arXiv:1604.03694v1 [physics.ao-ph] 13 Apr 2016 oiatdpnigo h lcmn ftasitrand transmitter of placement the on depending dominant seabed. a the transmitter low- on alternative both placed an if are is receiver path seabed communications The low-noise, one. water loss, the than lower is o xml,i w iesae1maata mblwthe below 2m loss. antici at through-water than apart 1km less the 1km significantly advantage. from be are key will divers a for attenuation be two surface, need can if path example, the and air For land The without buoys. effec to stations repeater This station surface submerged transmitter. submerged shallow antenna the a air- above between an from the air from communication the cross radiate aids in can to placed appears signal directly and electromagnetic ang boundary an refraction to-water the that and such losses are Propagation interface. water u odcinlast togatnaino electromagne of compone attenuation field waves. strong magnetic propagating to the leads on effect conduction direct but little is there so h Smrne eaiepreblt sapproximately is permeability Relative typic range. but interface variable mS/m air/water quite the is the no conductivity while water at 5S/m, fresh around inally refraction typically significa is of seawater a of angle has Conductivity this the and on material about impact any of of permittivity permittivity frequen highest relative with rapidly a increases and With water, high through water’s higher atte of is wave Plane because conductivity. electrical air and through permittivity propagation from ferent a;E aepoaain hlo seawaters. shallow propagation; wave EM nas; r rsne n iuae.Fnly iuainrslsa results simulation Finally, ones. simulated. experimental with framewor and compared propagation t Some presented are seabed. the antennas are on receiver placed loops and turns wa Transmitter electromagnetic seawaters. of shallow propagation of results experimental nvria eLsPla eGa Canaria, Gran de Palmas Las de Universidad n noaini omnctos(IDeTIC), Communications in Innovation and nmn elyet igepoaainpt ilbe will path propagation single a deployments many In iia feti ena h ebd hr t conductivit its where seabed, the at seen is effect similar A nte motn osdrto steefc fteair-to the of effect the is consideration important Another lcrmgei rpgto hog e ae svr dif- very is water sea through propagation Electromagnetic ne Terms Index Abstract hlo ewtr iuain n Measurements and Simulations Seawater: Shallow nttt o ehooia Development Technological for Institute uei iee,Gr Quintana, Gara Jimenez, Eugenio -al [email protected] e-mail: al ea al ot and Dorta Pablo Mena, Pablo netgto nRdoWv rpgto in Propagation Wave Radio on Investigation Teatospeetfl aesmltosand simulations wave full present authors —The a ams 51,Spain. 35017, Palmas, Las Cnutn eim newtrlo anten- loop underwater medium; —Conducting vnPerez-Alvarez Ivan .I I. NTRODUCTION ǫ r 8,wtrhsthe has water =80, nvria oienc eMadrid, de Politecnica Universidad -al [email protected] e-mail: -al [email protected] e-mail: ataoZz,Mrn Perez Marina Zazo, Santiago nuation lyin ally e in ves TITelecomunicacion, ETSI pated µ ard200 Spain. 28040, Madrid r en- m- cy. =1 nd tic ks re nt nt le y - t . fgoere ihto he n oelyr n applicatio and summarized. layers well more analy and are wave to three full two, extended [3] with the and was geometries [2] of work at [1], his in Sommerfeld and Later, A. geometries century. multilayer XX by the out I. of Fig carried begining in firstly illustrated was is geometries it as seabed the anten receiving on and placed transmitting be both will case our In receiver. on ob simulated. be to going aekn:a2c.rdu e un opatnamd of made antenna wate Sea loop coating. like turns teflon 1mm. ten a using radius isolated and 22cm. copper a kind: same ulwv nlsso rpgtn Mwvsi two-layers in waves EM propagating of analysis wave Full • are frameworks different four communication, this Along • • • nalcss rnmte n eevratna r fthe of are antennas receiver and transmitter cases, all In teuto ewe w oiotllospae nfree in placed loops space. horizontal two between Attenuation teuto ewe w oiotllosimre nsea in inmersed water. loops horizontal two between Attenuation teuto ewe w oiotllospae on layers two placed layer; air loops without problem. interface water horizontal sea to two seabed between Attenuation ebdt e ae nefc iha i ae vrsea over layer air on an problem. placed with layers interface three loops water; water sea horizontal to two seabed between Attenuation Tx i.1 emti ofiuaino testbed. of configuration Geometric 1. Fig. I S II. h SEABED SEA AIR MLTO FRAMEWORKS IMULATION -al [email protected] e-mail: ε µ σ ε cai ltomo h Canary the of Platform Oceanic 10 1 1 0 0 0 µ µ σ ε 2 2 2 d ed 50,Spain. 35200, Telde sad (PLOCAN), Islands dad Quevedo Eduardo k k 0 k Rx nas sis ns r σωµ is modelled as a dielectric with permittivity ǫr=81 and conduc- conductivity (attenuation constant α = 2 Neper/m). This tivity σ=4.5 S/m. Seabed, fine sand, is modelled as a dielectric is true for frequencies over 100 kHzp but not so true for with permittivity ǫr=3.5 and conductivity σ=1 S/m [4]. Height frequencies between 10 kHz and 100 kHz. For low frequencies of sea water layer is set to h=4 m. and distances or for very low frequencies, attenuation de- Simulations are carried out using a commercial MoM creases with frequency in both cases: free space and sea water, solver: FEKO. This tool supports the features needed for and it seems to be independent of the electrical properties of this analysis: planar Green functions for multilayered media, the medium. In this case a magnetostatic approach can explain dielectric coated wires and special basis functions for low this behaviour. frequency analysis. For antennas in free space and frequencies over 100 kHz, Simulations are carried out at five different distances (d=2, atenuattion for each frequency increases 18 dB when doubling 3, 4, 5 and 6 meters) with frequency sweeps from 10 kHz to distance. It is a typical near field dependence (eg 1/R3). 1 MHz. Electrically small loop antennas work as vertical magnetic dipoles. The electromagnetic fields generated by this source [1] III. SIMULATION RESULTS are (cylindrical coordinates): A. Antennas inmersed in homogeneous medium 1 ρz jk2 3k 3j In these simulations, antennas are radiating into two homo- = − − jkr Bρ 4 2 2 3 e geneous mediums: free space and sea water. Results of those πω r  r r r  1 2 2 2 3 3 simulations are shown in Fig 2 and in Fig 3. jk k j z jk k j jkr Bz = − − − − − − e 4πω  r r2 r3 r2  r r2 r3  Attenuation between two ten-turns loops. 1 1 Free space vs. sea water (σ=4.5 ε =81) ρ jk jkr r Eφ = − − e 4π r  r r2  -70

-80 These equations clearly show the aforementioned be- haviours: mainly magnetic field for low frequencies and dis- -90 tances and 1/R3 near field dependence for low distances. -100 B. Antennas inmersed in layered medium -110 d=2m (free space) dB -120 d=2m (sea water) Now we are going to compare the results from simulation d=3m (free space) of antennas in sea water with simulations of antennas placed -130 d=3m (sea water) d=4m (free space) on seabed. Both layers, sea water and seabed, are semi-infinite -140 d=4m (sea water) d=5m (free space) so it is a two layer geometry. -150 d=5m (sea water) d=6m (free space) Results of those simulations are shown in Fig 4 and in Fig 5. -160 d=6m (sea water) 100 200 300 400 500 600 700 800 900 1000

Freq [kHz] Attenuation between two ten-turns loops placed on seabed σ ε σ ε Sea water: =4.5 r=81. Seabed (sand): =1.0 r=3.5

Fig. 2. Free space vs. sea water. Full sweep. -70 d=2m (sea water) d=2m (sea water - seabed) -80 d=3m (sea water) d=3m (sea water - seabed) Attenuation between two ten-turns loops. -90 d=4m (sea water) σ ε Free space vs. sea water ( =4.5 r=81) d=4m (sea water - seabed) -100 d=5m (sea water) d=5m (sea water - seabed) -70 -110 d=6m (sea water)

dB d=6m (sea water - seabed) -120 -80 -130

-140 -90 d=2m (free space) -150 dB d=2m (sea water) -100 d=3m (free space) -160 d=3m (sea water) 100 200 300 400 500 600 700 800 900 1000 d=4m (free space) -110 d=4m (sea water) Freq [kHz] d=5m (free space) d=5m (sea water) d=6m (free space) Fig. 4. Sea water vs. two layer. Full sweep. -120 d=6m (sea water) 10 20 30 40 50 60 70 80 90 100 As it can be seen, the effect of seabed layer is to decrease Freq [kHz] the attenuation at all frequencies. This means that a surface Fig. 3. Free space vs. sea water. Low frequency sweep. wave, lateral wave, has been launched and energy is mainly travelling on the sea-seabed interface. Once again, for low fre- As it is expected for antennas inmersed in sea water, atten- quencies and distances or for very low frequencies, attenuation uation grows exponentially with frequency due to sea water seems to be independent from the medium. Attenuation between two ten-turns loops placed on seabed σ ε σ ε Sea water: =4.5 r=81. Seabed (sand): =1.0 r=3.5 Optical fiber Optical fiber Air

-70 d=2m (sea water) d=2m (sea water - seabed) Signal Spectrum d=3m (sea water) generator analyzer d=3m (sea water - seabed) -80 d=4m (sea water) d=4m (sea water - seabed) 10Base−T 10Base−T to fiber to fiber d=5m (sea water) Sea -90 d=5m (sea water - seabed) d=6m (sea water) Control Control

dB d=6m (sea water - seabed) board board -100 Loop antennas Batteries Batteries

Seabed -110 Fig. 6. Experimental testbed setup. -120

10 20 30 40 50 60 70 80 90 100

Freq [kHz] pressure and temperature using the fiber link. The generator is controlled from an external computer using Keysight VEE Fig. 5. Sea water vs. two layer. Low frequency sweep. software.

B. Receiver and antenna Full expressions for the fields generated by a vertical dipole The receiver is built using a handheld Keysight 9340B placed on the interface between two mediums can be found spectrum analyzer, a Beaglebone Black board, a 10Base-T in [1] and in [5]. We are not going to reproduce them here to fiber transceiver and a battery pack. All the equipment is because of their length and complexity. placed into a receptacle made of high pressure PVC pipe. The Simulations with three layers (seabed, sea water and air) loop antenna is the same used in the transmitter and the PVC has been carried out too. Results of those simulations, with receptacle is pressurized too. The analyzer is controlled from water heigh h=4m, are indistinguishable from those of the two an external computer using Keysight VEE software. layers. So, at least for our testbed, the seawater to air interface seems not to have any effect.

IV. EXPERIMENTAL RESULTS After reviewing a great number of studies about under- water propagation, we have found little information about experimental results in this frequency band. Therefore, a measurement system was designed and several experiments were carried out along 2015 and 2016. After debugging a lot of problems we came to a conclusion: the only way to measure without interferences in this band was to submerge all of the equipment in the sea and communicate with it using a fiber link. No copper cables from undersea to ground, even coaxial ones work like antennas! The selected location is in Taliarte Harbour (Telde, Canary Islands, Spain). This location was selected because PLOCAN’s headquarters are placed there and we can use a private pier. The testbed is shown in Fig 6 and a photograph of both systems is shown in Fig 7. Fig. 7. Transmitter and receiver. A full description and details of the design of the experi- mental seabed can be found in [6]. C. Results A. Transmitter and antenna Frequency sweeps were made between 10 kHz and 100 kHz The transmitter is built using a Keysight 33220A waveform (1 kHz IF bandwidth)and between 100 kHz and 1 MHz (3 kHz generator, a Beaglebone Black board, a 10Base-T to fiber IF bandwidth). Both antennas were placed on the seabed and transceiver and a battery pack with an inversor. All the equip- the distance between their centers was swept between 2 and ment is placed into a receptacle made of high pressure PVC 6 meters using one meter steps. Signal generator power was pipe. The loop antenna is made of enamelled coper covered set to 18 dBm for distances between 2 and 5 meters. For 6 with self-vulcanizing tape and it ts conected to the transmitter meters, signal generator power was set to 23 dBm. using a short patch of coaxial line. The PVC receptacle is After making the measurements, data from the spectrum pressurized and the control board sends information about analyzer needs to be calibrated with a well known source. The spectrum analyzer has a poor response below 40 kHz different frequencies and distances. Full wave simulations and (it’s rated for use from 100 kHz) and calibration curves have measurements are carried out for the same testbed geometry to be made in the lab to improve its response. These curves and two conclusions can be drawn. help extracting the effects of analyzer in the measurements. First, simulations predict that for frequencies over 100 kHz In Fig 8 a full sweep between 10 kHz and 1 MHz is shown. propagation takes place mainly on the seabed-seawater inter- In this figure, simulations of a two layer model (water-seabed) face. The simulated attenuation is greater in an homogeneous are compared with measurements. Heigth of water was four medium (seawater) than in a two layer medium (seawater- meters during the measurements. seabed). These predictions agreed with the measurements. Second, simulations predict that for low frequencies the Attenuation between two horizontal ten-turns loops placed on seabed. influence of the medium decreases with the frequency showing Sea water σ=4.5 ε =81 Seabed σ=1.0 ε =3.5 r r a behaviour that can be explained using a magnetostatic d=2m (meas) -70 d=2m. (sim) approach. This effect is stronger at short distances. These d=3m (meas) d=3m. (sim) predictions agree with the measurements too. -80 d=4m (meas) The good agreement between simulations and measure- d=4m. (sim) d=5m (meas) ments validates the simulation tool. It will let us to make d=5m. (sim) -90 d=6m (meas) ”numerical” experiments with antenna placement (vertical, d=6m. (sim) dB horizontal, etc...), with antenna geometry (radius of the loop, -100 number of turns, shape, etc...) and with frequency choice.

-110 ACKNOWLEDGMENT The authors would like to thank the work carried out by -120 Juan Domingo Santana Urbin (ULPGC) when making the 100 200 300 400 500 600 700 800 900 1000 pressurized PVC receptacles, loop antennas and all of the Freq[kHz] lab stuff needed to make the measurements. Also, the authors would like to thank the work carried out by Gabriel Juanes Fig. 8. Measurements vs. simulations. Full sweep. and Raul Santana (PLOCAN) when setting up measurement In Fig 9 a low frequency sweep between 10 kHz and testbed in the pier and into the sea. 100 kHz is shown. Results for 4, 5 and 6 meters and for This work has been supported by Ministerio de Economia y frequencies below 40 kHz are largely influenced by the Competitividad, Spain, under public contract TEC2013-46011- response of the spectrum analyzer at low power levels at C3-R. these frequencies. A new spectrum HF analyzer from Aaronia REFERENCES GmbH (1 Hz to 30 Mhz) has been acquired and a new [1] R. W. King, M. Owens, and T. T. Wu, Lateral Electromagnetic Waves. measurement campaign is being planned as of writing this New York, NY, USA: Springer-Verlag, 1992. paper. [2] J. R. Wait, Electromagnetic Waves in Stratified Media, ser. IEEE/OUP series on electromagnetic wave theory, J. B. Anderson, Ed. Piscataway, Attenuation between two horizontal ten-turns loops placed on seabed. NJ, USA: IEEE Press and Oxford Univeristy Press, 1996. σ ε σ ε [3] K. Li, Electromagnetic Fields in Stratified Media. Hangzou, China, Sea water =4.5 r=81 Seabed =1.0 r=3.5 Berlin, Germany: Zhejiang University Press and Springer-Verlag GmbH, d=2m (meas) 2009. -70 d=2m. (sim) d=3m (meas) [4] U. M. Cella, R. Johnstone, and N. Shuley, “Electromagnetic wave wireless d=3m. (sim) communication in shallow water coastal environment: Theoretical -80 d=4m (meas) analysis and experimental results,” in Proceedings of the Fourth ACM d=4m. (sim) d=5m (meas) International Workshop on UnderWater Networks, ser. WUWNet ’09. d=5m. (sim) New York, NY, USA: ACM, 2009, pp. 9:1–9:8. [Online]. Available: -90 d=6m (meas) http://doi.acm.org/10.1145/1654130.1654139 d=6m. (sim) dB [5] D. Margetis and T. T. Wu, “Exactly calculable field components of electric -100 dipoles in planar boundary,” Mathematical Physics, vol. 42, pp. 713–745, feb 2001. [6] P. Mena, P. Dorta, G. Quintana, E. Jimenez, I. Perez, S. Zazo, M. Perez, -110 L. Cardona, and J. Brito, “Experimental testbed for seawater channel characterization,” submitted to UComms ’16 Proceedings of the Third -120 Underwater Communications Conference, 2016.

10 20 30 40 50 60 70 80 90 100

Freq[kHz]

Fig. 9. Measurements vs. simulations. Low frequency sweep.

V. CONCLUSIONS This paper investigates the propagation of EM waves gen- erated by loop antennas horizontally placed on seabed at