mobile equipment not moving faster than 2 knots 3 min Class B “CS” shipborne mobile equipment moving 2-14 knots 30 s Search and rescue aircraft (airborne mobile equipment) 10 s Aids to navigation 3 min AIS base station 10 s * SO = Self-organized ; ** CS = carrier sense

Signal Specifications The important parameters specified in [1] for AIS transmissions are provided in Table 3.

DRDC Atlantic CR 2011-152 5

Table 3: AIS Transmission Parameters

Parameter Value Frequency AIS-1 (MHz) 161.975 Frequency AIS-2 (MHz) 162.025 Bit rate (bits/sec) 9600 Line code NRZI6 Modulation GMSK7 Number of training bits 24 Transmit Bandwidth Time Product |0.4 Receive Bandwidth Time Product |0.5 Class A Transmitter Power (W) 12.5 Channel bandwidth (kHz) 25

The following message types in Table 4 are relevant to the principal reports of interest. The reports are 256 bits long and, with the exception of message 27, a string of 24 bits within the message is used as a buffer to accommodate bit stuffing, distance variations, repeater delays and synchronization jitter. Of these 24 bits, 12 bits are allocated to distance delay. Noting that the bit rate is 9600 bits/s, it is easily verified that the delay corresponds to a distance of about 375 km or 202 nmi.

Table 4: Message Report Types (IDs)

Message ID Description 1 Class A scheduled position report 2 Class A assigned scheduled position report 3 Class A special report; interrogation response

5 Class A static report

18 Class B scheduled position report (cf 1,2,3) 19 Class B extended position report

24 Class B static report

27 Class A long range scheduled position report

Signal Modulation The AIS signals satisfy several requirements. The bandwidth of the signals is small so that they occupy a minimal part of the VHF spectrum and their sidelobes are also small so as to cause minimal interference with signals in neighbouring channels. To minimize costs, the performance is not sensitive to non-linearities in the system and it is possible to receive the information

6 Non-Return-to-Zero differential line code; a transition occurs when a logical zero is sent but not a logical one. 7 Gaussian Minimum Shift Keying.

6 DRDC Atlantic CR 2011-152

without regenerating a carrier. These requirements are met by using Gaussian Minimum Shift Keying (GMSK).

There is a hierarchy of modulation schemes but the requirement of lack of sensitivity to non- linearities limits consideration to phase and frequency modulations. The requirement to demodulate the signal without reconstructing a carrier tends to limit the modulation to frequency modulation using a binary shift keying. Differential Phase-Shift Keying (DPSK) would have been an option but the Bit Error Rate (BER) is higher [4]. The current specification requires that the logical data is coded using a Non-Return-to-Zero-Inverted (NRZI) line code and this implies that the data can be detected unambiguously using a simple non-coherent Differential Frequency Shift Keying (DFSK) demodulator. However, the power spectral density of a random signal processed with the NRZI line code does not go to zero at zero frequency. Therefore bit-stuffing is needed to remove the dc component (remove the mean value of the waveform).

A narrow bandwidth suggests continuous phase and Minimum Shift Keying (MSK). At first sight, MSK implies that symbols are transmitted by an up or down shift that introduces or removes exactly one cycle during the time that each symbol is transmitted. This ensures that the phase is continuous and that the signals representing each binary symbol are orthogonal so that they can be recovered easily. If the frequency shift is fm and the inter-symbol time is Tb, this leads to [4]:

f mTb h (2) where h is called the deviation ratio. When h = 1, we have Sunde’s Frequency Shift Keying (FSK). However, it is possible to reduce the bandwidth by a factor of 2 by setting h = ½. In this case neighbouring symbols are not independent but can be grouped in pairs to maintain phase continuity. The signal space diagram that represents MSK now has four message points.

As shown in [4], the error probability for coherent MSK (in which the carrier is reconstructed) is similar to that for Quadrature Phase Shift Keying (QPSK); for Additive White Gaussian Noise (AWGN) it is given by:

1 § E · P erfc¨ b ¸ (3) e ¨ ¸ 2 © N 0 ¹ where Eb is the signal energy per bit and N0/2 is the two-sided noise spectral density. For example, for the available thermal noise power we have the usual formula:

N 0 kT (4) The complementary error function is given by:

u 2 2 erfc(u) 1 e z dz (5) 1/ 2 ³ S 0

DRDC Atlantic CR 2011-152 7

Figure 1: Spectral densities as a function of normalized frequency

The sidelobes of a MSK signal are too large for communications purposes and so the symbol stream is filtered with a Gaussian profile in the frequency domain. This smoothes the modulation and reduces the sidelobes. The spectral densities for various time-bandwidth products are shown in Figure 1. The smoothing is characterized by the time-bandwidth product, WTb, where W is the 3 dB bandwidth.

The probability of an error is also affected by the Gaussian filtering but the reduction of the bandwidth is not very costly. Again the time-bandwidth product can be employed as a parameter [4] and Figure 2 shows the degradation, D, which is defined in units of decibels as:

§D · D 10 log10 ¨ ¸ (6) © 2 ¹ Once the parameter D has been determined from Figure 2 and (6), it can be inserted into a modification of (3):

1 § DE · P erfc¨ b ¸ (7) e ¨ ¸ 2 © 2N 0 ¹

8 DRDC Atlantic CR 2011-152

Figure 2: GMSK degradation.

When coherent demodulation cannot be used because of carrier phase jitter, which can be a result of rapid signal fading, non-coherent modulation can be employed. In this case envelope detectors are part of an optimum receiver and there are only two outcomes, either a logical 0 or 1. The probability of a bit error is now given by [4]:

1 § E · P exp¨ b ¸ b ¨ ¸ (8) 2 © 2N 0 ¹

For example, if WTb = 0.3, the degradation for coherent demodulation is about 0.46 dB and D = 1.8. The BERs for both coherent and (non-differential) non-coherent demodulation for this time- bandwidth product are shown in Figure 3; it is assumed that there is no co-channel interference. The Gaussian filtering makes only a very small difference to the coherent BER. It is obvious that there is a significant advantage in coherent demodulation.

DRDC Atlantic CR 2011-152 9

1.0E+00

1.0E-01

1.0E-02 BER 1.0E-03

1.0E-04

1.0E-05 -5-2.502.557.51012.5

Eb/N0 (dB)

Figure 3: Bit Error Rates for GMSK maximum likelihood coherent demodulation () and non- coherent demodulation ()

Hardware There are various manufacturers of Class A AIS transponders and receivers, such as CNS Systems of Sweden (ICAN in Canada) and Kongsberg of Norway. The ICAN VDL-6000 transponder is made by CNS and retails for $3800 CAD. This receiver is based on software- defined radio and the unit will be “firmware” up-gradable to accommodate the new AIS channels 3 and 4 designed for space-based AIS. Kongsberg Maritime markets the AIS-200 transponder at a cost of $5355 (includes antenna and GPS antenna). This is also software up-gradable.

There are other versions one of which is portable and another is designed for military blue force operations. In the latter case the AIS messages are encrypted to hide the position information from red forces; the encryption is either Advanced Encryption Standard (AES) or Blowfish.

There are a number of low-end manufacturers and their Class A units are not designed for software up-grade. This is because the receiver front end is based on analog circuitry.

According to one manufacturer, non-coherent DFSK is employed in inexpensive AIS receivers though the BER may be quite poor compared with coherent detection. However, the transponders still meet the ITU specification, which only requires a receiver sensitivity of -107 dBm for a Packet Error Rate (PER) of 20%. This manufacturer has tested various off-the-shelf receivers and found sensitivities ranging from -110 dBm to -114 dBm.

10 DRDC Atlantic CR 2011-152

Link Budget The link budget depends on the transmitter power, path length, antenna gains, cable losses, etc. Table 5 provides notional data based on a free space range of 120 km. This is an extreme terrestrial range and has been chosen to illustrate that large losses can be tolerated.

It has been assumed that the ship-board AIS and the shore based antennas are vertical monopoles or dipoles with some gain towards the horizon. Those channels neighbouring the AIS channels are likely to give rise to low levels of interference and a loss of 4 dB has been assumed. In the far range multi-path propagation is likely to give rise to rapid fading and this has been represented by a further loss of 4 dB. The required Signal to Noise Ratio (SNR) has been set at 12.5 dB. This corresponds to non-coherent demodulation with a BER of 10-4. Non-coherent demodulation would be needed in the presence of heavy rapid fading.

The receiver sensitivity in the table is -107 dBm, which is the minimum specified by the ITU. The result is a large margin of 19.8 dB. This suggests that large additional losses can be tolerated for trans-horizon paths or paths in which obstacles are present and signal losses are associated with diffraction.

Table 5: Class A Link Budget Parameters

Parameter Value Range (km) 120 Elevation Angle (deg) 0 Transmit Power (W) 12.5 Ship Antenna Gain (dB) 2 Receiver Antenna Gain (dB) 2 Total Cable Losses 6 Interference Loss (dB) 4 Fading Loss (dB) 4 Required SNR (dB) 12.5 Received Power (dBW) --117.2 Receiver Sensitivity from ITU (dBW) --137.0 Margin (dB) 19.8

DRDC Atlantic CR 2011-152 11

Troposphere Effects on AIS Technology

Several phenomena affecting VHF propagation can be found within the troposphere. Turbulent irregularities in the refractive index can scatter the signal. As the refractive index of air is based on a variety of natural factors, an atmospheric profile results which may cause ducting. Lightning strikes can cause reflections. Precipitation, referred to as hydrometeors, can also attenuate or scatter a signal. All of these phenomena are examined and their effects on VHF propagation, and hence AIS signal transmission, are evaluated.

Tropospheric Scatter VHF energy can be scattered via small-scale irregularities in the troposphere. These irregularities result from small variations in temperature and humidity from the ambient values. When large scale air masses of differing refractivity move alongside each other, a shear results along the interface [2]. Large eddies on the order of 100 m in size will form because of the turbulent mixing of air at the boundary. These large eddies create increasingly smaller eddies by a similar mechanism, until they approach a size on the order of 1 mm, at which point viscosity counterbalances the effect of the turbulence and the spawning process halts [5].

These eddies can defocus and refocus an impinging radio wave, which manifests itself as rapid amplitude and phase fluctuations. For larger aperture antennas this results in a perceived degradation in antenna gain. However, the effects of this phenomenon are only noticeable for a total antenna system gain of greater than 30 dB [6].

The more important effect of tropospheric turbulence is the resultant scatter—called “troposcatter”—when a small portion of the incident beam is deflected and creates a weak signal alongside the main beam. This signal may cause interference to an AIS receiver outside of the SOTDMA cell, or it may allow over-the-horizon transmission. In either case, this weak signal is the subject of an interference prediction.

12 DRDC Atlantic CR 2011-152

Figure 4: Geometry of tropospheric scattering

Tatarski [7] has outlined the underlying theory of radio wave propagation through the turbulent atmosphere. The model has been used to predict backscatter and forward scatter, but in the case of interference prediction in communication systems the theory is not deemed to be accurate enough [5]. Other theoretical models which deal with interference scenarios become very complex [2].

A simpler, semi-empirical model was therefore adopted by the ITU for the purposes of interference prediction. The model chosen for the ITU recommendation ITU-R P.452 was a derivative of the Yeh model. The basic transmission loss in decibels which is not to be exceeded p percent of the time is given as [2]:

Lbs 190 30log( f ) 20log(d) 0.573T 0.15N 0 Lc Ag (9) 10.1(log( p / 50))0.7 where f is the frequency in megahertz, d is the distance in kilometres, ș is the scattering angle in degrees, and N0 is the antenna gain correction factor.

Ag is the gaseous absorption factor calculated according to ITU-R P.676. It is based on a value of 3 grams per cubic metre because, while low, most of the troposcatter path is well above the earth’s surface. Note that for the purposes of AIS frequencies, Ag < 0.005 dB per kilometre, according to the graphs in ITU-R P.676. N0 is the surface refractivity which can be inferred from the map given in the text of the ITU-R recommendation P.452-14.

DRDC Atlantic CR 2011-152 13

For a typical AIS setup, with operating frequencies of 160 MHz, a range of 120 km, and a resulting scattering angle of about 3.5°, the loss which is not to be exceeded 0.01 percent of the time is:

Lbs 190 66.1 41.6 2 48 0.6 25.2 227.1dB (10) Hence this is the minimum loss for 99.99% of the time. For a 12.5W class A AIS transponder, this yields a received power of -216 dBW which is well below the typical AIS receiver sensitivity of -137 dBW. Hence troposcatter is deemed to be of small consequence in extending AIS transmission beyond the horizon.

Refractive Index of Air The basic theory of refractive index can be found in many undergraduate texts. The refractive index is closely related to the relative permittivity of the medium, H, being equal to its square root. Therefore the refractive index can be investigated by applying the methods of electromagnetism to atomic and molecular models. It is customary to begin by studying the relative permittivity in the static or dc (dielectric constant) case because this is simple. It turns out that the value of H-1 for dry air is about 5.75x10-4 and varies by less than 1% over the frequency range from dc to 24 GHz, for example [8]. It is quite close to this value even at optical frequencies.

There are three mechanisms that result in the polarization of air [9]. The first applies to non-polar gases and is due to the distortion of the electron orbits about the nucleus in a small electric field. The distortion gives rise to a dipole moment, p, for each molecule. This dipole moment is proportional to the field, E, at the molecule; in the application to radio propagation the deviations from vacuum permittivity are very small and we can assume that E is just equal to the applied field:

p DE (11) where D is the polarizability of a molecule. It follows that the induced electric moment per unit volume, P, is given by:

P n0ĮE (12) where n0 is the number density of molecules. However, from considerations of electric displacement, we have:

D P İ0 E İİ0 E (13) where H0 is the permittivity of free space. Therefore:

İ 1 n0ĮH 0 (14) When the left hand side is very small, this is identical to the Clausius-Mossotti equation, which applies more generally [8]:

14 DRDC Atlantic CR 2011-152

H 1 n D 0 (15) H 2 3H 0

The atoms and molecules He, H2, O2, N2 and CO2 have zero dipole moments in the absence of an electric field so that (14) applies. However, (14) can be recast using the ideal gas law:

PgV ng RT (16) where Pg is the pressure, V is the volume, ng is the number of moles, R is the gas constant and T is the temperature. For unit volume we have:

ng n0 / N A (17) where NA is Avogadro’s number. Therefore (15) becomes:

H 1 DPg N A DPg (18) H 2 3RTH 0 3kTH 0 where k is Boltzmann’s constant (R = NAk) . It follows that for dry air, the difference between the relative permittivity and vacuum is approximately proportional to pressure and inversely proportional to absolute temperature.

Water, H2O, is a polar molecule because the atoms form a triangle, unlike the previous molecules, which exhibit symmetry in their arrangement. This results in a permanent dipole moment, P. Therefore, when water vapour is introduced into air, there is an additional contribution to the permittivity.

The water vapour molecules with their associated dipoles experience collisions with their neighbours and these tend to randomize their orientation. When a static electric field is applied, there will be a tendency for the dipoles to align with the field. This can be calculated according to statistical mechanics, and the electric moment per unit volume is given by:

n P 2 E P 0 (19) 3kT Adding this contribution to that derived earlier we have for the static permittivity:

2 H 1 pDry e § P · D ¨D ¸ (20) Air ¨ H 2O ¸ H 2 3kTH 0 3kTH 0 © 3kT ¹ where pDry and e are the partial pressures of dry air and water vapour respectively.

In practice the permittivity can be modeled by introducing empirical constants, K1,2,3 [10]:

DRDC Atlantic CR 2011-152 15

pDry e e H 1 K K K (21) 1 T 2 T 3 T 2 In radio propagation studies, it is convenient to work in term of the refractive index, n, rather than the permittivity. Because both the refractive index and the permittivity are close to one, we have:

1 H (1 n 2 ) | 2(1 n) (22) According to [10], this yields the refractivity, N:

6 77.6 § e · N (n 1)10 ¨ p 4810.0 ¸ (23) T © T ¹ where T is in degrees K and both p (the total atmospheric pressure) and e are in millibars. The coefficients are based on 0.03% of carbon dioxide and the equation of state for air rather than the ideal gas law. Smith and Weintraub [10] claim that this expression for the refractivity is accurate to 0.5 per cent for radio frequencies up to 30 GHz within the normal ranges of temperature, pressure and humidity. This equation has been used to model atmospheric refractivity for propagation studies [11]. There is a long history of measurements and the resolution of problems. In particular, no evidence of dispersion for frequencies up to 24 GHz was found by the authors of [12]. Much effort has been spent on estimating the refractive index accurately at optical frequencies, e.g. [13], [14] and [15]. Some of this could be relevant.

It is noted that the proportion of CO2 in the atmosphere has increased to about 0.04% since Smith and Weitraub published their data. This suggests that some minor adjustments to their constants could be advantageous.

The polarization of a gas with polar molecules intimately involves collisions between the molecules. This is because the angular momentum of a molecule is quantized and an applied electric field of practical magnitude cannot excite a molecule out of its ground state. Therefore the orientation of the molecules cannot respond to the field unless a collision takes place. At low frequencies there are many collisions per cycle of the field and re-orientation under the field takes place many times. At angular radio frequencies, Z, of the order of the collision frequency, re- orientation of the molecule cannot occur smoothly and the polarizability will fall. As in the Debye theory of polarizability, the effect can be modeled as a viscous damping with a relaxation time, W. This introduces a factor into the polar component of the induced dipole moment equal to:

1 1 iZ (24) 1 iZW 1 Z 2W 2 This has an imaginary part that represents the radio frequency energy absorbed by the collisions. Simplistic theory suggests that significant roll off in the real parts of the relative permittivity and refractivity could start at microwave frequencies of about 1 GHz but for water vapour this is not observed.

To render (23) useful, it is necessary to express the partial pressure of water vapour in terms of relative humidity, which is the quantity that is readily available in meteorological data. This can

16 DRDC Atlantic CR 2011-152

be implemented using tables or estimated using the Clapeyron equation from thermodynamics. It relates the rate of change of saturated water vapour pressure, es, with temperature to the molar latent heat, L, and the molar volumes, Vv, of the vapour and the liquid, Vl, i..e.

de L s (25) dT T (Vv Vl )

For water Vv >> Vl and applying the ideal gas law yields:

de e L s | s (26) dT RT 2 On the assumption that L does not vary significantly with temperature, this can be integrated giving:

§ L § 1 1 ·· e e exp ¨ ¨ ¸¸ (27) s 0 ¨ ¨ ¸¸ © R © T0 T ¹¹ where e0 is the saturated vapour pressure at T0. For example, the actual water vapour pressure used in [11] is:

§ T 273· e 6.11 H exp¨19.7 ¸ (28) © T ¹ where the units of e are millibars and H is the relative humidity as a fraction. This result is from [16] and it appears to be accurate to better than 1% over the temperature range 0 º to 45 º. Clearly T0 is intended to be 0 ºC at which the saturated vapour pressure is 6.11 mbar (611 Pa). However, equating 0 ºC to 273 ºK rather than 273.16 ºK is not particularly accurate and could lead to small errors. A formula, due to Goff and Gatch is discussed in [17] but the values are not consistent with those of other workers. However, another useful formula for finding the vapour pressure is the Antoine equation [18]; for example the coefficients for a large number of organic liquids is given in [19].

For temperatures below freezing and down to at least -20 ºC, Ciddor [20] recommends the formula, which has a form that is similar to that above (as well as to the Antoine equation):

log10 es 12.537 2663.5/T (29) where es is now in units of Pa. Multiplying es by H gives the actual vapour pressure. It is easily verified that this gives the correct saturated vapour pressure at 0 ºC.

Fig. 1 shows the refractivity as a function of temperature and relative humidity at a pressure of 100 kPa and was calculated using (23) and (28). With high relative humidity, the refractivity rises with temperature because warm air can absorb more moisture.

DRDC Atlantic CR 2011-152 17

600

500

400

300 Refractivity 200

100

0 010203040 ºC

Figure 5: Refractivity of air as a function of temperature for relative humidities of 0% (), 50% (), 80% () and 100% ().

Atmospheric Profile The variation of the refractivity with height determines the propagation of VHF signals in air. This can be measured by a balloon-borne radiosonde. The preferred instrument is a microwave refractometer because this has a frequency response of tens of Hertz. This type of refractometer is based on the resonant frequency of a cavity, which depends on the dielectric constant of the air. Otherwise the pressure, temperature and humidity can be measured and the refractivity calculated using (23).

The profiles are discussed by Hall [16] and Barclay [2]. In temperate regions under normal atmospheric conditions the temperature decreases with height at a rate of about 6.5º/km and the pressure decreases approximately exponentially. This leads to a refractivity that decreases with altitude at a rate of about 40 km-1. Signals are refracted slightly downwards and tend to follow the curvature of the earth. This raises the effective radius of the earth by a factor of approximately 4/3. For lower refractivity decreases, the signals will be refracted away from the “4/3 earth” and this is called “sub-refraction”. For greater decreases, the range of the signals is extended and this is called “super-refraction”.

Under conditions of temperature inversion, the refractivity decreases at a rate of more than about 157 km-1 and ducting can occur. This leads to propagation of VHF signals over very long

18 DRDC Atlantic CR 2011-152

distances because the signals remain trapped within the duct and fall off with distance much less rapidly. Temperature inversion is a situation where the temperature increases with altitude rather than decreases. It can occur when x Warm dry air lies over a cool surface, such as the sea; x The land surface cools under clear skies; x A warm front forces warm air over a cold surface; this is often associated with an anticyclone (high pressure); x Sea breezes undercut warm air over land; and x Cold downdraughts are associated with cumulonimbus clouds as may occur when there are heavy showers or thunderstorms.

It should be noted that a relationship involving the wavelength and the layer thickness must also be satisfied for ducting to occur. The ducts can extend up to 3 km in altitude.

Batianeh and Macario [11] describe a refractivity model in which an isothermal atmosphere is assumed so that the pressure profile is given by:

§ gh · p p0 exp¨ ¸ (30) © RT ¹ where p0 is the pressure at the sea surface, g is the acceleration due to gravity, h is the height and T is the average temperature in the lower troposphere. In their model the temperature normally decreases linearly with height and the combined effect of the pressure and temperature variation is that the refractivity also decreases. Under conditions of temperature inversion, the temperature first increases with height from the surface and then decreases. The humidity also decreases with height and this causes the refractivity to fall with height; the refractivity lapse rate is high and refraction may be increased to an extent where ducting occurs. Details of the model can be found in the paper.

Refraction

In the simplest model, the refractive index of the atmosphere can be described by concentric spherical surfaces of constant index centered on the earth. The phase of an electromagnetic wave passing through one of the surfaces must be continuous through it. If the phase of the wave at the surface of constant refractive index is I:

I Zt k r (31) where Z is the angular frequency, t is the time, k is the angular wavevector and r the position vector along the surface. Therefore the refraction through the surface is described by k r = constant. Because the frequency of the wave is a constant and the velocity of light in the medium is inversely proportional to the refractive index, k is proportional to n. The basic equation is Snell’s law:

nkˆ rˆ const (32)

DRDC Atlantic CR 2011-152 19

where the “hat” indicates a unit vector. In polar coordinates (r,T), this becomes:

ˆ ˆ ˆ nk r rˆ nrk ș ș const (33) However, the first term is zero because the surfaces of constant index are perpendicular to the radial direction. Thus we have [16] the polar coordinate form of Snell’s law:

nr cosD const (34) where D is the angle of the wavevector to the earth’s surface. When the refractive index is a function of height, h, differentiation with respect to r yields:

1 1 dn cosD (35) r n dh If we set D = 0, and the radial coordinate, r, to the earth radius, we can find a refractivity lapse rate that just causes the ray to follow the surface of the earth; it corresponds to about 157 km-1 as already noted and corresponds to ducting.

Software

Some useful information on atmospheric and tropospheric effects at long range is provided in [21]. The formulae for finding the propagation losses under simple atmospheric conditions are provided. This is accompanied by a spreadsheet [22] from which the propagation loss under the refractivity (profile constant over the path) conditions can be estimated8. The spreadsheet estimates the propagation loss statistics for interference signals over land and sea paths. However, the calculations are the same if the “interfering signal” is simply an AIS signal.

AIS receiver characteristics are defined in [1]. This document states that the minimum receiver sensitivity for a 20% Packet Error Rate (PER) is -107 dBm or -137 dBW. This takes into account the thermal and other noise associated with bandwidth of the receiver to achieve the stated PER. The signal power, Pr, received from a transmitter over a free space path is given by:

2 Pt Gt Gr O Pr 2 (36) (4Sr) Lh where Pt is the transmitted power, Gt is the transmitter antenna gain, Gr is the receiver antenna gain, O is the wavelength, r is now the path length and Lh represents other losses in the hardware. In this case the path loss, L, is given by:

O2 L (37) (4Sr) 2

8 Though saved as an “xls” file, the spreadsheet is designed to function using Excel 2007 and has been tested. There seemed to be problems running under Excel 2003. If the antenna heights are set very high and other parameters adjusted appropriately, the program calculates the correct free space propagation loss.

20 DRDC Atlantic CR 2011-152

The spreadsheet provides the path loss in more general circumstances that include refraction due to variations in the atmospheric refractivity and diffraction at the earth’s surface. As an example, consider the data in Table 6 for an ocean area off Vancouver Island

Table 6: AIS Parameters

Parameter Value Transmitter Power (dBW) 11.0 Transmitter Antenna Gain (dBi) 2 Receiver Antenna Gain (dBi) 2 Wavelength (m) 1.86 Path length (km) 120 Average Refractivity Lapse Rate (km-1) 44* Maximum Monthly Mean Refractivity Lapse Rate(km-1) 55* Sea Level surface Refractivity 325* Height of Transmitting Antenna (m) 15 Height of Coastal Receiving Antenna (m) 50 Cable and Other Losses (dB) 3 Percent of Time that Signal Exceeds Calculated Value (%) 10 *From the data in the spreadsheet for Vancouver.

Now entering this data with appropriate latitude and longitude, the spreadsheet gives a propagation loss of 148.9 dB. The loss will be less than this for 10% of the time. Therefore the received power will be greater than 11 + 2 + 2 – 3 – 149 = -137 dBW for 10% of the time. Alternatively the result can be interpreted that the range will be greater than 120 km for 10% of the time. To establish a match to the AIS specifications generally requires some manual iteration in the range or other parameters.

Ray Methods The simplest approach to predicting the propagation characteristics of radio waves is to employ ray theory. This is based on the geometric optics model. In free space the waves propagate in straight lines within ray tubes. The ray tubes vary in cross-section so as to be consistent with the conservation of energy. Ray models are scalar models and cannot properly represent polarization effects though the electric or magnetic vectors of the electromagnetic wave can point in a fixed direction with respect to the ray. However, the theory can be extended to accommodate scattering and polarization changes at discrete points along the ray. Based on geometric optics, ray models cannot easily encompass leakage of waves from the top of a layer or scattering from a rough ocean surface.

A principal effect of the troposphere is to bend the rays in an arc. If this follows the earth’s surface, the ray will appear to propagate parallel to the surface. Therefore a simple coordinate transformation is often used which renders the earth flat. This corresponds to a refractivity gradient of close to -157 km-1. If the refractivity is less than this, the rays in this coordinate system are curved upwards. If it is more than this, the rays are bent downwards and are reflected from the earth’s surface; they tend to be confined in a duct. The model is known as the “flat earth model”.

DRDC Atlantic CR 2011-152 21

In the flat earth model the effect of the coordinate transformation is to change the refractivity from N to a new value M by adding 157 km -1, which is an effective refractivity value for this model.

Applications to AIS must take into consideration that the transmitter and receiver will always be located close to the earth. This implies that elevated ducts should not be important. This simplifies propagation treatments because the height of a ray above the earth’s surface will be confined within narrow limits where the flat earth approximation is valid; no elevation angle corrections are needed.

When the vertical gradient of the refractivity profile is less than that needed for ducting, the rays from a transmitter can be calculated in a trivial manner (especially when the profile is constant along the surface of the earth) and the receiver will either lie on a ray path or not. The amplitude of the received wave can be found by computing rays along a ray tube that passes through the receiver. If ducting occurs, the rays will be refracted and bounce within the duct; they can be computed in the same way though with a little more difficulty.

However, when the receiver is not directly in the path of a ray, such as when it is over the radio horizon, ray methods are not useful because, at least in their simple form, they ignore diffraction and the result is binary. A situation where the signal is simply received or not is not realistic. Therefore, while ray methods can be useful when there is ducting or when there are variations in the vertical refractivity profile along the surface of the earth, the parabolic equation method is preferred.

Ducting A typical duct is of thickness, t = 25 m with a change of į1 = 10 N-units, this represents a gradient of 400 km-1. There is a wavelength cut-off [16] given by:

1/ 2 § GN · 3 / 2 Omax 0.0025 ¨ 0.157¸ t (38) © t ¹ For this example, the cut-off is 0.15 m, which is far less than the AIS wavelength of about 2 m. Therefore the AIS signals will not be able to propagate in such a duct. According to (38), the thickness of the duct would have to be about 150 m with a change in refractivity of 50 N-units. This would occur less frequently.

Rays can only enter a duct in a limited range of angles. This reduces the power that can be transmitted over long distances. Rays propagating at angles greater than a critical angle, șc, will leave the top of the duct and those that start below a critical angle will hit the ground. Hall states that:

6 șc 2ǻ0 u10 (39)

Rays are trapped between ±șc, where the units are radians [16] DQGǻM is the decrease in effective refractivity between the transmitter height and the top of the duct. According to Hall, the critical angle never exceeds 0.5 degrees.

22 DRDC Atlantic CR 2011-152

The additional path attenuation (over and above the free space loss) associated with a duct can be positive or negative and according to Hall is given by:

C1d 10log d Lc (40) where d is the path length (in kilometres) within the duct. At AIS wavelengths, the constant C1 is approximately 0.1 dB/km. The second term is negative because the power falls off more slowly than in free space. For the present purposes, the last term [16] is given by:

10log(2T c /T B ) (41)

+HUHșB is the half power antenna beam width. In the case of an AIS system, where the antennas are almost isotropic the antenna beam width nominally approDFKHVʌUDGLDQV However, estimates in terms of solid angles would be an improvement.

Ducts are produced by evaporation and advection. Tropical heating produces a humid layer over the sea with a drier layer above. This is prevalent in the afternoon with a typical thickness of 15 m. Ducts may be an almost permanent feature of tropical seas. They can also form when hot dry air from land blows over the sea in the evening; the typical thickness is 25 m. Other mechanisms are described in [16].

Parabolic Equation Methods The parabolic equation method is described in [2]; in [23] it is treated in detail. The method can be applied to propagation through the troposphere over sea and land. It is usually applied to two- dimensional problems so that the model for the signal is based on a scalar wave equation so that polarization is either horizontal or vertical. Further simplification is achieved by assuming that the waves propagate close to some preferred direction. In other words the problem involves paraxial waves. Also the tropospheric variations in refractivity are usually handled in such a way as to effectively flatten the earth.

The advantage of the parabolic method is that Fast Fourier transform or finite difference techniques can be employed to render solutions by a marching technique in which solutions are found at increasing range. For applications to propagation over the sea, the method can be implemented on a desktop computer in seconds. The model can include scattering off the rough surface of the sea and the leakage of waves from a tropospheric duct.

Details of the theory and applications are described by Levy [23]; the steps are as follows:

1. Reduce Maxwell’s equations to a second order Helmholtz equation with a refractive index as a function of range and height. 2. Assume an oscillatory variation in the range direction. 3. Assume paraxial propagation in the range direction. 4. Factor the resulting equation into first order forward and backward propagating terms. 5. Approximate a square root operator, often by a simple power series. 6. Approximate propagation by the split-step Fourier method or the finite difference method.

DRDC Atlantic CR 2011-152 23

Levy explains that the first three steps allow the second order differential equation arising from the scalar approximation to Maxwell’s equations (the Helmholtz equation) to be factored into two parts: namely forward and backward waves. While it may be legitimate to ignore the backward wave in some applications, such as propagation over the sea, this is not true generally. Therefore a pair of first order coupled equations must be solved.

When backward waves and edge currents are neglected, Levy shows that the parabolic method applied to a semi-infinite screen in a vacuum reduces to the standard Fresnel diffraction result.

The split-step method handles propagation by separating the medium in the range direction into vacuum segments separated by phase screens. The vacuum regions are sufficiently thin that the refractive index does not vary significantly with range over the width of the segment. Boundary conditions at the top and bottom of the segment must be satisfied. For example, when the sea is smooth and an earth flattening approach is used, the boundary condition over the sea is that the field is zero at the bottom of each segment. At the top of a segment the waves must be outgoing only.

The treatment of propagation over a rough sea is only important when there is at least one reflection off the sea surface. This will be true for any trans-horizon path. Reflection depends on the relative magnitudes of the root mean square sea height and the wavelength. This can be expressed in terms of a Rayleigh roughness parameter, Ȗ given by:

J 4SV sinD / O (42) where ı is the rms sea height, Į is the grazing angle and O is the wavelength; the rms sea height can be expressed entirely in terms of the wind speed. If this is small, perturbation techniques are valid. However, the wavelength of AIS signals is about 2 m, which is of the same order as sea height variations in high sea states. Therefore a roughness reduction factor, ȡ, can be introduced to account for the reduction of the reflection coefficient for a flat surface:

§ J 2 · § J 2 · exp I U ¨ ¸ 0 ¨ ¸ (43) © 2 ¹ © 2 ¹ where I0 is a modified Bessel function.

The parabolic equation method requires the surface impedance for its lower boundary condition. The surface impedance, į, for a plane wave can be expressed in terms of the reflection coefficient, R, which is modified by the reduction factor (as outlined by Levy, p 170):

1 R G sinD (44) 1 R Hybrid models, which combine geometric optics solutions over high altitude paths and parabolic methods, are discussed by Levy but, for the purposes of analyzing AIS signals, the parabolic method is probably the most suitable. Criteria are provided to indicate an appropriate solution domain.

24 DRDC Atlantic CR 2011-152

An example of the use of parabolic methods for propagation over a sea surface is provided in [24]. This was based on a parabolic method with a forward wide angle propagator. A comparison was made between measurements at 818 MHz and simulations. The authors claim that excellent agreement was obtained in one case but in the other the ducting conditions were not properly met and agreement was not as good.

Lightning The literature on the effects of lightning reflections is very limited. There has been anecdotal evidence (found on certain web resources such as [25]) that scattering can occur which result in ranges of up to 700 miles. The ionization that occurs during a lightning strike results in a scattering mechanism similar to that of meteor trails (see section 5.3). The strike must occur roughly halfway between the transmitting and receiving station, as well as during the length of the transmission, in order for interference to result. Hence it is deemed that this interference mechanism will have adverse AIS effects which are unlikely to occur.

Hydrometeors When water condenses in the atmosphere, it can take on a variety of forms: rain, fog, clouds, snow, and hail. Collectively these are known as hydrometeors. Their effect on radio wave propagation is dependent on both the system frequency and the size and type of particle that the wave encounters. [2]

Of the mentioned hydrometeors, rain is the most important in the VHF band [6]. The rain drop size distribution is central to theoretical analysis. The distribution function is denoted N(D)dD , which represents the number of drops with diameters between D and D + dD per cubic metre. The model most commonly chosen is an exponential form:

N(D) N 0 exp(/D) (45) Where lambda is related to the rainfall rate R in millimetres per hour via:

/ 4.1R 0.21 (46) The rainfall rate is also related to the integration of the drop size distribution function [6]:

3 3 R 0.6u10 ³ D V (D)N(D)dD (47) D Where V(D) is the terminal velocity of a drop of size D in metres per second. The Marshall- Palmer distribution yields a typical value for N0 of around 8000 mm per cubic metre [6]. Various other forms of the rain drop size distribution function have been proposed. None are regarded as been consistent with physical reality, but all are considered accurate enough for modeling.

The far field scatter is determined by a scattering function S(ș). For the purposes of forward scatter, S(0) is the parameter of interest. In the Rayleigh scattering region (ʌ' Ȝ), then [2]:

DRDC Atlantic CR 2011-152 25

3 S(0) jPZ P0 / 4Sc (48) where Ȧ = 2ʌI is the radial frequency of the impinging electromagnetic wave, ȝ0 is the permeability of free space, and c is speed of light in a vacuum. The induced dipole moment, P, for a spherical water drop is:

P 3zH 0v (49) where z = (İr – 1)/(İr + 2) is a complex value based on the relative permittivity of water, and v is the volume of the particle. Also of interest is the total extinction cross section of a particle, Cext , which is the portion of incident energy that is absorbed and scattered. Cext is given as

O2 C Re[S(0)] (50) ext S Combining these three equations, the expression for the total extinction cross section becomes:

S 2 D 3 C Im[z] (51) ext 4O Finally, the total attenuation for a path of length L can be found from [2]:

f A 4.34L Cext N(d)dD (52) ³0 Evaluation of this integral for a rainfall of 10 millimetres per hour at a typical AIS frequency of 160 MHz (and a corresponding value of approximately 0.0001 for Im[z] as extrapolated from the graph in [2]) yields an attenuation of approximately 0.000018 dB per kilometre, which is negligible for a typical signal path length on the order of hundreds of kilometres.

Interference from backscattering of a radio wave in a common scattering volume is possible. To find the power received from the backscatter, the radar cross section of a single drop is found to be [6]

2 6 5 O § SD · 2 S 2 6 V d ¨ ¸ z z D (53) S © O ¹ O4 For convenience, a parameter known as the atmospheric reflectivity factor is introduced

Z ND 6 (54) where N is the number of particles per unit volume so that Z has units of mm6/mm3, and hence a multiplying coefficient of 10-18. The cross section per volume is then given as:

5 S 2 V z Z (55) D O4

26 DRDC Atlantic CR 2011-152

Z has been related to the rainfall rate R by this relation from the ITU:

Z 400R1.4 (56) Hence the ratio of received power to transmitted power becomes:

2 P G G O2VV G G S 2VZ z r t r D t r (57) 3 2 2 2 2 2 Pt 64S Rt Rr 64O Rt Rr where V is the common scattering volume between the receiving antenna at a distance of Rr, and 2 transmitting antenna at a distance of Rt. For frequencies up to 100 GHz, |z| = 0.93 to a very good approximation [6].

Note that this expression assumes a constant rainfall throughout the entire common scattering volume. A more accurate representation results when an integration of constituent radar reflectivities is taken over the entire scattering volume. This is further outlined in ITU-R P.452- 14 [26].

It can be inferred from the above equation that the received power decreases with square of increasing wavelength. Barclay summarizes an experiment conducted at 11.2 GHz between Baldock and Chilbolton in the UK – a range of about 131 km. The transmission loss does not exceed 130 dB for 0.001 percent of the time [2]. A typical AIS frequency of 160 MHz has a wavelength which is 72 times greater than at 11.2 GHz. Hence the further transmission loss for AIS frequencies would be 20log(72) = 37 dB more than at 11.2 GHz, or approximately 167 dB. For a 12.5 W class A AIS transponder and isotropic antennas, this corresponds to a received power of -156 dBW which is below the typical AIS receiver sensitivity of -137 dBW.

DRDC Atlantic CR 2011-152 27

Ionosphere Effects on AIS Technology

The ionosphere is a portion of the upper atmosphere which has adequate ionization to affect the propagation of VHF signals. Two regions of the ionosphere—the D region and the sporadic E region—are most likely to cause scattering and reflections. Meteors entering the ionosphere also produce ionized trails which can reflect VHF energy. These effects and their impact on VHF propagation are evaluated in the context of AIS transmission.

D Region The D Region of the ionosphere extends from 50 to 90 km above sea level and is produced by /\PDQĮDQG;-rays. Scatter in the D Region can occur at heights of 70 to 90 kilometres. Upper troposhere turbulence can cause fluctuations in the electron density of this region, which in turn can lead to the scattering of RF energy [27]. Usually the scatter occurs at small angles to the incident beam, and hence is called “forward scatter.” Bailey et al. [28] give the ratio of power received (Pr) to power transmitted (Pt) in the general form:

4 2 P f p A(sin(F / 2)) r m 2 n (58) Pt f l (sin(J / 2)) Where A is the total gain of the antenna system, f is the operating frequency, l is the ray path OHQJWKIURPWKHWUDQVPLWWHUWRWKHLRQRVSKHUHOD\HULQTXHVWLRQȖLVWKHVFDWWHULQJDQJOHDQG is the angle between the incident electric vector and the direction of the scattered wave. The plasma frequency fp is defined in (61) in section 5.3. It is found that for high and mid latitudes, the value of m can vary with time between 7 and 9.5. The value of n can vary between 4 and 12 depending on the season. It should be noted that the received scattered power decays as an inverse to the square of the path length, since the scattering volume increases for an increase in range.

Davies [27] summarizes the long term variations of the received signal strengths presented by Bailey et al [28]. It was found that for signals between 25 MHz and 108 MHz, and for distances between 1000 and 2000 km, the recorded system losses between transmitter and receiver were between 140 dB and 210 dB. As the above equation states, the received power decreases sharply for an increase in frequency.

For the purposes of AIS interference, a particular setup outlined in Bailey et al. [28] is considered. In it, the path length is 1251 km, the system frequency is 107.8 MHz, and a system loss of 115 dB is measured relative to the free space path. The loss is then extrapolated based on (58). A value of m = 7 is chosen for the frequency dependent factor, since it would provide the least amount of further path loss. Adjusting for a typical AIS frequency of 160 MHz, and adding the inverse distance loss for a path length of 300 km yields a further path loss of:

70log(160 /108) 20log(300) 61dB (59) Thus received signals from a 12.5W Class A AIS transponder would have a power level of 11 - 115 – 61 = -165 dBW which is much less than the specified AIS receiver sensitivity of -137

28 DRDC Atlantic CR 2011-152

dBW. It is hence unlikely that scattered AIS signals from the D region would arrive at a high enough power to effectively extend the transmission range.

Sporadic E

The sporadic E region (Es) of the ionosphere is comprised of thin layers (often less than 1 km) of enhanced ionization 100-120 kilometres above sea level. These layers vary greatly by season, geographical location, and time of day – hence sporadic. These layers are dense enough to affect radio wave propagation. In general, there are three types of sporadic E layers [2]: 1) Mid Latitude: caused by wind shears and possibly meteors interacting with the geomagnetic field; 2) Equatorial: instabilities are caused by large electron drift velocities from the electrojet; and 3) Auroral: produced by the precipitation of kilovolt electrons;

The effects of propagation within the sporadic E layer have been deemed to be negligible at frequencies greater than 90 MHz, except as a potential interference source [27]. Further, it has been well documented that the impact of these sporadic E effects decrease with increasing frequency. Hence most of the literature has examined the statistical impact of sporadic E effects at frequencies much lower than the ~160 MHz which an AIS system employs.

For a vertically incident signal to be reflected back to earth, there must be adequate ionization in a particular sporadic E cloud to support a critical frequency fEs. Any frequencies above fEs will pass through the sporadic E layer. A signal obliquely approaching the cloud is found to have a lower equivalent vertically incident frequency by dividing the incident frequency by the secant factor arising from the geometry of the transmitter-receiver setup [2].

Davis et al. [29] examined the sporadic E effect on VHF propagation from Cedar Point to Sterling over a period of four years. The frequencies selected were 27.775 MHz and 49.8 MHz. Plotted are the probabilities that a given received signal intensity will exceed the ordinate. These probabilities are of approximately an order of magnitude lower at 49.8 MHz (between 0.01% and 1% of the time) versus at 27.775 MHz (between 0.1% and 10% of the time).

This is in accordance with the Philips rule, also known as the frequency dependence rule, which states the probability of fEs being greater than the signal frequency f – and hence permitting reflection of the signal back to earth—is related to the signal frequency f (in MHz) by the following [26]

log( p( fEs ! f )) a bf (60) Where the parameters a and b are empirically determined. Dazhang et al. [30] have calculated values for these parameters for 18 different sites in China. It was found that the frequency dependent coefficient b had values between -0.1683 and -0.4943, with an average of approximately -0.2, confirming the rapid drop-off of probability of fEs reflection with increased signal frequency. Substitution of a typical AIS frequency of ~160 MHz into the geometry of the Cedar Point to Sterling setup would yield a likelihood of sporadic E interference approximately

DRDC Atlantic CR 2011-152 29

40,000 times less likely than at 49.8 MHz. Hence sporadic E reflections are insignificant when considering AIS signals.

Meteor Trails A consideration of the scattering of signals at AIS frequencies by meteor trails is important because in principle it can lead to interference from AIS transmissions outside of the SOTDMA cell as well as from land-based radio transmissions. The range can be over 200 km and interference via a single meteor trail might occur for a second or two.

Meteor scattering has been reviewed by Davies [27]. The most important meteors for radio propagation are sporadic meteors, which arrive randomly in time and from random points on the celestial sphere. In contrast, shower meteors tend to arrive at specific times when the earth crosses the orbital path of a debris stream. Shower meteors often seem to radiate from specific constellations, for which they are named. Davies states that reflections are observed at frequencies in the range 30-110 MHz, which is somewhat less than the AIS frequency of about 160 MHz. Therefore, at the outset a serious problem is not expected.

A meteor trail can be regarded as a reflecting cylinder of ionized air at altitudes from about 80 km up to 120 km with a length of up to 50 km and a diameter of several metres. The trail is characterized by the density of free electrons per unit length and this density increases with the original mass of the meteor. The frequency of occurrence of a meteor decreases rapidly with mass so that meteors with a mass greater than 1 gram intersect the earth about 105 times per day and produce a line density of more than 1017 electrons/m.

An ionized trail in the upper atmosphere spreads in radius due to diffusion. The initial radius is of the order of 1 m and expands to tens of metres in a few seconds, depending on the height. If the radio frequency is less than the plasma frequency, the trail scatters like a conducting metallic cylinder because the radio wave cannot penetrate it. This is known as an “over-dense” trail. As the trail expands the scattering increases until the plasma frequency, which is proportional to the electron number density, falls below the radio frequency and the radio waves penetrate the trail. In this latter case, the scattering can be described in terms of individual electron scattering and the trail known as “under-dense”. The scattered amplitude is now reduced because of interference effects in the radial direction over the entire cross-section. The wavelength of AIS signals is about 1.9 m so that interference will be important for under-dense trails of radius greater than 1 m.

The incident and reflected rays make equal angles to a normal to the cylinder and all three vectors must lie in the same plane. Thus geometrical conditions must be met if bistatic communication is to take place. However, these conditions will be met frequently.

The plasma frequency fP depends on the electron density, ne:

2 2 nee f P 2 (61) 4S mH 0

30 DRDC Atlantic CR 2011-152

-19 -31 where e and m are the charge (1.6u10 C) and mass (9.1u10 kg) of an electron and H0 is the permittivity (8.85u10-12 F/m) of free space. Solving this for the electron density using the AIS 14 -3 frequency of about 160 MHz yields ne = 3.2u10 m .

Davies provides a table of meteor frequencies and electron line densities. The electron density just calculated with an initial trail radius of a few metres corresponds to meteors of mass 10-2 g that intercept the earth about 107 times per day. However, most of the trails will not be visible to an AIS receiver and the number in a radius of about 200 km is about 2500 per day.

For the purpose of estimating the signal strength of an AIS transmission, the trail can be considered as an over-dense, perfectly conducting reflector with radius 5 m located at a distance of 100 km from both the AIS transponder and the terrestrial ground station. Apart from a geometrical factor, which must be applied for bistatic communication and which is of the order of unity, the effective length of the trail, L, is equal to one half of the Fresnel zone size, i.e.:

1/ 2 § RO · L | ¨ ¸ (62) © 2 ¹ where R is the distance of the reflection point to the receiver (and transmitter in this case) and O is the AIS wavelength. Therefore, applying the geometrical optics approximation [31], the RCS, V, is given by:

2SaL2 V | SaR (63) O where a is the trail radius. (The definition of RCS is not the same as that in [27] but is consistent with that normally used in radar work such as [32] and [31]; Davies’ version is less by a factor of 4S.) For a = 5 m, this gives a practical upper limit to the RCS of about 1.5u106 m2.

In mobile applications we can assume that the antennas are approximately omni-directional though some small gain can be achieved by restricting the antenna pattern in the vertical direction. Therefore the received power, Pr, is related to the transmitted power by (e.g. [27] and see remark above):

G G P O2V G G P O2 a P r t t r t t (64) r 64S 3 R 4 64S 2 R 3

Assuming isotropic antennas (Gr,t = 1), the received power from a 12.5 W Class A AIS transponder is about -155 dBW. This must be compared with the specified AIS receiver sensitivity of -137 dBW; the signal is 18 dB smaller than the receiver sensitivity.

Therefore under normal circumstances reflections of AIS signals from meteor trails can be ignored unless the meteor is very large. This is quite rare (see Davies [27]). Also the AIS transmission has to occur within the time (one or two seconds) that the meteor trail exists. However, in high shipping densities, the rate of transmission of AIS signals into a single TDMA slot could approach or even exceed one hundred per second. Then meteor scatter would provide a seriously noisy background for a significant fraction of the time.

DRDC Atlantic CR 2011-152 31

Interfering land-based transmissions could be a problem if the transmitter power is of the order of 1 kW or more but this is unlikely because the internationally agreed spectrum allocations do not permit these signals in the AIS band.

32 DRDC Atlantic CR 2011-152

Other Propagation Effects

Other effects which are not native to the troposphere or the ionosphere can also impact VHF propagation. VHF energy can arrive at a receiver via more than one path either constructively or destructively adding, an effect called multipath. As a signal propagates over land, various obstacles along the path can cause diffraction and attenuation. The signal can also be diffracted over a smooth surface such as the sea, allowing it to follow along the curvature of the earth. These various propagation effects are considered in the context of AIS signal transmission and their respective impacts evaluated.

Multipath The signal strength at the receiver is proportional to the transmitted power. For AIS Class A signals the power in the ITU specification is 12.5 W. However, in practice it may be substantially less than this because the transmitter is degraded and the cabling and antenna are corroded by salt water. Therefore some care must be exercised in interpreting observed signal strengths.

When a signal arrives over several different paths, the average power is additive, which is a positive effect. On the other hand there is the possibility of cancellation, which may be sufficiently severe to cause message corruption or signal loss. It should be noted that AIS is designed as a collision avoidance system and that signals are transmitted from a ship under way at intervals of seconds. Therefore AIS will function for the purpose of collision avoidance even with a greatly reduced transmitted power as well as severe multi-path effects. This is not helpful when the objective is reliable reception at long range.

Multi-path causes three effects. These are variations in signal strength at the receiver, changes in phase that may be sufficiently rapid to affect carrier recovery in a coherent receiver and a spread in the information carried by the signal in time (group delays). It is a result of the signal propagating over different paths of different length so that the received complex amplitude is the sum of two or more contributions. These can combine in different ways in the complex plane so as to augment the received amplitude or reduce it to the extent that a signal may fade out entirely.

For AIS, there will typically be propagation paths involving rays reflecting from different parts of the transmitting ship superstructure as well as from the ocean surface. An AIS antenna may take the form of a vertical whip and ideally this would be mounted above and away from other parts of the ship. Unfortunately this is often not the case and the signal may be obscured and, in some directions, may be diffracted around obstructions. Overall this leads to signal losses and multipath propagation.

Consider a receiver that is ship based or shore based. In both cases reflections can occur that contribute to multi-path. In the case of a shore based receiver, efforts are generally made to ensure that the antenna is not obscured and that multi-path propagation is minimized. However, reflections from the sea surface at low angles of depression may be difficult to avoid and reflections from the terrain are possible.

DRDC Atlantic CR 2011-152 33

At grazing incidence, a surface such as the sea may have significant roughness but can act as if it were smooth to the radio waves. Therefore it can act like a highly reflective mirror with only a small fraction of the energy being scattered randomly and most being reflected specularly [2]. This is discussed at length by Hall [16], Chapter 4. It is stated that roughness with a standard deviation in height of V0 introduces a factor, k, reduction in the specular component of the reflected amplitude given by:

2 § 1 ª4SV sinD º · k exp¨ 0 ¸ (65) ¨ « » ¸ © 2 ¬ O ¼ ¹ where D is the grazing angle and O is the electromagnetic wavelength. The scattered or diffuse amplitude component tends to be Rayleigh distributed and its phase is incoherent.

Multi-path propagation can be based on a model in which each path contributes signal amplitude that can be regarded as a vector in the complex plane. Signals from different paths contribute vectors with arbitrary phase. Moreover, as the ship and sea moves, the phase is apt to vary in time. Therefore the model is one in which random vectors change their orientation. This suggests that, at any instant of time and if the number of paths is large, the complex amplitude is approximately normally distributed (“Gaussian”) and the phase is uniformly distributed over the interval [0, 2S]. The phase variations are described in terms of a characteristic or correlation time.

The theory is provided in [4] and [33]. When the complex amplitude is normally distributed, the signal amplitude is Rayleigh distributed and the signal power is exponentially distributed (the exponential distribution is identical to the chi-squared distribution with two degrees of freedom). When the direct signal path dominates the received signals and other paths are relatively small and randomly phased, the received signal is better represented by a constant vector with noise added; a more appropriate distribution is the Ricean distribution. In other cases the Nakagami-m distribution is a better fit to data. The Nakagami-m distribution is related to the gamma distribution in probability theory [34]. It is a one-sided distribution and the parameters are the mean and a shape parameter. This distribution appears not to be grounded in theory but is based purely on the fact that it can be made to fit data. For heavy fading, the log-normal distribution may be used.

Expressions for the distributions are given in [33]. The Rayleigh density is given by:

x § x 2 · p(x) exp¨ ¸ (66) 2 ¨ 2 ¸ V © 2V ¹ where V2 is the variance related to the mean, ȝ, by ȝ = 1.253 V. The exponential probability density, p, is given by:

p(x) exp(x / P) / P (67) The Ricean density for the signal amplitude is:

34 DRDC Atlantic CR 2011-152

x § x 2 Q 2 · § xQ · p(x) exp¨ ¸ I (68) 2 ¨ 2 ¸ 0 ¨ 2 ¸ V © 2V ¹ ©V ¹ Here Ȟ is the amplitude of the signal from the direct path and V2 corresponds to a Rayleigh distributed noisy component from the other paths. The function, I0, is a modified Bessel function of the first kind of order zero. When Ȟ = 0, the Ricean density goes to the Rayleigh density. The problem with the Ricean density is that it requires two parameters, neither of which may be easy to predict.

The Nagakami-m density for the signal amplitude is:

2(m /V 2 ) m x 2m1 § mx 2 · p(x) exp¨ ¸ (69) ¨ 2 ¸ *(m) © V ¹ where V2 is now the mean square amplitude and the parameter m is called the “fading figure”. Again there are two parameters, which may be difficult to predict.

As shown in [4], the effects of phase variations depend on the relative magnitudes of the fading correlation time and either the reciprocal of the channel coherence bandwidth or the symbol duration. If the correlation time is very large, the variations in phase can be regarded as slow and the effect will be the same over the entire frequency spectrum of the signal. Also the phase will be roughly constant over each symbol. This leads to a classification in which there are four categories: Flat-Flat, Frequency-Flat, Time-Flat and Non-Flat.

For AIS signals, which are based on GMSK modulation, the 3 dB bandwidth is significantly smaller than 10 kHz (Graphs for GMSK spectra are provided in [4]). The bit rate is 9600 bit/s so that the symbol interval is about 0.1 ms. Ships travel at velocities up to 10 m/s and the wavelength of AIS transmissions is about 1 m; the Doppler shifts are up to 10 Hz and the correlation time can be expected to be of the order of the reciprocal of this, namely 0.1 s. Therefore we expect multi-path to be Flat-Flat except under special circumstances (e.g. reflections from aircraft, significant tropospheric scatter as described in [21] and [26] as well as high winds and sea states). This implies that fading is usually slow and the principal concern is insufficient received power. The conclusion is consistent with a statement in [21].

The effect of multi-path depends somewhat on the receiver. A table and graphs are provided in [4] and [33] for the coherent and non-coherent FSK systems. It is shown that, in terms of Bit Error Rate (BER), the performance of coherent detection is better than non-coherent detection in the presence of Flat-Flat fading. However, though modern AIS transponders typically use coherent detection, older models are usually based on non-coherent detection9.

In summary, multi-path depends on how a transmitting ship’s antenna is mounted, the ship superstructure, the wind speed, which can drive a thin layer of water at speeds approaching the wind speed, and the receiver configuration, including antenna height and surrounding terrain. The way in which the signals are demodulated in the receiver affects the detected signals. Any

9 From a conversation with the terrestrial AIS transponder design firm Fidus, Ottawa.

DRDC Atlantic CR 2011-152 35

rapid changes in the path characteristics may result in rapid fading, which can be represented as a loss of signal. The best modeling solution is to employ existing simple general algorithms and expressions that are based on practical experience [26].

For example, in [21] the following expression is used for the trans-horizon multi-path loss, L, in decibels:

L 2.6>@1 exp(0.1d) log10 (E / 50) (70) where d is the total path length in kilometres and E is the probability that the loss is not exceeded as a percentage. For example, if d = 50 km or more, the exponential term is negligible and a loss of 1.8 dB will be exceeded for 10% of the time; a loss of 4.4 dB will be exceeded for 1% of the time. This at least provides an order of magnitude of the loss to be expected in spite of the large number of unknowns and parameters that are difficult to predict in practice.

Propagation Over Land The propagation of AIS signals is affected when obstacles obscure the direct line of sight to the ship. However, it may still be possible to receive the AIS signals because they are diffracted around an obstacle, though there may be very large losses. The presence of an obstacle close to the line of sight path can cause a loss but it can also result in an increase in the signal; again this is due to diffraction effects. In the case of land, the obstacle may be a hill and for a ship it could be part of the ship’s superstructure.

There are several models for diffraction. The simplest can be understood in terms of the Huygens construction of wave fronts. This type of approximation is due to Fresnel and is a scalar model of electromagnetic propagation; it is incapable of handling polarization effects. Though it is easy to obtain results for simple geometries, such as an occluding disc or knife edge, it is more difficult to apply it more generally in three dimensions.

Maxwell’s equations, which describe electromagnetic wave propagation, are vector equations. Direct solutions of these equations for practical problems are far too computationally intensive because the region of scattering must be represented to a resolution better than one half wavelength. Therefore approximation is needed. In many cases the vector equations may be reduced to scalar equations. This is valid if the geometry allows the electric or magnetic vector to lie always along one coordinate axis. This applies in two-dimensional problems. In some cases, for example when a scatterer is infinitely conducting, the reflection coefficient is independent of polarization. Then, scalar models can be applied successfully. Problems may occur when the scatterers have finite conductivity and when there is significant cross-polarization of the signals during the scattering process.

The reality is that, when an electromagnetic wave is incident upon a scatterer, the scatterer blocks part of the wavefront and excites currents in it; these latter reradiate energy. The calculation of these currents requires boundary conditions to be solved and this is usually quite difficult, especially if there is a need to obtain results in closed form. Solutions have only been achieved for a few canonical cases, such as the semi-infinite plane and the wedge [35], and then only when the wavelength is small compared to the size of the object. However, this covers many practical

36 DRDC Atlantic CR 2011-152

cases. Small wavelength approximation theory is the domain of ray theory; other models are based on this. One advantage of some ray theories is that they can handle polarization.

The Geometric Theory of Diffraction (GTD) was developed by Keller [36] and is valid as the wavelength goes to zero compared with the other dimensions. The diffracted rays are just like any other rays and the only difficulty is the determination of the initial field at the point of diffraction. The theory provides this as a factor by which the incident ray is multiplied on diffraction. As noted, canonical models are used to assist in representing real objects. These are based on some exact solutions and include the vertex of an edge and the tip of a cone. Unfortunately GTD tends to fail near to the shadow boundary.

The Uniform Theory of Diffraction (UTD) is another ray model, which is more accurate. However, both GTD and UTD can be difficult to apply partly because the scattering factors are quite complicated compared with Fresnel diffraction but mostly because the calculation of all possible ray paths may be difficult. The presence of caustics, where rays coalesce and the signal amplitude goes to infinity, can also cause complications.

In practice, the diffraction of radio signals is often treated with sufficient accuracy by Fresnel diffraction. This is partly because diffraction by terrain, buildings, etc. involves a number of uncertainties, such as the conductivity of the material and the shape of the scatterer, which is usually not known to within a fraction of a wavelength.

A useful review of GTD and some UTD is provided in [37].

Diffraction losses are discussed by Hall [16] and Barclay [2]. It is assumed that the obstacle is in the far field of both transmitting and receiving antennas so that the Fresnel approximation to diffraction applies; the treatment is in terms of Fresnel zones and the simplest case is knife-edge diffraction. Once the geometry has been established, the Fresnel zones, which are contours of constant phase, can be calculated. Determination of the signal strength consists of integrating the complex amplitudes over the aperture left unobscured by the obstacle. The integrals for a straight edge are of the form (Fresnel integrals):

Q § S x 2 · I(Q ) ³ exp¨i ¸ dx (71) 0 © 2 ¹ When Ȟ goes to infinity, the integral I(f) = 0.5 + i0.5; in between -f and f, we have a Cornu spiral. To evaluate the diffraction from a straight edge, which is a two-dimensional problem we need:

f § S x 2 · ³ exp¨i ¸ dx (72) Q © 2 ¹ This yields the familiar diffraction pattern, which exhibits oscillations at points clear of obstruction and tapers off rapidly in the shadow. Near the shadow boundary there is about a 6 dB loss in the power. Therefore it is important in the design of terrestrial communications systems to leave sufficient clearance of the direct path above any obstacle. This needs to take into account

DRDC Atlantic CR 2011-152 37

any effects of tropospheric refraction effects, such as ducting. In practice the loss, L, in the region of attenuation can be approximated by [2]:

2 (73) L 6.9 20.0log10 (Q 0.1) 1Q 0.1

For example, if the path from transmitter to hilltop is of length d1 and that from hilltop to receiver is d2, and the hill is represented as a knife edge of height h above the direct path, and the loss is given by (73), then Ȟ is given by:

2 § 1 1 · Q h ¨ ¸ (74) O © d1 d 2 ¹ It is pointed out by Barclay [2] that the clearance required to avoid attenuation from diffraction on point-to-point links corresponds to a value of Ȟ = -0.85.

When there is a sequence of diffracting obstacles, the signal may be estimated using a method of Deygout [38]. This method starts by identifying the most prominent obstacle that gives the greatest value of Ȟ. Diffraction loss is calculated for this obstacle alone. Then the path is subdivided for the next most prominent obstacle and the process is repeated to determine a further loss, which is added (in decibels) to the first. This is then repeated until all obstacles are taken into account.

Diffraction Over Sea On VHF trans-horizon paths, diffraction takes place over the earth’s surface. This extends the path beyond the cut-off implied by geometric optics. The extent to which the path length is increased depends on the signal power, the curvature of the earth, tropospheric refraction, the wavelength and the sensitivity of the receiver. The topic has been discussed by Ekstrom [39] and is applicable to smooth terrain, such as the sea. The loss is expressed as a reduction factor, F, on the electric field.

Normalized distances X and Z are defined by:

1/ 3 § 2S · 2 / 3 X 0.7937d ¨ ¸ kre © O ¹ (75) 2 / 3 § 2S · 1/ 3 Z 1.26h ¨ ¸ kre © O ¹ where d is the distance, h is the antenna height, O is the wavelength, re is the earth radius, all in metres, and k is the refractive-gradient earth-radius factor, which is nominally 4/3.

The reduction factor is given by:

F V (X ) |U (ZT ) || U (Z R ) | (76) where the subscripts T and R denote transmitter and receiver respectively and:

38 DRDC Atlantic CR 2011-152

V (X ) 2 SX exp 2.025X 1 2 |U (Z) | Z 1 sinI (hl) 2 hl 2S p (77) l O 1/ 2 2S >@(H 1) 2 (60VO) 2 p b O >@H 2 (60VO) 2 Here b = 0 for horizontal polarization and b = 1 for vertical polarization. Also İ is the relative permittivity of the surface material (typically the sea), ı is its conductivity in mho/m and:

S 1 H 1 I tan 1 horizontal polarization 4 2 60VO (78) 5S 1 H 1 I tan 1 vertical polarization 4 2 60VO The range of validity is limited to distances larger than that of the sum of the radio horizons, taking into account the curvature of the earth modified by refractivity gradients. This propagation distance according to geometric optics and taking into account the refractivity is given by:

dGO 2kre hT hR (79) Using the parameters in Table 7, it turns out to be 38.6 km. There are also some other formal conditions that are usually satisfied for AIS signals but this should be verified by referring to [39].

Figure 6 shows the loss as a function of range for AIS signals as a function of range calculated from the above equations10. AIS signals are vertically polarized, the refractivity coefficient for the earth radius is assumed to have the standard value and the heights of the antennas are 15 m for the ship and 30 m for the shore based receiver. The parameters, some of which are from [39], are provided in Table 7. Table 7: Parameters for Diffraction Loss

Polarization Vertical Receiving Antenna Height (m) 30 Transmitting Antenna Height (m) 15 Earth Radius (km) 6400 Refractivity Coefficient, k 4/3 Relative Permittivity 81 Conductivity (mho/m) 4 Frequency (MHz) 162 Transmit Power (W) 12.5

10 Numerical examples in [39] were used to verify a spreadsheet calculation.

DRDC Atlantic CR 2011-152 39

-40

-45

-50

-55

-60 Diffraction Loss (dB)

-65

-70 40 50 60 70 80 90 100 Range (km)

Figure 6: Loss due to refraction around the earth

The minimum sensitivity of an AIS receiver, as specified by the ITU, is -107 dBm. The power, PR, received from a Class A transmitter at an unobstructed range r is given by:

G G P O2 P T R T (80) R (4Sr) 2 where GT,R are the antenna gains. Therefore when the antennas are isotropic, the unobstructed received power at a range of 50 km would be about -69 dBm. This is reduced by diffraction to about -112 dBm and so would be just below the nominal detection threshold of the receiver. Therefore we can conclude that diffraction over the sea around the curvature of the earth is significant in extending the range of AIS signals. It should be taken into account in AIS signal modeling.

40 DRDC Atlantic CR 2011-152

ITU Model Applied to AIS

The ITU has published a model for VHF propagation in the form of a spreadsheet containing Visual Basic codes. The model implements the ITU Recommendation ITU-R P.452-14 [26]. This can be applied to AIS signals though it is aimed at predicting the interference from extraneous signals.

The recommendation applies to signals at frequencies above 100 MHz and permits the propagation losses to be estimated over line of sight as well as trans-horizon paths. It includes diffraction over a smooth earth, diffraction over obstacles, tropospheric scatter, ducting (including elevated ducts) and hydrometeor scatter.

Predictions can be made for yearly average propagation or worst month. A key parameter is the probability that the loss will not be exceeded. This is the same as the probability that the signal strength of a desired signal is greater than the calculated value. Alternatively, the calculations can be interpreted in terms of the probability that a signal can be received from a certain range, though some iteration may be needed.

The basic input data, which is entered in Step 1, is shown in Table 8. The inputs are in the yellow boxes and refer to AIS signals off the east coast of Canada. Table 8: Input Data

User Preferred Parameter input resolution Description f 0.161 0.01 Frequency (GHz) Required time percentage(s) for which the p 10 0.001 calculated basic transmission loss is not exceeded (%) Latitude of transmitting (interfering) station ĳ 45.0 0.001 t (degrees) Longitude of transmitting (interfering) station ȥ -62.0 0.001 t (degrees) Latitude of receiving (interfered-with) station ĳ 45.0 0.001 r (degrees) Longitude of receiving (interfered-with) station ȥ -63 0.001 r (degrees) Transmitting antenna centre height above ground h 15 1 tg level (m) Receiving antenna centre height above ground h 15 1 rg level (m) Transmitting antenna gain in the direction of the Gt 2 0.1 horizon along the great-circle interference path (dBi) Receiving antenna gain in the direction of the Gr 2 0.1 horizon along the great-circle interference path (dBi)

DRDC Atlantic CR 2011-152 41

The latitudes and longitudes are only used to estimate the variability of the refractive index lapse rate and do not determine the range of the signal. Therefore they do not have to be specified accurately.

Following this in Step 2, it is necessary to specify whether the calculation is for a yearly average loss or for a worst month loss. If the latter is required, it is necessary to enter the fraction of the path that is over water.

In Step 3, two parameters are entered. The first is the average refractive index lapse rate through the lowest 1 km of the atmosphere. This can be established by referring to Figure 11 of the spreadsheet, which provides a world map with contours of average refractivity lapse rate. Alternatively the maximum mean values of refractive index lapse rate are in Figure 12 for the worst month predictions. The second parameter is the surface refractivity and this can be found in Fig. 13.

In Step 3b, data is entered to resolve an ambiguity in the theory. Entries are only needed when the path is over water, which will be the case when the application is to AIS. However, because paths of less than 5 km in length are of little interest here, a large value (e.g., 500 km) can be entered into the first box and this can serve as a default value.

Step 4 is the terrain profile analysis. The first column represents range. It is important to note that the final row in this sheet gives the ground range of the transmitter. The second column represents the height of the terrain over the mean sea level. For paths over the sea, this takes the value zero. The third column represents the radio-climatic zone as shown in Table 9.

Table 9: Radio-climatic zones

Zone Type Code Definition Coastal Land A1 Coastal land and shore areas, i.e., land adjacent to the sea up to an altitude of 100 m relative to mean sea or water level, but limited to a distance of 50 km from the nearest sea area. Where precise 100 m data are not available an approximate value, i.e. 300 ft, may be used Inland A2 All land, other than coastal and shore areas defined as “coastal land” above Sea B Seas, oceans and other large bodies of water (i.e. covering a circle of at least 100 km in diameter)

The test profile sheet in the spreadsheet should be examined to verify that all data has been entered correctly and that the numbers are reasonable.

As an example we can use the data in Table 8 to calculate the loss over the sea over a range of 120 km. For 10% of the time, the loss will not be any greater than what is calculated. This entails using 120 entries in Step 4 spaced at intervals of 1 km with heights of zero. The radio- climatic zone is ‘B’ for all entries in the third column. When the Calculate button is clicked, the average yearly loss is 150.5 dB. If the worst month analysis is needed, Step 3 is changed and the

42 DRDC Atlantic CR 2011-152

fraction of the path over sea is entered as ‘1’. The result is 141.3 dB. This is a smaller loss and indicates a higher signal level by 9.2 dB.

The analysis of AIS can be completed by comparing the received signal strength with the specified AIS receiver sensitivity of -107 dBm. The path loss calculated by the spreadsheet, Lp, is related to the received signal strength, PR, by:

PR GRGT PT L p (81) where PT is the transmitter power and GR,T are the antenna gains. If the antenna gains are each 2 dB, the transmitter power is 12.5 W (41 dBm) and the path loss is 150.5 dB, the received signal is -105.5 dBm. This is higher than the lower limit of -107 dBm and therefore the AIS signal will be detectable for over 10% of the time even over a path length of 120 km.

The Visual Basic codes are visible by typing (ALT)-(F11) and may be copied. Otherwise access to the spreadsheet is restricted by password protection.

DRDC Atlantic CR 2011-152 43

Conclusions

The literature review has summarized a number of phenomena which may affect VHF propagation, and hence the transmission of AIS signals. These phenomena can have the effect of producing interference signals, and can also have the effect of extending the signal range beyond the line-of-sight.

Three phenomena were determined most likely to have a significant effect on AIS transmission. x Diffraction over the sea around the curvature of the earth, as outlined in section 6.3, is significant in extending the range of AIS signals and should be considered. x Ducting resulting from the varying refractivity of air, as outlined in section 4.2, can also extend the transmission range. x Multipath effects can cause significant variability in the received signal strength. These effects are principal components of the ITU Recommendation ITU-R P.452-14 [26].

Two other components of ITU-R P.452-14—tropospheric scatter and scatter from hydrometeors—were not found, based on the calculations in section 4.1 and section 4.4 respectively, to have a significant effect in extending the transmission range of typical AIS signals.

The other effects considered, notably scattering resulting from ionospheric layers, meteor trails, and lightning, were deemed to be unlikely to either cause interference or extend the transmission range of AIS signals.

Based on these findings, it is recommended that the software implementation of ITU-R P.452-14, available in [22], be considered for modelling the mentioned VHF propagation effects in the context of AIS transmissions.

44 DRDC Atlantic CR 2011-152

References .....

[1] "Technical Characteristics for an Automatic Identification System using Time-Division Multiple Access in the VHF Maritime Mobile Band," International Telecommunications Union 2010. [2] L. Barclay, Propagation of Radio Waves, 2nd ed. London, United Kingdom: The Institution of Electrical Engineers, 2003. [3] A. Harati-Mokhtari, et al., "Automatic Identification System (AIS): A Human Factors Approach," Journal of Navigation, vol. 60, pp. 373-389, 2007. [4] S. Haykin, Communications Systems, 4th ed.: John Wiley & Sons, Inc., 2001. [5] M. Hall and L. Barclay, Radiowave Propagation. London, United Kingdom: Peter Peregrinus Ltd., 1989. [6] L. Boithias, Radio Wave Propagation. London, United Kingdom: North Oxford Academic Publishers Ltd, 1987. [7] V. I. Tatarski, Wave Propagation in a Turbulent Medium. New York: McGraw-Hill, 1961. [8] B. I. Bleaney and B. Bleaney, Electricity and Magnetism: Oxford University Press, 1959. [9] P. Debye, Polar Molecules: Dover, 1929. [10] E. K. Smith and S. Weintraub, "The Constants in the Equation for Atmospheric Refractive Index at Radio Frequencies," Proceedings of the IRE, pp. 1035-1037, 1953. [11] M. H. Bataineh and C. V. Macario, "Modelling Refractivity Variation in the VHF/UHF Bands," Communications, vol. 1, pp. 267-271, 1996. [12] G. Birnbaum and S. K. Chatterjee, "The Dielectric Constant of Water Vapor in the Microwave Region," Journal of Applied Physics, vol. 23, pp. 220-223, February 1952. [13] B. Edlén, "The refractive index of air," Metrologia, vol. 1, pp. 71-80, 1966. [14] K. P. Birch and M. J. Downs, "An Updated Edlen Equation for the Refractive Index of Air," Metrologia, vol. 30, pp. 155-162, 1993. [15] K. P. Birch and M. J. Downs, "Correction to the Updated Edlen Equation for the Refractive Index of Air," Metrologia, vol. 31, pp. 315-316, 1994. [16] M. P. M. Hall, Effect of the Troposphere on Radio Communications. London, United Kingdom: Peter Peregrinus Ltd., 1979. [17] W. K. Prusaczyk, "Precise Water Vapor Pressure Value Calculations," Computers and Biomedical Research, vol. 19, pp. 129-130, 1989. [18] R. Reid, et al., The Properties of Gases and Liquids, 4th ed.: McGraw-Hill, 1987. [19] C. L. Yaws and H. C. Yang, "To Estimate Vapor Pressure Easily," Hydrocarbon Processing, vol. 68, pp. 65-68, 1989. [20] P. E. Ciddor, "Refractive Index of Air: New Equations for the Visible and Near Infrared," Applied Optics, vol. 35, pp. 1566-1573, March 1996. [21] "Propagation Prediction Techniques and Data Required for the Design of Trans-Horizon Radio-Relay Systems," International Telecommunications Union ITU-R P.452-14, 1992. [22] ITU. Point to Point (Interference) Propagation (Rec. P452). Available: http://www.itu.int/ITU-R/index.asp?category=study-groups&rlink=rsg3-software- ionospheric&lang=en [23] M. Levy, Parabolic equation methods for electromagnetic wave propagation. London, United Kingdom: The Institution of Electrical Engineers, 2000.

DRDC Atlantic CR 2011-152 45

[24] N.-H. Jeong and J.-K. Pack, "Statistical Modeling of Atmospheric Refractivity for Ducting Channel from Meteorological Observation Data in Marine Environments," presented at the 2003 Asia-Pacfic Conference on Applied Electromagnetics (APACE 2003), Shah Alam, Malaysia, 2003. [25] G. Hauser. Propagation. Available: http://www.anarc.org/wtfda/propagation.htm [26] "Prediction Procedure for the Evaluation of Interference between Stations on the Surface of the Earth at Frequencies above about 0.1 GHz," International Telecommunications Union ITU-R P.452-14, 2009. [27] K. Davies, Ionospheric Radio. London, United Kingdom: Peter Peregrinus Ltd., 1990. [28] D. K. Bailey, et al., "Radio Transmission at VHF by Scattering and Other Processes in the Lower Ionosphere," Proceedings of the IRE, vol. 43, pp. 1181-1230, 1955. [29] R. M. Davis, et al., "Sporadic E at VHF in the USA," in Proc. IRE, 1959, p. 762. [30] H. Dazhang, et al., "Model of Es Occurrence Probability in China at 1985-2006," in 4th International Symposium on Electromagnetic Compatibility, 2007. [31] L. V. Blake, Radar Range Performance: Artech House, 1986. [32] E. F. Knott, et al., Radar Cross Section: Artech House, 1985. [33] J. G. Proakis, Digital Communications, 4th ed.: McGraw-Hill, 2001. [34] W. Feller, An Introduction to Probability Theory and Its Applications, Vol. II: J. Wiley & Sons, Inc., 1971. [35] R. G. Kouyoumjian, "Asymptotic High-Frequency Methods," Proc. IEEE, vol. 53, pp. 864-876, August 1965. [36] J. B. Keller, "Geometrical Theory of Diffraction," J. Opt. Soc. America, vol. 52, pp. 116- 130, 1962. [37] R. C. Hansen, Geometrical Theory of Diffraction: IEEE Press, 1981. [38] J. Deygout, "Multiple Diffraction of Microwaves," IEEE Trans. Antennas Propagation, vol. AP-14, p. 480, 1966. [39] J. L. Ekstrom, "VHF-UHF Propagation Performance Predictions for Low Altitude Communication Links Operating Over Water," in Proceedings of the IEEE Mililtary Communications Conference, 1993, pp. 605-608.

46 DRDC Atlantic CR 2011-152

List of symbols/abbreviations/acronyms/initialisms

AES Advanced Encryption Standard AIS Automatic Identification of Ships AWGN Additive White Gaussian Noise BER Bit Error Rate CISTI Canada Institute for Scientific and Technical Information CSTDMA Carrier Sense Time Domain Multiple Access DFSK Differential Frequency Shift Keying DND Department of National Defence DPSK Differential Phase Shift Keying DRDC Defence Research & Development Canada DRDKIM Director Research and Development Knowledge and Information Management ETA Estimated Time of Arrival FSK Frequency Shift Keying GMSK Gaussian Minimum Shift Keying GTD Geometric Theory of Diffraction IALA International Association of Lighthouse Authorities IEEE Institute of Electrical and Electronics Engineers IMO International Maritime Organization ITU International Telecommunications Union LRDC London Research and Development Corporation MMSI Maritime Mobile Service Identity MSK Minimum Shift Keying NRZI Non Return to Zero Inverted PER Packet Error Rate QPSK Quadrature Phase Shift Keying R&D Research & Development RCS Radar Cross Section RF Radio Frequency RFP Request For Proposal

DRDC Atlantic CR 2011-152 47

SNR Signal-to-Noise Ratio SOLAS Safety Of Life At Sea SOTDMA Self-Organized Time Domain Multiple Access TDMA Time Domain Multiple Access UTD Uniform Theory of Diffraction VHF Very High Frequency

48 DRDC Atlantic CR 2011-152

Distribution list

Document No.: DRDC Atlantic CR 2011-152

LIST PART 1: Internal Distribution by Centre 1 F. DESHARNAIS 1 M. MCINTYRE 1 T. HAMMOND 2 A.W. ISENOR (1 CD, 1 hard copy) 1 S. WEBB 1 A. MACINNIS 1 D. SCHAUB 1 L. LAPINSKI 3 DRDC Atlantic Library (1 hard copy, 2 CDs)

12 TOTAL LIST PART 1

LIST PART 2: External Distribution by DRDKIM

1 Library and Archives Canada, Atten: Military Archivist, Government Records Branch

1 NDHQ/DRDKIM 2-2-5

1 Stéphane Paradis DRDC Valcartier 2459 Pie-XI Blvd North, Québec, QC G3J 1X5

1 Andrew Wind, DRDC CORA MARLANT HQ 3rd Floor, Room 311 PO Box 99000 Stn Forces, Halifax, NS B3K 5X5

1 C-CORE 1 Morrissey Road, St. John's, NL A1B 3X5 Attn: Chris Fowler

1 Commanding Officer TRINITY JOSIC, PO Box 99000 Stn Forces, Halifax, NS B3K 5X5

1 Commanding Officer OIC Maritime Operations Centre, JTFP HQ MARPAC/JTFP, PO Box 17000 Station Forces, Victoria, BC B9A 7N2

7 TOTAL LIST PART 2

19 TOTAL COPIES REQUIRED

DRDC Atlantic CR 2011-152 49

This page intentionally left blank.

50 DRDC Atlantic CR 2011-152

DOCUMENT CONTROL DATA (Security classification of title, body of abstract and indexing annotation must be entered when the overall document is classified) 1. ORIGINATOR (The name and address of the organization preparing the document. 2. SECURITY CLASSIFICATION Organizations for whom the document was prepared, e.g. Centre sponsoring a (Overall security classification of the document contractor's report, or tasking agency, are entered in section 8.) including special warning terms if applicable.)

C-CORE UNCLASSIFIED 1 Morrissey Road (NON-CONTROLLED GOODS) St. John's, NL DMC A A1B 3X5 REVIEW: *&(& JUNE 2010

3. TITLE (The complete document title as indicated on the title page. Its classification should be indicated by the appropriate abbreviation (S, C or U) in parentheses after the title.)

VHF Propagation Study

4. AUTHORS (last name, followed by initials – ranks, titles, etc. not to be used)

Green D.; Tunaley J. K. E.; Fowler C.; Power D.

5. DATE OF PUBLICATION 6a. NO. OF PAGES 6b. NO. OF REFS (Month and year of publication of document.) (Total containing information, (Total cited in document.) including Annexes, Appendices, etc.) September 2011 62 39

7. DESCRIPTIVE NOTES (The category of the document, e.g. technical report, technical note or memorandum. If appropriate, enter the type of report, e.g. interim, progress, summary, annual or final. Give the inclusive dates when a specific reporting period is covered.)

Contract Report

8. SPONSORING ACTIVITY (The name of the department project office or laboratory sponsoring the research and development – include address.)

Defence R&D Canada – Atlantic 9 Grove Street P.O. Box 1012 Dartmouth, Nova Scotia B2Y 3Z7

9a. PROJECT OR GRANT NO. (If appropriate, the applicable research 9b. CONTRACT NO. (If appropriate, the applicable number under and development project or grant number under which the document which the document was written.) was written. Please specify whether project or grant.)

11hl; 11ho W7707-115279

10a. ORIGINATOR'S DOCUMENT NUMBER (The official document 10b. OTHER DOCUMENT NO(s). (Any other numbers which may be number by which the document is identified by the originating assigned this document either by the originator or by the sponsor.) activity. This number must be unique to this document.)

R-11-020-868 DRDC Atlantic CR 2011-152

11. DOCUMENT AVAILABILITY (Any limitations on further dissemination of the document, other than those imposed by security classification.)

Unlimited

12. DOCUMENT ANNOUNCEMENT (Any limitation to the bibliographic announcement of this document. This will normally correspond to the Document Availability (11). However, where further distribution (beyond the audience specified in (11) is possible, a wider announcement audience may be selected.))

Unlimited

13. ABSTRACT (A brief and factual summary of the document. It may also appear elsewhere in the body of the document itself. It is highly desirable that the abstract of classified documents be unclassified. Each paragraph of the abstract shall begin with an indication of the security classification of the information in the paragraph (unless the document itself is unclassified) represented as (S), (C), (R), or (U). It is not necessary to include here abstracts in both official languages unless the text is bilingual.)

14. KEYWORDS, DESCRIPTORS or IDENTIFIERS (Technically meaningful terms or short phrases that characterize a document and could be helpful in cataloguing the document. They should be selected so that no security classification is required. Identifiers, such as equipment model designation, trade name, military project code name, geographic location may also be included. If possible keywords should be selected from a published thesaurus, e.g. Thesaurus of Engineering and Scientific Terms (TEST) and that thesaurus identified. If it is not possible to select indexing terms which are Unclassified, the classification of each should be indicated as with the title.)

VHF; AIS; Automatic Identification System; Very High Frequency

This page intentionally left blank.

Recommendation ITU-R P.617-4 (12/2017)

Propagation prediction techniques and data required for the design of trans-horizon radio-relay systems

P Series Radiowave propagation

ii Rec. ITU-R P.617-4

Foreword

The role of the Radiocommunication Sector is to ensure the rational, equitable, efficient and economical use of the radio- frequency spectrum by all radiocommunication services, including satellite services, and carry out studies without limit of frequency range on the basis of which Recommendations are adopted. The regulatory and policy functions of the Radiocommunication Sector are performed by World and Regional Radiocommunication Conferences and Radiocommunication Assemblies supported by Study Groups.

Policy on Intellectual Property Right (IPR)

ITU-R policy on IPR is described in the Common Patent Policy for ITU-T/ITU-R/ISO/IEC referenced in Annex 1 of Resolution ITU-R 1. Forms to be used for the submission of patent statements and licensing declarations by patent holders are available from http://www.itu.int/ITU-R/go/patents/en where the Guidelines for Implementation of the Common Patent Policy for ITU-T/ITU-R/ISO/IEC and the ITU-R patent information database can also be found.

Series of ITU-R Recommendations (Also available online at http://www.itu.int/publ/R-REC/en)

Series Title BO Satellite delivery BR Recording for production, archival and play-out; film for television BS Broadcasting service (sound) BT Broadcasting service (television) F Fixed service M Mobile, radiodetermination, amateur and related satellite services P Radiowave propagation RA Radio astronomy RS Remote sensing systems S Fixed-satellite service SA Space applications and meteorology SF Frequency sharing and coordination between fixed-satellite and fixed service systems SM Spectrum management SNG Satellite news gathering TF Time signals and frequency standards emissions V Vocabulary and related subjects

Note: This ITU-R Recommendation was approved in English under the procedure detailed in Resolution ITU-R 1.

Electronic Publication Geneva, 2017

RECOMMENDATION ITU-R P.617-4

Propagation prediction techniques and data required for the design of trans-horizon radio-relay systems (Question ITU-R 205/3) (1986-1992-2012-2013-2017)

Scope This Recommendation contains a propagation prediction method for the planning of trans-horizon radio-relay systems.

Keywords Anomalous/layer-reflection, diffraction, trans-horizon, tropospheric scatter

The ITU Radiocommunication Assembly, considering a) that for the proper planning of trans-horizon radio-relay systems it is necessary to have appropriate propagation prediction methods and data; b) that methods have been developed that allow the prediction of most of the important propagation parameters affecting the planning of trans-horizon radio-relay systems; c) that as far as possible these methods have been tested against available measured data and have been shown to yield an accuracy that is both compatible with the natural variability of propagation phenomena and adequate for most present applications in system planning, recommends that the prediction methods and other techniques set out in Annex 1 be adopted for planning trans-horizon radio-relay systems in the respective ranges of parameters indicated.

Annex 1

1 Introduction The only mechanisms for radio propagation beyond the horizon which occur permanently for frequencies greater than 30 MHz are those of diffraction at the Earth’s surface and scatter from atmospheric irregularities. In addition propagation due to ducting or layer-reflection may occur occasionally. Attenuation for diffracted signals increases very rapidly with distance and with frequency, and the anomalous propagation probability is relatively small, eventually the long term principal mechanism is that of tropospheric scatter. These mechanisms may be used to establish “trans-horizon” radiocommunication. Because of the dissimilarity of the three mechanisms it is necessary to consider diffraction, ducting/layer reflection and tropospheric scatter paths separately for the purposes of predicting transmission loss and enhancements. 2 Rec. ITU-R P.617-4

This Annex relates to the design of trans-horizon radio-relay systems. One purpose is to present in concise form simple methods for predicting the annual and worst-month distributions of the total transmission loss due to tropospheric scatter and ducting/layer reflection, together with information on their ranges of validity. Another purpose of this Annex is to present other information and techniques that can be recommended in the planning of trans-horizon systems.

2 Integral digital products Only the file versions provided with this Recommendation should be used. They are an integral part of the Recommendation. Table 1 gives details of the digital products used in the method.

TABLE 1 Digital products

Filename Ref. Origin Latitude (rows) Longitude (columns)

First row Spacing Number First col Spacing Number (ºN) (degrees) of rows (ºE) (degrees) of cols DN50.txt Att.1 Annex 1 P.452 90 1.5 121 0 1.5 241 N050.txt Att.1 Annex 1 P.452 90 1.5 121 0 1.5 241 The “First row” value is the latitude of the first row. The “First col” value is the longitude of the first column. The last column is the same as the first column (360° = 0°) and is provided to simplify interpolation. “Spacing” gives the latitude/longitude increment between rows/columns. The files are contained in the Supplement file R-REC-P.617-4-201712-I!!!ZIP.

3 Transmission loss for diffraction paths For radio paths extending only slightly over the horizon, or for paths extending over an obstacle or over mountainous terrain, diffraction will generally be the propagation mode determining the field strength. In these cases, the methods described in Recommendation ITU-R P.526 should be applied.

4 Transmission loss distribution due to tropospheric scatter Signals received by means of tropospheric scatter show both slow and rapid variations. The slow variations are due to overall changes in refractive conditions in the atmosphere and the rapid fading to the motion of small-scale irregularities. The slow variations are well described by distributions of the hourly-median transmission loss which are approximately log-normal with standard deviations between about 4 and 8 dB, depending on climate. The rapid variations over periods up to about 5 min are approximately Rayleigh distributed. In determining the performance of trans-horizon links for geometries in which the tropospheric scatter mechanism is predominant, it is normal to estimate the distribution of hourly-median transmission loss for non-exceedance percentages of the time above 50%. A simple semi-analytical technique for predicting the distribution of average annual transmission loss in this range is given in § 4.1. The method for conversion of these annual time percentages to those for the average worst month is given in § 4.2. Attachment 1 includes additional supporting information on seasonal and diurnal variations in transmission loss, on frequency of rapid fading on tropospheric scatter paths and on transmission bandwidth. Rec. ITU-R P.617-4 3

4.1 Average annual median transmission loss distribution The following step-by-step procedure is recommended for estimating the average annual median transmission loss L(p) not exceeded for percentages of the time p. The procedure requires the link parameters of great-circle path length d (km), frequency f (MHz), transmitting antenna gain Gt (dB), receiving antenna gain Gr (dB), horizon angle t (mrad) at the transmitter, and horizon angle r (mrad) at the receiver:

Step 1: Obtain the average annual sea-level surface refractivity N0 and radio-refractive index lapse-rate dN for the common volume of the link in question using the digital maps of Fig. 1 and Fig. 2, respectively. These maps are available electronically from the ITU-R SG 3 website under the specification in § 2.

FIGURE 1

Average annual sea-level surface refractivity, N0

80 380

60 370

40 360

20 350

u t 0

i t 340

–20 330

–40 320

–60 310

–80 300

–1 50 –1 00 –50 0 50 100 150 Longitude P.0617-0 1 4 Rec. ITU-R P.617-4

FIGURE 2 Average annual radio-refractive index lapse-rate through the lowest 1 km of the atmosphere, dN

80 75

60 70

40 65

60 20

e 55

t 0

a 50

L –20 45

–40 40

–60 35

30 –80 25 –1 50 –1 00 –50 0 50 100 150 Longitude P.0617-02 Step 2: Calculate the scatter angle θ (angular distance) from

  e  t  rmmmmmmmrad (1)

where t and r are the transmitter and receiver horizon angles, respectively, and

3 e  d  10 / kammmmmmmrad (2) with: d : path length (km)

a : 6 370 km radius of the Earth k : effective earth radius factor for median refractivity conditions (k = 4/3 should be used unless a more accurate value is known).

Step 3: Estimate the aperture-to-medium coupling loss Lc from:

Lc = 0.07 exp [0.055(Gt  Gr)]mmmmmmdB (3) where Gt and Gr are the antenna gains. Step 4: Estimate the average annual transmission loss associated with tropospheric scatter not exceeded for p% of the time from:

Lbs p  F 22log f  35log  17log d  L c  Y p dB (4) where (5) F0.18  N0  exp  hsb h  0.23  dN dB Rec. ITU-R P.617-4 5

0.67 0.035N exp h h   log q 50 p  50  00 b     Yq   0.67 (6) 0.035N exp  h h   log 100  q 50 p  50  00 b      1 ℎ = 10−6휃2푘푎 푘푚 (7) 0 8 with:

hs: height of the Earth’s surface above sea level (km)

hb: scale height (km) which can be determined statistically for different climates conditions. For reference purpose a global mean of the scale height may be defined by hb=7.35 km.

4.2 Average worst-month median transmission loss distribution For reasons of consistency with the average annual transmission loss distribution, this distribution is best determined from the average annual distribution by means of a conversion factor. The procedure is as follows: Step 1: If the annual statistics time percentage is given, calculate the time percentage conversion of annual statistics to worst-month statistics for tropospheric scatter from Recommendation ITU-R P.841. If the worst-month time percentage is given, an inversion calculation is needed. Step 2: Calculate the worst-month median transmission loss for the given time percentage, substituting the given or solved annual statistics time percentage into § 4.1.

5 Transmission loss and enhancement distribution due to ducting/layer reflection Ducting and layer reflection may cause an enhancement of the signal which can effect system design. The following calculation is the same as Recommendation ITU-R P.2001-2, Attachment D: Anomalous layer reflection model.

5.1 Characterize the radio-climatic zones dominating the path Calculate two distances giving the longest continuous sections of the path passing through the following radio-climatic zones:

dtm: longest continuous land (inland or coastal) section of the path (km)

dlm: longest continuous inland section of the path (km). Table 2 describes the radio-climatic zones needed for the above classification.

TABLE 2 Radio-climatic zones

Zone type Code Definition Coastal land A1 Coastal land and shore areas, i.e. land adjacent to the sea up to an altitude of 100 m relative to mean sea or water level, but limited to a distance of 50 km from the nearest sea area. Inland A2 All land, other than coastal and shore areas defined as “coastal land” above. Sea B Seas, oceans and other large bodies of water (i.e. covering a circle of at least 100 km in diameter). 6 Rec. ITU-R P.617-4

Large bodies of inland water A “large” body of inland water, to be considered as lying in Zone B, is defined as one having an area of at least 7 800 km2, but excluding the area of rivers. Islands within such bodies of water are to be included as water within the calculation of this area if they have elevations lower than 100 m above the mean water level for more than 90% of their area. Islands that do not meet these criteria should be classified as land for the purposes of the water area calculation. Large inland lake or wet-land areas Large inland areas of greater than 7 800 km2 which contain many small lakes or a river network should be declared as “coastal” Zone A1 by administrations if the area comprises more than 50% water, and more than 90% of the land is less than 100 m above the mean water level. Climatic regions pertaining to Zone A1, large inland bodies of water and large inland lake and wetland regions, are difficult to determine unambiguously. Therefore administrations are invited to register with the ITU Radiocommunication Bureau (BR) those regions within their territorial boundaries that they wish identified as belonging to one of these categories. In the absence of registered information to the contrary, all land areas will be considered to pertain to climate Zone A2. For maximum consistency of results between administrations it is recommended that the calculations of this procedure be based on the ITU Digitized World Map (IDWM) which is available from the BR.

5.2 Point incidence of ducting Calculate a parameter depending on the longest inland section of the path:

 –4 2.41  τ  1  e4.1210  dlm  (8)  

Calculate parameter μ1 characterizing the degree to which the path is over land, given by: 0.2  –dtm  μ  1016– 6.6τ 10–(2.48 1.77)  1   (9)   where the value of μ1 shall be limited to μ11.

Calculate parameter μ4, given by:

 (0.935 0.0176 mn ) logμ1 10 for mn  70 μ4   0.3 logμ1 (10) 10 for mn  70 where φmn is the path mid-point latitude.

The point incidence of anomalous propagation, β0 (%), for the path centre location is now given by:

100.015 mn  1.67μ μ % for   70 β   1 4 mn 0 (11) 4.17μ1 μ4 % for mn  70

5.3 Site-shielding losses with respect to the anomalous propagation mechanism Corrections to transmitter and receiver horizon elevation angles:

gt  0.1dlt (12)

gr  0.1dlr (13) Rec. ITU-R P.617-4 7

where dlt, dlr (km) are the terminal to horizon distances. For LoS paths set to distances to point with largest knife-edge loss The losses between the antennas and the anomalous propagation mechanism associated with site-shielding are calculated as follows. Modified transmitter and receiver horizon elevation angles:

st  t  gt mrad (14)

sr  r  gr mrad (15) Transmitter and receiver site-shielding losses with respect to the duct:

1/2 1/3 Ast  20log1 0.361st  f dlt  0.264st  f dB st>0 (16) A  0 st dB otherwise (17)

1/2 1/3 Asr  20log1 0.361sr  f dlr  0.264sr  f dB sr>0 (18) A  0 dB otherwise (19) sr

5.4 Over-sea surface duct coupling corrections Obtain the distance from each terminal to the sea in the direction of the other terminal:

dct = coast distance from transmitter km (20)

dcr = coast distance from receiver km (21)

The over-sea surface duct coupling corrections for the transmitter and receiver, Act and Acr respectively, are both zero except for the following combinations of conditions:

2 Act  3exp 0.25dct 1 tanh0.0750  hts dB

if ( 0.75) and (dct ≤ dlt) and (dct ≤ 5 km) (22)

Act  0 dB otherwise (23)

2 Acr  3exp 0.25dcr 1 tanh0.0750  hrs dB

if ( 0.75) and (dcr ≤ dlr) and (dcr ≤ 5 km) (24)

Acr  0 dB otherwise (25) where  is the fraction of the path over sea, hts, hrs are the transmitter, receiver, height above mean sea level.

5.5 Total coupling loss to the anomalous propagation mechanism The total coupling losses between the antennas and the anomalous propagation mechanism can now be calculated as: A 102.45  20logf d  d  A  A  A  A  A ac lt lr lf st sr ct cr dB (26)

Alf is an empirical correction to account for the increasing attenuation with wavelength in ducted propagation: 8 Rec. ITU-R P.617-4

2 Alf  45.375 137.0 f  92.5 f  Db if f < 0.5G Hz (27)

Alf  0 dB otherwise (28)

5.6 Angular-distance dependent loss Specific angular attenuation within the anomalous propagation mechanism:

5 1/3 d 5  10 k  a  f dB/mrad (29) Adjusted transmitter and receiver horizon elevation angles:

at  mint , gt  mrad (30)

ar  minr, gr  mrad (31) Adjusted total path angular-distance: 1000d a  at   ar ka mrad (32) Angular-distance dependent loss:

Aad  d a dB (33)

5.7 Distance and time-dependent loss The loss in the anomalous propagation mechanism dependent on both great-circle distance and percentage time is calculated by first evaluating the following. Distance adjusted for terrain roughness factor:

dar  mind  dlt  dlr,40 km (34) Terrain roughness factor:

5 3  exp 4.610 hm 1043 6dar hm>10 m (35)

3  1 otherwise (36) where hm is the path roughness parameter given in Attachment 2. A term required for the path geometry correction: 9 3.1   0.6  3.510 d  (37) If α < −3.4, set α = −3.4. Path-geometry factor:   500d2   2 2   ka hte hre  (38)

If 2 > 1, set 2 = 1. hte, hre are the effective transmitter, receiver, height above smooth surface given in Attachment 2. Time percentage associated with anomalous propagation adjusted for general location and specific properties of the path: Rec. ITU-R P.617-4 9

  0 2 3 % (39) An exponent required for the time-dependent loss:

1.076exp106 d1.13 9.51 4.8log  0.198log2      2.0058 log1.012 (40) The time-dependent loss:

 A  12  1.2  0.0037d log p 12 p  50 at       q dB (41) where q=100-p.

5.8 Basic transmission loss associated with ducting Basic transmission loss associated with anomalous propagation is given by:

Lba  Aac  Aad  Aat dB (42)

6 Estimation of total transmission loss distribution For dynamic range calculations requiring estimates of the distribution for lower time percentages, pure tropospheric scatter cannot be assumed. The transmission loss values not exceeded for very small percentages of time will be determined by the anomalous propagation mechanism. Tropospheric scatter and the ducting/layer-reflection propagation mechanism are largely correlated and are combined power-wise at these time percentages. The basic transmission loss of the two mechanisms can be combined to give a total loss with equations (4) and (42). 퐿(푝) = −5log (10−0.2퐿푏푠 + 10−0.2퐿푏푎) dB (43)

7 Diversity reception The deep fading occurring with tropospheric scatter propagation severely reduces the performance of systems using this propagation mode. The effect of the fading can be reduced by diversity reception, using two or more signals which fade more or less independently owing to differences in scatter path or frequency. Thus, the use of space, angle, or frequency diversity is known to decrease the percentages of time for which large transmission losses are exceeded. Angle diversity, however, can have the same effect as vertical space diversity and be more economical.

7.1 Space diversity Diversity spacing in the horizontal or vertical can be used depending on whatever is most convenient for the location in question. Adequate diversity spacings h and v in either the horizontal or vertical, respectively, for frequencies greater than 1 000 MHz are given by the empirical relations:

2 2 1/ 2  h  0.36 D  4Ih  m (44)

2 2 1/ 2  v  0.36 D  4Iv  m (45) where D is the antenna diameter in metres and Ih = 20 m and Iv = 15 m are empirical scale lengths in the horizontal and vertical directions, respectively. 10 Rec. ITU-R P.617-4

7.2 Frequency diversity For installations where it is desired to employ frequency diversity, an adequate frequency separation f (MHz) is given for frequencies greater than about 1 000 MHz by the relation:

2 2 1/ 2  f  1.44 f / d  D  Iv  MHz (46) where: f: frequency (MHz) D: antenna diameter (m) : scatter angle (mrad) obtained from equation (1)

Iv: 15 m the scale length noted above.

7.3 Angle diversity Vertical angle diversity can also be used in which two or more antenna feeds spaced in the vertical direction are employed with a common reflector. This creates different vertically-spaced common volumes similar to the situation for vertical space diversity. The angular spacing  r required to have approximately the same effect as the vertical spacing v (m) in equation (45) on an approximately symmetrical path is:

 r  arc tan (v / 500d) (47) where d is the path length (km).

8 Effect of the siting of stations The siting of transmission links requires some care. The antenna beams must not be obstructed by nearby objects and the antennas should be directed slightly above the horizon. The precise optimum elevation is a function of the path and atmospheric conditions, but it lies within about 0.2 to 0.6 beamwidths above the horizon. Measurements made by moving the beam of a 53 dB gain antenna away from the great-circle horizon direction of two 2 GHz transmitters, each 300 km distant, demonstrated an apparent rate-of-decrease of power received of 9 dB per degree. This occurred with increases of scattering angle over the first three degrees, in both azimuth and elevation, for each path, and for a wide range of time percentages.

Attachment 1 to Annex 1

Additional supporting material

1 Seasonal and diurnal variations in transmission loss In temperate climates, transmission loss varies annually and diurnally. Monthly median losses tend to be higher in winter than in summer. The range is 10 to 15 dB on 150-250 km overland paths but diminishes as the distance increases. Measurements made in the European parts of the Russian Federation on a 920 km path at 800 MHz show a difference of only 2 dB between summer and winter Rec. ITU-R P.617-4 11 medians. Diurnal variations are most pronounced in summer, with a range of 5 to 10 dB on 100-200 km overland paths. The greatest transmission loss occurs in the afternoon, and the least in early morning. Oversea paths are more likely to be affected by super-refraction and elevated layers than land paths, and so give greater variation. This may also apply to low, flat coastal regions in maritime zones. In dry, hot desert climates attenuation reaches a maximum in the summer. The annual variations of the monthly medians for medium-distance paths exceed 20 dB, while the diurnal variations are very large. In equatorial climates, the annual and diurnal variations are generally small. In monsoon climates where measurements have been carried out (Senegal, Barbados), the maximum values of Ns occur during the wet season, but the minimum attenuation is between the wet and dry seasons.

2 Frequency of rapid fading on tropospheric scatter paths The rapid fading has a frequency of a few fades per minute at lower frequencies and a few hertz at UHF. The superposition of a number of variable incoherent components would give a signal whose amplitude was Rayleigh distributed and this is found to be nearly true when the distribution is analysed over periods of up to 5 min. If other types of signal form a significant part of that received, there is a modification of this distribution. Sudden, deep and rapid fading has been noted when a frontal disturbance passes over a link. Reflections from aircraft can give pronounced rapid fading. The frequency of the rapid fading has been studied in terms of the time autocorrelation function, which provides a “mean fading frequency” for short periods of time for which the signal is stationary. The median value of the mean fading frequency was found to increase nearly proportionally to path length and carrier frequency, and to decrease slightly with increasing antenna diameter. Measurements have also shown that the rapidity of fading is greatest when the hourly median transmission loss is greater than the long-term median. In general, it was found that the fading rate decreased with decreasing transmission loss below the long-term median, the lowest fading rates occurring for events in which duct propagation was predominant. It is the most rapid fading for hourly-median transmission loss values larger than the long-term median that is most important, and the few measurements available (at 2 GHz) give median fading rates between about 20 and 30 fades/min.

3 Transmissible bandwidth The various discontinuities which give rise to scatter propagation, create propagation paths which may vary in number and in transmission time. Accordingly, the transmission coefficients for two adjacent frequencies are not entirely correlated, which leads to a distortion of the transmitted signal. The transmissible bandwidth is the bandwidth within which the distortion caused by this phenomenon is acceptable for the transmitted signal. This bandwidth therefore depends both on the nature of the transmitted signal (multiplex telephony, television picture, etc.) and on the acceptable distortion for this signal. Studies carried out in France show that: – increasing the antenna gain widens the transmissible bandwidth to the extent where the gain degradation increases also (i.e. for gains exceeding approximately 30 dB); – all other things being equal, the transmissible bandwidth depends on the atmospheric structure and hence on the climatic zone in question; – the transmissible bandwidth becomes narrower as the distance increases, but this is governed by a law which is not the same for all climates; 12 Rec. ITU-R P.617-4

– the transmissible bandwidth becomes narrower when there are positive angles of departure, and wider when these angles are negative.

Attachment 2 to Annex 1

Effective heights and path roughness parameter

The following modelling is the same as Recommendation ITU-R P.2001-2 Section 3.8 effective heights and path roughness parameter. The effective transmitter and receiver heights above terrain are calculated relative to a smooth surface fitted to the profile, as follows. Calculate the initial provisional values for the heights of the smooth surface at the transmitter and receiver ends of the path, as follows:

n 1di  d i 1 h i  h i 1  (2.1) i2

n  2di  d i 1  h i22 d i  d i  1  h i  1 d i  d i  1  (2.2) i2

 2v d  v  h   1 2  stip  2   d  m amsl (2.3)

 v2  v1d  hsrip     d 2  m amsl (2.4)

Where di is the distance from transmitter of i-th profile point (km), hi is the height of i-th profile point above sea level (m), i:1, 2, 3 ... n, index of the profile point, n is the number of profile points.

If hhts stip 1, re-evaluate hstip using:

ℎ푠푡푖푝 = ℎ푡푠 − 1 m amsl (2.5)

Where hts=h1+htg, htg is the height of electrical centre of transmitting.

If hhrs srip 1, re-evaluate hsr using:

ℎ푠푟푖푝 = ℎ푟푠 − 1 m amsl (2.6)

Where hrs=hn+hrg, hrg is the height of receiving antenna above ground. The slope of the least-squares regression fit is given by: hh m  srip stip m/km (2.7) d The effective heights of the transmitter and receiver antennas above the smooth surface are now given by: h h h m te ts stip (2.8) h h h m re rs srip (2.9) Rec. ITU-R P.617-4 13

Calculate the path roughness parameter given by:

hmax h  h  md m m i stip i  (2.10) where the profile index i takes all values from ilt to ilr inclusive. The ilt and ilr are profile indices of transmitter and receiver horizon distances.

______