Deakin Research Online

Transcription

1 Deakin Researh Online This is the published version: Shaneyfelt, Ted, Joordens, Matthew A., Nagothu, Kranthimanoj and Jamshidi, Mo 2008, RF ommuniation between surfae and underwater roboti swarms, in WAC '08: Proeedings of the World Automation Congress '08, I, Pisataway, N.J., pp Available from Deakin Researh Online: Reprodued with the kind permission of the opyright owner. Copyright: 2008, I.

2 RF Communiation between Surfae and Underwater Roboti Swarms Ted Shaneyfelt, Student Member I, Matthew A. Joordens, Member I, Kranthimanoj Nagothu, Student Member I and Mo Jamshidi, Fellow I Autonomous Control ngineering (AC) Center and C Department The University of Texas San Antonio, TX, USA mail: {ted.shaneyfelt, matthew.joordens,kranthimanoj.nagothu}@utsa.edu,moj@waong.org Abstrat-In order for underwater robots to ommuniate with land and air based robots on an equal basis, high speed ommuniations is required. If the robots are not to be tethered then wireless ommuniations is the only possibility. Sonar ommuniations is too slow. Unfortunately radio waves are rapidly attenuated under water due to phenomena suh as skin depth. These experiments attempt to extend the range of underwater radio ommuniations. Index Terms- Antenna radiation patterns, Underwater radio ommuniation, Underwater vehile ommuniation, underwater antennas, Robots 1. INTRODUCTION The Autonomous Control ngineering (AC) enter of the University of Texas at San Antonio is working towards a Net-Centri System of Systems of Roboti Swarms with ooperation among swarms in air, land, and sea. [I], [2]. Roboti swarms ould be used as part of a System of Systems for appliations suh as augmentation of an environmental researh knowledge base [3] or sensor network type appliation [4]. A protool [5] for underwater robots [6] has been developed, whih provides for ommuniations among robots within a single swarm. This paper investigates a means of extending the protool for ommuniating with the surfae to failitate Net-Centriity of the System of Systems. Physial layer ommuniations an be aomplished by a number of tehnologies, inluding wire-line, and wireless methods. Wire-line ommuniations ould be established with all submarines, but the restritions on mobility suh as entanglement suggest wireless ommuniations should be aommodated at some point, whether at the end of a wire, or from a ommuniations devie at the surfae. One ommuniation is made to the surfae, a surfae vessel ould relay signals to other swarms outside of the water. A possible hybrid approah is shown in Fig. 1. Fig. 1 Hybrid Communiations Wireless ommuniations inludes sonar, laser, and radio frequeny. Laser suffers from being diffiult to point, as the Underwater Autonomous Vehiles (UAVs) need to loate one another, and even then murky water ould be a ommon problem. Sonar travels well underwater, but the signal travels slowly with low bandwidth. High end ommerial underwater sonar modems are limited to 9600 baud. A low ost alternative may be radio ommuniations. The main disadvantage of radio ommuniations is that it suffers from heavy attenuation underwater due to the skin depth assoiated with impure water' s ondutivity. Skin depth [7] measures the distane a radio signal an travel through a medium before attenuating to e-! of the original signal power, and is alulated by: 1 distane = (1) w ~ J1+C:J2-1 Manusript reeived February 4, This work was supported in part by University of Texas, San Antonio, Autonomous Control ngineering and by Deakin University. Where w is the frequeny, f1 is the permeability of the water, is the permittivity, (J is the eletrial ondutivity.

3 The AC enter has been using XBee Pro modules for ommuniations among land based vehiles, and is experimenting with using these modules for ommuniations underwater. II. XPRIMNT OVRVIW A previous experiment using XBee Pro modules and omni diretional antennas showed that underwater ommuniations was possible between two robots 25' apart and 9' deep [5]. The XBee Pro modules are 100mW units that use the 2.40Hz radio frequeny for ommuniations. The problem with this is that water readily absorbs radio frequenies around the 2.4 OHz region, whih is why mirowave ooking units operate at 2.450Hz. Therefore experiments using this frequeny range will show the worst ase senario. A previous experiment done with the antenna inside a robot asing, that is, surrounded with a poket of air and then a plasti asing before reahing the water, showed that ommuniations was only possible for robots 8' apart. The next experiments done with the antenna alone immersed in the water resulted in the 25' ommuniations range as mentioned above. All experiments sine have always used the antenna alone in the water. The XBee Pro modules themselves and the ables are shielded to ensure that the antenna is the only objet radiating energy. The following experiments were designed to see if different antenna types ould be used to extend the range of the ommuniations system. The two new antennas were a parabola antenna with an 8dB gain and a Yagi with a 9dB gain. The results of these experiments will be used to improve simulation models of swarm robotis at the AC lab [8]. Ill. XPRIMNT 1 This experiment was arried out to look at extending the horizontal range of the ommuniations system. A. Pool Type The pool used was one that the authors have regular aess to. It is a hlorinated pool being 75 ' long by 36' wide by 9' deep at the deepest point. See Fig. 2. Fig. 2 Chlorinated pool setup As the authors only had 2 lanes available to them so the experimental area's width was redued to 12'. Also the deep part of the pool only extended 20' down the pool, after whih the pool' s depth rapidly rose to 4'. This redued the range possible at depth. Finally the shield antenna ables only allowed a maximum distane of 45 ' to be attained. B. Methodology One antenna was fastened on to a telesoping pole. The pole was attahed to the side of the pool at its deep end suh that the antenna pointed down the length of the pool. This antenna, the base antenna, ould then be moved up and down to the required depth. The other antenna was held by a SCUBA diver at the requested depth and distane and pointed bak at the other antenna. To ensure that the antennas were pointing at eah other a length of line was used. In the base of eah antenna were plaed two eyelet srews in line with the diretion of the antenna. The line was tied to the pole and through the eyelets of the base antenna. The line was then passed through the eyelets of the mobile antenna. When the diver tightened the line, the antennas were brought into alignment. By marking the line at regular intervals the diver ould also use the line to measure the distane between the two antennas. A buoy with another line was used to measure depth. This line was weighted to ause it to hang down straight. The down line was also marked to allow the diver to determine the depth of the antenna. The diver, using underwater sonar ommuniations gear to reeive instrutions from the surfae, was able to plae the antenna where requested by the surfae party. The surfae party ould then read the signal strength and paket suess rate when given the go ahead by the diver. The base antenna was plaed at the required depth and then the diver moved the antenna away in 5' inrements. One the maximum distane was reahed the antennas were moved down by a foot and the proedure was repeated. C. Observations The diver found that handling the antenna, the distane line and the depth line was umbersome and hene error prone. This was amplified when the diver was in water deep enough that he ouldn't touh the bottom or any walls, having to float at the required depth and pull the distane/alignment line without moving himself. Pulling the alignment line meant that the diver had to swim to stop moving toward the line. This made it harder to maintain the requested depth. If the diver tried to stead himself on the depth line, this would pull down the line making the depth measurement less aurate. This an be seen in Fig. 3 whih is part of the log from the diver's omputer showing the depth. IV. XPRIMNT 2 This experiment was arried out to determine the vertial range of the ommuniations system.

4 o P T H lapsd DIV TIM o P T H 10 lapsd DIV TIM MirUe, POC Model DATAMASK Dive S~ A IIle: 2Set Fig. 3 xerpt from diver's log showing the diffiulty in maintaining a onsistent depth. A. Poo/Type The pool used a saltwater pool owned by Dive World San Antonio. As a SCUBA pool it had a greater depth of 23 '. The pool was about 20' wide and 40' long and round at both ends. The deep part of the pool was about 20' in diameter. The remaining part of the pool was half at 9' and half at 5'. The whole pool was available to the authors but only the deep setion was used. B. Methodology One antenna was fastened on to a telesoping pole. The pole was attahed to the side of the pool aross the surfae suh that the antenna pointed down toward the bottom of the pool. This antenna, the base antenna, was just under the surfae of the water. The diver was given the mobile antenna and held the antenna, pointing up, at the requested depth. The diver, using underwater sonar ommuniations gear to reeive instrutions from the surfae, was able to plae the antenna where requested by the surfae party. The surfae party ould then read the signal strength and paket suess rate when given the go ahead by the diver. The diver used a line from a buoy to stead himself. The diver would move the antenna up or down in either l' or 3' inrements. The depth was determined by the diver's depth gauge. C. Observations Before this experiment was run, beause this pool had not been used before, one of the underwater robots was testing in the pool under remote ontrol. The robot was ontrolled using a model airraft radio ontrol with a frequeny of 75MHz. It was found that the radio lost ontat with the robot in just l' of water, whilst in the other pool the ontat was maintained down to 9' This raised the question of whether the salt water, as opposed to hlorinated, was worse for radio ommuniations. POC Model DATAMASK DiYeS~ Rale:2S"", Fig. 4 xerpt from diver's log showing the more onsistent depth. ah step is the diver moving up or down between measurements By using a SCUBA mask with built in omputer and heads up display, the diver ould see his depth and keep wathing the antenna. This meant that he had one hand to hold the antenna and one to steady himself on the down line from the buoy. As the depth sensor was at eye level in the SCUBA mask the diver ould keep himself at the right depth and keep the antenna at eye level resulting in a more aurate antenna depth. In this way it didn't matter if the down line moved a bit as it wasn't used in the measurement. This an be seen in Fig. 4 whih is part of the log from the diver's omputer showing the depth. ah step shows the diver moving between depths for eah measurement. V. XPRIMNT RSULTS Measurements were initially to be measured horizontally by measuring tape, and vertially by a marked line extending downward from a buoy was to be used to measure depth. At the time, the only measuring tape readily available to the authors was a metalli fish tape, whih was suspeted of ausing interferene, so some measurements were made to verify this as shown in Fig. 5. The interferene of the tape prompted the authors to abandon the use of metalli measuring tape in favor of string tied off at regular intervals. o ~....s::. 110 <II... Signal Affeted by Metali Measuring Tape (S'at 2'depth) I I I along tape diag tape no tape ~ "i ~.:- iii Fig. 5 Signal strength measurements demonstrating interferene of metalli tape along the transmission path

5 Horizontal Transmission in Chlorinated Water with Paraboli Antennas o ! ~ ; to ~ ~ iii ~ Distane (It) - 2ft depth --3 ft depth --4 ft depth - 5 ft depth - 6ft depth - 7ft depth Fig. 6 Signal strength measurements using paraboli antennas in hlorinated water for a horizontal transmission path The horizontal measurements using the line extending from the buoy were used in the shallow hlorinated pool. The data olleted is shown in Fig. 6. Vertial transmission was performed in a saltwater pool using various antennas with the results shown in Fig. 8. All three antennas gave surprisingly steady results for inreasing depth up to the bottom of the pool, where the omnidiretional antenna signal dropped off. The Yagi antenna performed worse than expeted at all depths ompared to the omnidiretional antenna. The paraboli antenna performed better than the omnidiretional antenna, as expeted, but the next measurement measuring signal strength for misaligned antennas gave ounterintuitive results. The results in Fig. 9 show that the diretion the parabola was pointing did not have the expeted strong effet on the signal strength. Diretional Antenna Misallignment o _0 CD ~ i f j:: I 1iQ 't ~ t;; to ;, v; Misallignment (degrees)... Yagi -. Paraboli Fig. 9 The effet of misalignment on a paraboli antenna reeption Fig. 7 Fresnel zone indiated by elliptial region near water surfae These measurements show a general degradation in signal strength as depth inreases. This degradation is most pronouned within 3' from the surfae and within 3' of the bottom of the pool. The degradation at those depths might be explained partially by the Fresnel zone [9] interseting the surfae or floor of the pool, as shown in Fig. 7. Where the Fresnel zone runs along the water surfae, the surfae of the water may at as a wave guide. The drifting of the data samples for 3' depth in deep water may be due to the diffiulty the diver had in maintaining desired depth during data olletion. 0_0 _ 20.0 CD ~ s:... Vertial Transmission Through Saltwater ~ Antenni! Type ----Paraboli ~ Omni t;; to 80.0 '" bii v; , _.... Depth (m)... YAGI VI. CONCLUSIONS We have found ommuniations in a saltwater pool to be more diffiult than in a hlorinated pool, as the ondutivity of the saltwater is higher, adversely affeting the skin-depth. Nevertheless, we have suessfully transmitted vertially in saltwater to the maximum depth available to us in the pool. Future researh inludes testing in mild oeani saltwater off the windward oast of the Big Island of Hawaii. We have found that although the diretional antenna gave better signal strength readings than the omni diretional antenna, the diretion in the pool was not influential as expeted. Our diver's observations explain anomalies in depth readings in the shallow water pool, and have improved our method of depth measurement for data olletion. Oean water olletion of data will need to be arried out using the diver's in-mask depth-gage and a rope will be suspended from kayaks for diver stabilization. The resulting data will hopefully be available at the WAC 2008 onferene. We have demonstrated viability of sub to surfae ommuniations by using underwater radio ommuniations. Whether radio ommuniations is used for sub to surfae or some other method is used to ommuniate from under the water to the surfae, or a ombination thereof, this exerise has demonstrated sub to surfae ommuniations as a proof of onept for a Net-Centri System of Systems involving air, land, and sea-based swarms of robots. Fig. 8 Signal strength measurements using various antennas in saltwater for a vertial transmission path

6 ACKNOWLDGMNT The authors wish to thank Serkan Dursun for his expertise on radio propagation theory, and Dive World West for the use of their saltwater pool. [1] RFRNCS Ted Shaneyfelt, Mo Jamshidi, "Net-Centri System of Systems Involving Roboti Swarms aross Air, Land, and Sea," in University of Texas at San Antonio 2007 ngineering, Siene, and Business Conferene, San Antonio, Texas, USA, Innovations) for 3 years in testing, produt planning, seurity, and ontrol. After a year of developing Windows based doument proessing software with Profitability of Hawaii, he returned to the University of Hawaii at Hilo to teah for 7"h years and developed a series of Web Tehnology ourses and related materials while mentoring high-shool roboti teams whih won numerous US Congressional reognition awards. Mr. Shaneyfelt is urrently studying for a Ph.D. at The University of Texas at San Antonio as a student member of the I. [2] [3] [4] [5] [6] Robert J. Cloutier, Mihael J. DiMario, Hans W. Polzer, "Net Centriity and System of Systems", System of Systems ngineering - Innovation for the 21st Century,(m. Jamshidi, ed.) Chapter 9 Wiley & Sons 2008 S. S. rdogan, Ted Shaneyfelt, Andrew Honma and Cam Muir, "Integrated Knowledge Base for nvironmental Researh", Pro. I International Conferene on Autonomi and Autonomous Systems International C01iferene on Networking and Servies (ICAS'05 and ICNS'05), Otober 23-28, 2005, Papeete, Tahiti, Frenh Polynesia. Prasanna Sridhar, Asad M. Madni, and Mo Jamshidi, "Hierarhial Aggregation and Intelligent Monitoring and Control in Fault Tolerant Wireless Sensor Networks", I Systems Journal, 2007, pp Kranthimanoj Nagothu, Matthew Joordens, Mo Jamshidi, "Communiations for underwater Robotis Researh platforms," in Pro. I Systems Conferene, Montreal, Canada, 2008,p. 6. M. Joordens, "Design of a low ost Underwater Roboti Researh Platform," in paper submitted to I SoS Conferne, Monterey, CA, USA, 2008, p. 6. M. A. Joordens (Member -I, Fellow - The Institution of ngineers Australia) earned his Bahelor of ngineering (eletroni) degree at Ballarat University in 1988 and a Masters of ngineering (by researh) in Virtual Reality at Deakin University in He began his areer with Industrial Control Tehnology designing ontrol systems to automate various different industrial proesses. For 5 years he designed miroproessor based ontrol systems for ompanies suh as Ford, Pilkington Glass, Webtek and Blue Cirle Southern Cement. He then moved to Deakin University and wrote their first eletronis units. Using his industrial experiene he designed one of the first Australian ngineering degrees in Mehatronis that still runs at Deakin as Mehatronis and Robotis. He urrently letures units in Digital eletronis, Miroontrollers, Robotis and Atiifiial Intelligene after 15 years at Deakin. He is urrently researhing underwater swarm robotis in the USA. Mr. Joordens is a Fellow of the Institution of ngineers, Australia and an I member. [7] [8] [9] Griffiths, David, "Introdution to letrodynamis," 3rd edition, Upper Saddle River, New Jersey: Prentie Hall, 394. ISBN x. OCLC J. Prevost, M. Joordens and Mo Jamshidi, "Simulation of Underwater Robots using MS Robot Studio " in paper submitted to I SoS Conferne, Monterey, CA, USA, 2008, p. 5. Andersen,.J, "Fresnel Zones for Ground-Based Antennas,". in Pro. I Trans. Antennas and Propagation., Vo1.12, no. 4, pp , July Ted Shaneyfelt (Student Member-I) earned the first Bahelor of Siene in Computer Siene degree at the University of Hawaii at Hilo in 1986, and earned his Master of Siene degree in letrial ngineering (Computer ngineering) at the University of California, San Diego in After a year of legay mainframe experiene with ASI, Mr. Shaneyfelt worked in teleommuniations for 7 years with Hughes Network Systems. There he developed design and manufaturing test system for VSAT satellite ground equipment, he designed the RF hannel simulator ommerialized by Hewlett-Pakard, he designed the embedded user interfae for North Ameria's first-deployed dual-mode ellular telephone, and he developed the first prototype whih led to the development of the OnStar system, along with other onept vehile projets for GM and Hughes. He ontinued working in teleommuniations with Sony, MCSI (later aquired by Motorola as Broadband Kranthimanoj Nagothu (Student Member- I) reeived the Bahelor of ngineering degree in letronis and Communiations ngineering from Anna University, Chennai, India in 2006 where he is urrently pursuing a Master of Siene degree in letrial ngineering. In 2006, he joined the Autonomous Control ngineering Center as a Graduate Researh Assistant. His urrent researh areas are ommuniations for Underwater Robotis Swarms, Sensor Networks and Systems of Systems. He has been atively partiipating in organizing international onferenes. Mr. Nagothu reeived several awards inluding the best paper award for the SB student onferene 2007 held at University of Texas at San Antonio. Mo Jamshidi (Fellow-I, ASM, AAAS, NY AS, TWAS, Assoiate Fellow AIAA) reeived his Ph.D. Degree in letrial ngineering from the University of Illinois at Urbana-Champaign in February He holds Honorary Dotorate Degrees from Azerbaijan National University, Baku, Azerbaijan (1999), the University of Waterloo, Canada (2004) and Tehnial University of Crete, Greee, (2004). Currently, he is the Luther Brown ndowed Chaired Professor at the University of Texas at the San Antonio Campus, San Antonio, Texas, USA. He was the founding Diretor of the Center for Autonomous Control ngineering (AC) at the University of New Mexio (UNM), and moved to the University of Texas, San Antonio in early He has been the Diretor of the ICSoS - International Consortium on

7 System of Systems ngineering (isos.org) - sine Mo is an Adjunt Professor of ngineering at Deakin University, Australia. He is also Regents Professor meritus of C at UNM. In 1999, he was a NATO Distinguished Professor in Portugal of Intelligent Systems and Control. Mo has written over 560 tehnial publiations inluding 62 books and edited volumes. Six of his books have been translated into at least one foreign language other than nglish. He is the Founding ditor, Co-founding ditor or ditor-in-chief of 5 journals. Mo is ditor-in-chief of the new I Systems Journal (to be inaugurated in 2007) and founding ditor-in-chief of the I Control Systems Magazine. Mo is a Member of Senior Member of the Russian Aademy of Nonlinear Sienes, Hungarian Aademy of ngineering and an assoiate fellow of the AIAA. He is a reipient of the I Centennial Medal and I CSS Distinguished Member Award. Mo is urrently on the Board of Governors of the I Soiety on Systems, Man and Cybernetis and the I Systems Counil. In 2005, he was awarded the I SMC Soiety's Norbert Weiner Researh Ahievement A ward and in 2006 the I SMC Distinguished Contribution Award. In 2006 he was awarded a "Distinguished Alumni in ngineering" at the Oregon State University (one of his Alama Matters), Corvallis, Oregon, USA.