Prototype Design and Experimental Evaluation of Autonomous Collaborative Communication System for Emerging Maritime Use Cases

senso s A icle P o o ype Design and Expe imen al E alua ion o Au onomous Collabo a i e Communica ion Sys em o Eme ging Ma i ime Use Cases Ji i Poko ny 1,2,* , Khanh Ma 2, Salwa Saa i 1,2 , Jakub F olka 1, Jose Villa 3, Mikhail Ge asimenko 2, Ye geni Kouche ya y 2and Ji i Hosek 1   Ci a ion: Poko ny, J.; Ma, K.; Saa i, S.; F olka, J.; Villa, J.; Ge asimenko, M.; Kouche ay y, Y.; Hosek, J. P o o ype Design and Expe imen al E alua ion o Au onomous Collabo a i e Communica ion Sys em o Eme ging Ma i ime Use Cases. Senso s 2021,21, 3871. h ps://doi.o g/10.3390/ s21113871 Academic Edi o : And ey V. Sa kin Recei ed: 30 Ap il 2021 Accep ed: 29 May 2021 Published: 3 June 2021 Publishe ’s No e: MDPI s ays neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a il- ia ions. Copy igh : © 2021 by he au ho s. Licensee MDPI, Basel, Swi ze land. This a icle is an open access a icle dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion (CC BY) license (h ps:// c ea i ecommons.o g/licenses/by/ 4.0/). 1 Depa men o Telecommunica ions, Facul y o Elec ical Enginee ing and Communica ion, B no Uni e si y o Technology, Technicka 12, 616 00 B no, Czech Republic; [email p o ec ed].cz (S.S.); [email p o ec ed].cz (J.F.); [email p o ec ed].cz (J.H.) 2Uni o Elec ical Enginee ing, Tampe e Uni e si y, Ko keakoulunka u 7, 337 20 Tampe e, Finland; [email p o ec ed] (K.M.); [email p o ec ed] (M.G.); e geny[email p o ec ed] (Y.K.) 3Mecha onics Resea ch G oup, Tampe e Uni e si y, Ko keakoulunka u 6, 337 20 Tampe e, Finland; [email p o ec ed] *Co espondence: xpoko 26@ u b .cz Abs ac : Au oma ed sys ems ha e been seamlessly in eg a ed in o se e al indus ies as pa o hei indus ial au oma ion p ocesses. Employing au oma ed sys ems, such as au onomous ehicles, allows indus ies o inc ease p oduc i i y, bene i om a wide ange o echnologies and capabili ies, and imp o e wo kplace sa e y. So a , mos o he exis ing sys ems conside u ilizing one ype o au onomous ehicle. In his wo k, we p opose a collabo a ion o di e en ypes o unmanned ehicles in ma i ime o sho e scena ios. P o iding high capaci y, ex ended co e age, and be e quali y o se ices, au onomous collabo a i e sys ems can enable eme ging ma i ime use cases, such as emo e moni o ing and na iga ion assis ance. Mo i a ed by hese po en ial bene i s, we p opose he deploymen o an Unmanned Su ace Vehicle (USV) and an Unmanned Ae ial Vehicle (UAV) in an au onomous collabo a i e communica ion sys em. Speci ically, we design high-speed, di ec ional communica ion links be ween a e es ial con ol s a ion and he wo unmanned ehicles. Using measu emen and simula ion esul s, we e alua e he pe o mance o he designed links in di e en communica ion scena ios and we show he bene i s o employing mul iple au onomous ehicles in he p oposed communica ion sys em. Keywo ds: au onomous ehicles; ma i ime use cases; USV; UAV; collabo a i e communica ion sys em; di ec ional wi eless links; p o o ype design 1. In oduc ion Digi aliza ion and digi al ans o ma ion a e shaping indus ies ac oss he wo ld. Among hese, he ma i ime indus y can bene i om he p ocess o mode nizing i s exis ing p ac ices o imp o e ope a ional e iciency [ 1 ]. As pa o he sys ems and solu ions in which essel ope a o s a e in es ing, au onomous ehicles and obo ics a e being u ilized no only in po s and ha bo s, bu also o aid he sea anspo se ices in o sho e a eas [ 2 ]. The deploymen o su ace, ae ial, and unde wa e au onomous sys ems open possibili ies o imp o emen s in cu en ma i ime ope a ions and assis in se e al se ices, such as ma i ime sea ch and escue, onboa d applica ions, na iga ion, and lee managemen [ 3 ]. Howe e , hese bene i s a e condi ional upon he u iliza ion o a wi eless communica ion sys em ha in ol es hese en i ies and suppo s he communica ions be ween hem. This end in unmanned- ehicle-aided o sho e sys ems is he main mo i a ion behind launching esea ch in his a ea. Speci ically, he goal is o cons uc a p o o ype o a collabo a i e communica ion sys em composed o di e en au onomous ehicles ope a ing Senso s 2021,21, 3871. h ps://doi.o g/10.3390/s21113871 h ps://www.mdpi.com/jou nal/senso s Senso s 2021,21, 3871 2 o 20 in an o sho e en i onmen . This sys em is buil upon su ace, ae ial, and unde wa e componen s ep esen ed by a Unmanned Su ace Vehicle (USV), Unmanned Ae ial Vehicle (UAV), and Au onomous Unde wa e Vehicle (AUV), espec i ely. As pa o he inal p o o ype, a collabo a i e communica ion sys em should be implemen ed o connec hese h ee subsys ems using di e en adio access echnologies. The o e all communica ion layou o he sys em is depic ed in Figu e 1. P ospec i e applica ions o he AUV include unde wa e mapping, en i onmen al su eying, and iden i ica ion o unde wa e haza ds o na iga ion. Howe e , mainly he USV and UAV a e conside ed in ou p e ious and cu en wo ks on he au onomous collabo a i e communica ion sys em. In [ 4 ], he design o a high-speed di ec ional communica ion link be ween he G ound Con ol s a ion (GC) and USV was ou lined and i s pe o mance was e alua ed. To u he enhance he p oposed au oma ed o sho e sys em, al e na i e solu ions should be aken in o accoun o main ain he GC–USV connec ion in cases whe e he Line-O -Sigh (LOS) link canno be es ablished. Hence, we build upon he wo k in [ 4 ] by in ol ing he ae ial componen ep esen ed by he UAV o complemen he p e iously p oposed GC–USV se up and, hus, o enable a ully ledged au oma ed o sho e sys em [5]. GC USV AUV UAV Di ec ional Wi-Fi (Links 1, 2, 3) Omni-di ec ional Wi-Fi (Link 4) GPS (Links 5, 6) Landing beacon (Link 7) 1 1 2 2 3 3 5 5 6 6 4 4 7 7 Figu e 1. O e all layou o he p oposed wi eless communica ion sys em. 1. GC–USV di ec ional Wi-Fi; 2. USV–UAV di ec ional Wi-Fi; 3. GC–UAV di ec ional Wi-Fi; 4. UAV–AUV nondi ec ional Wi-Fi; 5. USV–sa elli e link; 6. UAV–sa elli e link; 7. Beacon o UAV landing. The es o his a icle is o ganized as ollows: Sec ion 2p o ides a desc ip ion o he s a e-o - he-a o he indus ial applica ions o UAVs, speci ically in he ma i ime indus y, hei di e en communica ion scena ios, and he challenges o hei deploymen wi h he e iew o ele an ela ed esea ch wo ks. In Sec ion 3, we desc ibe he conside ed wi eless communica ion modes be ween he GC, USV, and UAV and discuss he communica ion enable s o hese modes, speci ically in e ms o loca ion-based beam-s ee ing capabili ies and adio p opaga ion models. We hen de ail, in Sec ion 4, he p o o ype a chi ec u e design o he GC, USV, and UAV wi h hei di e en communica ion modules. Fu he , Sec ion 5desc ibes he mechanical, powe , and ne wo king componen s and me hods u i- lized in he p o o ype implemen a ion o he au oma ed o sho e sys em. The pe o mance o he p oposed sys em is e alua ed in Sec ion 6, whe e we compa e he e alua ion esul s ob ained om measu emen campaigns o hose ob ained om simula ions o selec ed analy ical models. This compa ison will allow us o iden i y he models ha can be e cha ac e ize he adio p opaga ion in he s udied scena ios and ha can, hus, be used o simula ion-based e alua ion o u u e enhancemen s o he communica ion sys em. Addi ionally, he bene i s o employing a UAV in he p oposed communica ion sys em a e e alua ed on a eal scena io and he echnique o de e mining he op imal UAV loca ion is desc ibed. Senso s 2021,21, 3871 3 o 20 2. S a e-o - he-A and Rela ed Wo k UAVs, commonly known as d ones, a e cha ac e ized by hei mobili y, adap i e al i- ude, ad anced wi eless communica ion, and sensing capabili ies [ 6 ]. Such cha ac e is ics mo i a e se e al indus ies o employ UAVs in di e en ope a ing scena ios, including bu no limi ed o he manu ac u ing indus y [ 7 ], ag icul u e [ 8 ], and public sa e y [ 9 ]. In ac , he global ma ke o d ones o indus ial applica ions is p ojec ed o g ow o a ound USD 43 billion by 2024, eaching he compound annual g ow h a e o 20% be ween 2018 and 2024 [ 10 ]. Simila ly o o he indus ies, he applica ions o UAVs in ma i ime use cases can a y om sea ch and escue p ocedu es, da a ga he ing o na iga ion and lee managemen , and emo e moni o ing ia ideo su eillance [ 11 ] o cha ging shipbo ne senso s using wi eless powe ans e [12]. As pa o he deploymen op ions, UAVs can be used as ae ial base s a ions o enhance he co e age [ 13 ] and capaci y o wi eless ne wo ks [ 14 ]. This op ion can be aluable in o sho e a eas whe e LOS links wi h he e es ial access ne wo k canno be es ablished. Addi ionally, UAVs can ope a e as lying mobile e minals and be used as elay nodes in ou -o -co e age and Non-Line-O -Sigh (NLOS) scena ios [15]. In such si ua ions, essels can make use o he exis ence o p oxima e UAVs o access he ne wo k ia elaying. In b ie , UAV-assis ed communica ions can be u ilized o empowe he pe o mance o wi eless ne wo ks owa d eliable, massi e da a ansmissions, and addi ional co e age anges o ma i ime applica ions [16]. Howe e , he abo emen ioned bene i s come wi h a ious challenges in he deploy- men o UAVs. These challenges a e in he scope o se e al esea ch wo ks, such as ai - o-g ound channel modeling [ 17 ], ajec o y planning o UAVs pe o ming explo- a ion asks in disas e scena ios [ 18 ], dynamic placemen o elay UAVs [ 19 ], and he challenge o ene gy-e icien ope a ion in UAV ne wo ks due o he limi ed onboa d en- e gy [ 20 , 21 ]. To ackle some o hese challenges, ce ain esea ch wo ks ha e p oposed he use o So wa e-De ined Ne wo king (SDN) o se e al con ol unc ions in Flying Ad Hoc Ne wo ks (FANETs), including bu no limi ed o opology managemen and collision a oidance [22,23]. Al hough in ol ed in a ious applica ions, UAVs a e u ilized as he only ype o unmanned ehicle in mos o he sys ems s udied in he e iewed esea ch wo ks. In his a icle, we p opose he use o wo ypes o unmanned ehicles, namely, USV and UAV, in a collabo a i e communica ion sys em o ma i ime use cases. We de ail he UAV deploymen scena ios in he p oposed au onomous communica ion sys em in he ollowing sec ions. 3. P oposed Communica ion Modes and Solu ion Componen s Based on he LOS/NLOS scena ios, wo communica ion modes be ween he GC and he USV in he p esen ed au oma ed o sho e sys em a e p oposed, namely, a LOS communica ion mode and a elay communica ion mode. The i s mode is based on LOS communica ions be ween he GC and he USV and is depic ed in Figu e 2a. The second communica ion mode u ilizes elaying ia UAV in an NLOS scena io, as demons a ed in Figu e 2b. GC USV Di ec ional an enna GC-USV LOS link (a) GC–USV LOS communica ions. GC USV UAV Di ec ional an enna Omni-di ec ional an enna GC-UAV LOS link GC-USV NLOS link UAV-USV LOS link (b) GC–USV communica ions ia UAV elaying. Figu e 2. Illus a ion o he p oposed communica ion modes in he au onomous collabo a i e sys em. Senso s 2021,21, 3871 4 o 20 To es ablish high-speed links, he GC, USV, and UAV a e equipped wi h di ec ional an ennas and dynamic beam-s ee ing capabili ies. In Sec ion 3.1, he loca ion-based beam- s ee ing algo i hm u ilized in ou communica ion sys em is in oduced. In addi ion, p opaga ion models a e u ilized o p edic he channel pa h losses and, hus, o cha ac e ize he beha io o adio p opaga ion in he communica ion sys em wi h he conside a ion o ce ain en i onmen e ec s. Fu he de ails on he selec ed p opaga ion models a e p o ided in Sec ion 3.2. 3.1. Loca ion-Based Beam-S ee ing Algo i hm As depic ed in Figu e 2, he u iliza ion o di ec ional an ennas depends on he commu- nica ion mode. In he LOS communica ion mode, bo h he USV and GC a e equipped wi h di ec ional an ennas. In o de o align he beams o hese an ennas in he sys em, an algo- i hm o beam-s ee ing based on de ices’ loca ions was de eloped in [ 4 ]. This algo i hm included wo main s eps: (1) calcula ion o he ac ual beam-s ee ing angle; (2) ansmission o he con ol commands wi h he new angle alue. In he elay communica ion mode, bo h he USV and he UAV a e equipped wi h one di ec ional an enna and one omnidi ec ional an enna while he GC ca ies one di ec ional an enna. Using he same algo i hm p oposed in [ 4 ], he d one’s yaw can be se in a way ha he ins alled an enna is ho izon ally aligned wi h he GC. Fo il ing o he an ennas on he UAV and he GC, Equa ion (1) , ob ained om [ 24 ], can be used o calcula e he ele a ion angle be ween wo poin s wi h ele a ions ele 1 and ele 2 abo e he g ound: Ele a ion angle =180 πele 2−ele 1 d−d 2R, (1) whe e R is he Ea h’s adius and d is he dis ance be ween wo poin s ( la 1, lon 1) and ( la 2, lon2), which can be calcula ed using he ollowing Ha e sine o mula [24]: d=2 sin−1 ssin2la 2−la 1 2+cos(la 1)cos(la 2)sin2lon2−lon1 2!. (2) Finally, o ind he il ing o he di ec ional an ennas on he UAV and he GC, he ho - izon al angle calcula ion pa in he algo i hm p oposed in [ 4 ] should be eplaced by he ele a ion angle using Equa ion (1). 3.2. Radio P opaga ion Models The choice o adio p opaga ion models in he p esen ed communica ion sys em depends on he adio links in bo h LOS and elay communica ion modes, namely, GC–USV and GC–UAV–USV links. Speci ically, he F ee Space Pa h Loss (FSPL) model is used o he simula ion o he communica ions be ween he UAV and he o he de ices. The GC–USV link can be cha ac e ized wi h LOS communica ions in he nea sea-su ace en i onmen . The selec ed p opaga ion models a e adjus ed o he 5 GHz equency bands since he IEEE 802.11ac s anda d is u ilized as he connec i i y solu ion in his communica ion sys em. On op o he o e ed capaci y and co e age, he choice o Wi-Fi echnology can be jus i ied by i s low cos and high a ailabili y ha make i sui able o ou expe imen al p o o yping. 3.2.1. FSPL Model The FSPL model desc ibes an ideal adio condi ion be ween a ansmi ing and a ecei ing an enna, whe e only LOS link exis s wi hou o he sou ces o sca e ing, di ac- ion, o e lec ion [ 25 ]. I is one o he commonly used models o cha ac e iza ion o he g ound- o-ai adio p opaga ion [ 26 ]. The e o e, his model is used o de e mine pa h loss in he communica ion om su ace de ices o he UAV based on Equa ion (3) [25]. LFSPL =32.44 +20log10( c) + 20log10(d), (3) Senso s 2021,21, 3871 5 o 20 whe e LFSPL— ee space pa h loss (dB); c—ca ie equency (MHz); d—dis ance be ween ansmi e and ecei e (km). 3.2.2. Nea Sea-Su ace P opaga ion Models To cha ac e ize he adio p opaga ion o e he GC–USV link in he nea sea-su ace en i onmen , he channel model om he s udy o Lee e al. [ 27 ] is u ilized. Au ho s in [ 27 ] p oposed a combina ion o 2- ay and 3- ay p opaga ion models and showed ha he u iliza ion o each model depends on a b eak poin db eak . This me ic indica es he ansi ion poin be ween he wo models. In de ail, he pa h loss p edic ion abili y o he 2- ay model in a nea sea-su ace LOS en i onmen de e io a es as he p opaga ion dis ance inc eases beyond db eak . In his la e case, he 3- ay model becomes a be e op ion o he pa h loss p edic ion. The b eak poin db eak can be es ima ed using Equa ion (4) , whe e λ is he wa eleng h in me e s, and h and h a e he ansmi ing and ecei ing an enna heigh s in me e s, espec i ely [27]. db eak =4h h λ. (4) The ep esen a ion o he u ilized 2- ay pa h loss model is depic ed in Figu e 3and he p edic ed pa h loss L2- ay (dB) can be calcula ed using Equa ion (5). L2- ay =−10log10(λ 4πd22 sin2πh h λd2). (5) Figu e 3. Rep esen a ion o he 2- ay adio p opaga ion. On op o he di ec and e lec ed ays, he 3- ay model akes in o conside a ion he e ac ed ay caused by he duc ing e ec s, as shown in Figu e 4. This model assumes ha he e apo a ion duc laye is ho izon ally homogeneous and he is i s e ec i e heigh . Based on Equa ions (6) and (7), he 3- ay pa h loss L3- ay (dB) can be p edic ed. L3- ay =−10log10(λ 4πd2 [2(1+∆)]2), (6) wi h ∆=sin2πh h λdsin2π(he−h )(he−h ) λd. (7) Senso s 2021,21, 3871 6 o 20 Figu e 4. Rep esen a ion o he 3- ay adio p opaga ion. 4. Sys em P o o ype A chi ec u e To enable LOS and elay communica ion modes, he communica ion modules o all h ee pla o ms (USV, GC, and UAV) mus be equipped wi h long- ange high- h oughpu anscei e s. The long- ange aspec can be enabled wi h dynamic s ee ing o di ec ional an- ennas. Each module combines a ious mechanical, ne wo king, and elec ical componen s o enable he an enna s ee ing capabili y. The USV, GC, and UAV ha e communica ion modules wi h di e en a chi ec u e designs ha a e be e depic ed in Figu es 5–7, espec i ely. The USV consis s o a single-boa d compu e , a mo ion senso , wo se o mo o s, and a se o con olle ins alled on he compound pla e oge he wi h he di ec ional an enna. These wo se o mo o s a e esponsible o he e ical and ho izon al s ee ing o he an enna. Fo ho izon al beam-s ee ing o he USV an enna, he posi ioning da a (i.e., posi ion and o ien a ion) is collec ed om a Global Posi ioning Sys em (GPS) compass module by he USV main compu e using Robo Ope a ing Sys em (ROS). Subsequen ly, he da a om he GPS compass is sen o he single-boa d compu e ia E he ne . By assuming ha he loca ion o he GC is ixed, his compu e calcula es he s ee ing angle by using he posi ions o he essel and he GC, hen sends a con ol signal in he ROS o ma o he se o con olle . The con olle hen ac i a es a se o mo o o o a e he USV an enna owa ds he GC. When he anscei e s on bo h sides a e connec ed h ough a Wi-Fi link, he USV’s GPS coo dina es a e sen o he g ound sys em and collec ed by he con olle boa d in GC. The GC s ee ing sys em also includes a DC mo o , a mo o d i e , and a se o mo o . The DC mo o s ee s he an enna ho izon ally o he di ec ion o he au onomous essel. Single-boa d compu e Se o Mo o (Azimu h) Se o Mo o (Ele a ion) Se o con olle Mo ion Senso GPS module USV Cen al Compu e USV Di ec ional An enna Powe Regula o Ex e nal powe sou ce E he ne Se ial I2C E he ne Powe line Da a line Con ol line USV Omni-di ec ional An enna E he ne Figu e 5. A chi ec u e design o he USV. The an enna sys ems on bo h he USV and GC a e capable o e ical o a ion. Since he essel is loa ing on wa e , he il ing angle o he an enna may a y. To sol e his, a 6-axis senso is u ilized o es ima e he compensa ed e ical angle. The embedded compu e con ols he o he se o o o a e he an enna in o de o keep i s able agains Senso s 2021,21, 3871 7 o 20 he wa es. The se o mo o in he GC sys em s ee s he an enna owa d he lying UAV in he case when he sys em wo ks in elay communica ion mode. Single-boa d compu e Se o Mo o (Ele a ion) DC Mo o (Azimu h) GPS module USV Cen al Compu e GC Di ec ional An enna Powe egula o Ex e nal powe sou ce E he ne DC Mo o D i e Ro a y Encode Powe line Da a line Con ol line Figu e 6. A chi ec u e design o he GC. In NLOS scena ios, a UAV can be deployed o es ablish a elay link. The UAV s ee ing sys em includes an embedded compu e and a se o mo o . The compu e ecei es GPS coo dina es o he UAV, including la i ude, longi ude, and al i ude, o calcula e he il ing angle. Then, i con ols he se o o s ee he an enna e ically. Fo ho izon al o a ion, he yaw-con ol capabili ies o he d one a e used o keep he an enna cons an ly di ec ed owa d he GC. All he subsys ems ha e a sepa a e powe egula o o p o ide a sui able powe supply o each componen . Single-boa d compu e Se o Mo o (Ele a ion) UAV Di ec ional An enna Powe egula o UAV’s Ba e ies E he ne Powe line Da a line Con ol line UAV Omni-Di ec ional An enna GPS module E he ne USB Figu e 7. A chi ec u e design o he UAV. 5. Sys em P o o ype Implemen a ion The mechanical s uc u e o he communica ion pa o he USV and GC was designed in ou p e ious wo k [ 4 ]. The design o he an enna o a ion mechanism o he USV is desc ibed in de ail in [ 28 ]. In his wo k, he USV communica ion sys em was modi ied by using mo e igid ma e ials and mo e p ecise s ee ing pa s o imp o e he esis ance agains he wind and o inc ease he p ecision o o a ion, as shown in Figu e 8a. Two aluminum b acke s we e used o hold he an enna. This p e en s he an enna om swinging when he wind is s ong o when he au onomous essel is mo ing a high speed. Two gea s we e placed on he b acke o he e ical o a ion. One was connec ed o he se o mo o h ough a sha and he o he was ins alled on he sha o he an enna. A bel linked hose wo gea s o ansmi he o a y mo ion om he se o mo o o he an enna. In his Senso s 2021,21, 3871 8 o 20 ame e sion, wo ROBOTIS Dynamixel MX-28 se o mo o s we e used o ho izon al and e ical o a ion o he essel an enna. The MX-28 se o ea u es he acking capabili ies o i s speed, empe a u e, sha posi ion, ol age, and load. The sha posi ion can be main ained and modi ied acco dingly o each indi idual se o wi h he con ol algo i hm on he AX-12 ac ua o . I allows con olling he mo o ’s esponse in e ms o speed and s eng h. The se o’s buil -in mic ocon olle manages he con ol o senso s and an ennas. The se o p oduces a high s all o que o 1.5 Nm and high no-load speed o 60 RPM, which a e sui able o beam-s ee ing o a ion. (a) (b) Figu e 8. Final mechanical design o bo h an ennas on he USV and UAV. ( a ) Mechanical pla o m o he di ec ional an enna on he USV; (b) Moun ing pla o m o he di ec ional an enna o he UAV. The di ec ional an enna on he UAV was a ached o a ca bon- ibe pla o m. The de- sign o his ame ollowed he s uc u e o he UAV so ha i could be moun ed on o i , as demons a ed in Figu e 8b. This moun ing pla o m consis s o se e al pla es and has enough space o place o he elec ical de ices on i . The b acke s holding he an enna and he gea s we e 3D p in ed. The se o mo o sha was connec ed o one gea and he an enna was ins alled o he adjacen gea , which enabled he e ical o a ion. 5.1. Single-Boa d Compu e s and Mic ocon olle s The USV and GC a e bo h con olled by he Beaglebone G een single-boa d compu e s. They a e esponsible o exchanging GPS messages, unning beam-s ee ing algo i hms and execu ing measu emen sc ip s. Table 1shows he ha dwa e speci ica ion o he Beaglebone G een boa d. Table 1. Ha dwa e speci ica ions o he Beaglebone G een boa d. Fea u e Value P ocesso AM335x 1 GHz ARMR Co ex-A8 RAM 512 MB DDR3 On-boa d s o age 4 GB eMMC Accele a o suppo NEON loa ing-poin and 3D g aphics accele a o Mic o USB 1, o Powe ing and da a communica ion USB 1, o Hos ing GPIO 2 ∗46 pin heade s Ne wo king 1 E he ne po Ope a ing empe a u e 0 o 75 °C Senso s 2021,21, 3871 9 o 20 The UAV sys em is con olled by he Udoo X86 ULTRA e sion single-boa d compu e . One ask o he con olle on he UAV is o wa ding he ne wo k packe s in he elay communica ion mode. The Mik o ik SXT AC an enna and he Wi-Fi USB dongle we e, espec i ely, connec ed o he E he ne and USB po s o his boa d. Udoo X86 was selec ed due o i s Gigabi E he ne ne wo k in e ace and USB 3.0 o p o ide a high da a- ans e a e. In addi ion, he chosen embedded boa d is compa ible wi h A duino 101 pla o m so ha i can con ol he se o mo o using A duino so wa e. Besides, his boa d has powe ul ha dwa e ha can be u ilized in o he au onomous asks such as ideo p ocessing. The echnical speci ica ions o he Udoo X86 boa d a e p esen ed in Table 2. Ano he con olle used in he sys em was he A bo ix-M Robocon olle . This obo con olle is an ad anced solu ion o Dynamixel se os and o he high-accu acy obo ic ac ua o s. I inco po a es a obus AVR mic ocon olle , a wi eless in e ace, dual mo o d i e s, and 3-pin heade s o hobby se os wi h digi al and analog I/O. The A bo ix-M Robocon olle is o con olling he wo, newly ins alled Dynamixel MX-28 se o mo o s. Table 2. Ha dwa e speci ica ions o he Udoo X86 boa d (A ezzo, I aly). Fea u e Value CPU In el®Pen ium N3710 up o 2.56 GHz GPU In el®HD G aphics RAM 8 GB DDR3L Dual Channe Video in e aces 1∗HDMI 1.4 (CEC), 2∗Mini DisplayPo ++ On-boa d s o age 32 GB eMMC solde ed on-boa d Ne wo king 1∗Gigabi E he ne LAN in e ace 1∗M.2 Key E slo o op ional Wi eless Module Audio in e aces HD Audio Codec ALC283CG Mic ophone + Headphone Combo Connec o (TRRS) P eampli ied s e eo speake ou pu , S/PDIF ou pu USB 3∗USB 3.0 ype-A socke s O he in e aces 2∗HSUART po s, 2∗I2C in e ace, 1∗SDIO in e ace 1∗LPC in e ace Pla o m compabili y A duino™ 101-Compa ible h ough s anda d A duino™ Pins layou , compa ible wi h A duino™ shields 5.2. Powe Managemen The powe supply inpu a ies o all he de ices o he sys em. Thus, i is necessa y o ins all di e en powe egula o s o each speci ic de ice. Fo he USV sys em, powe was supplied by he ship accumula o and connec ed o he ol age egula o s in he communica ion module o he mo ing essel. The componen s on he UAV we e p o ided wi h he powe om he UAV’s ba e y. In o de o show he powe demands o u ilized equipmen , an o e iew on he inpu powe equi emen s o each componen is p esen ed in Table 3. The GC and USV powe supply sys ems ha e no changed om he p e ious design excep o he ins alla ion o AX-12 se os and he A bo ix-M Robocon olle , as shown in Figu e 8a. The inal elec ical design o he UAV is p esen ed in Figu e 9. Senso s 2021,21, 3871 16 o 20 GC UAV USV Figu e 16. UAV elay scena io in Tampe e Uni e si y, He an a campus. While he UAV was ai bo ne, he wo an ennas on he g ound (USV and GC) we e success ully connec ed and exchanged packe s o each o he using he elay link. The Udoo boa d ins alled on he UAV played he impo an ole o a ou ing de ice and o wa ded packe s be ween he USB Wi-Fi dongle and Gigabi E he ne in e ace connec ed o he Mik oTik an enna. 6.2.2. Measu emen Resul s The RSS le els a he UAV di ec ional an enna and omnidi ec ional an enna on he g ound a e shown in Figu e 17. The heigh o he UAV was cons an a 22.5 m h oughou he whole ligh . The e a e ou posi ions shown in Figu e 17. A he beginning o he ligh (posi ion 0), he UAV was al eady a he heigh o 22.5 m abo e he g ound. Howe e , he di ec ional an enna on he UAV was no aligned wi h he di ec ional an enna on he GC un il posi ion 2. Be o e ha , he links be ween all h ee pa s (GC, UAV, and USV) had al eady been es ablished, e en hough he RSS o he di ec ional an enna on he UAV was ela i ely low. This was due o he ac ha he dis ance be ween GC and UAV is sho compa ed o he scena io a a lake. The RSS o he omnidi ec ional an enna is highe han ha o he di ec ional an enna on he UAV. As he beam s a ed o be aligned a posi ion 2, he signal le el o he di ec ional an enna con inued inc easing un il posi ion 3. Figu e 17. RSS le el in he elay communica ion mode measu emen . The connec ion be ween he GC and USV depends on wo g ound- o-ai adio links— be ween GC and UAV, and be ween USV and UAV. The measu emen esul s show ha one o hem (wi h lowe RSS) will always be a bo leneck link, which limi s he h oughpu o he o e all elay sys em. The e o e, he posi ion o he UAV should be well selec ed o op imize he bo leneck adio link. 6.2.3. UAV Posi ioning Analysis In comme cial solu ions, he UAV mo es au onomously; howe e , he op imal elaying posi ion s ill needs o be calcula ed. The op imal loca ion can po en ially inc ease he RSS le els. In his sec ion, he simula ion e lec ing he RSS wi h a ying UAV posi ions is discussed and e alua ed. Senso s 2021,21, 3871 17 o 20 The elay scena io wi h he posi ion o he UAV a e illus a ed in Figu e 18. As discussed in Sec ion 3.2, he channel model used o simula e he GC–UAV and UAV–USV connec ions is he FSPL model, since he g ound- o-ai adio links in his si ua ion a e in an LOS and open-space en i onmen . In he simula ions, he scena io om Sec ion 6.1 is used as well as he pa ame e s. The l1 and l2 a e se o 3 km and 1 km, espec i ely. The UAV is assumed o ly a an al i ude h = 22.5 m abo e he g ound, which is highe han he a e age le el o ees and o he in as uc u es on he island. The alues d1 and d2 a e calcula ed based on he pa ame e s abo e. In he simula ion, he wo alues l1 and l2 a y om 0 o 4 km, which indica es he ins alled loca ion o he GC and mo emen o he USV. The RSS le els o he GC–UAV and UAV–USV links a e calcula ed based on he FSPL model and hen compa ed o each o he . The one wi h a lowe signal le el is he bo leneck, which will be plo ed agains d1and d2. Figu e 18. Simula ion scena io o ob aining he op imized UAV posi ion be ween GC and USV. The bo leneck RSS plo o he communica ion sys em is shown in Figu e 19. The la e shows ha he RSS inc eases as he d1 and d2 dis ance dec eases. A ea A in he igu e implies he cases when he bo leneck RSS only depends on d1 , ega dless o he d2 alue. In a ea B, he RSS alues a e subjec o he UAV–USV dis ance. Fo each cons an d1 , as he dis ance d2 be ween he UAV and he USV inc eases (c osses he ed line), he UAV–USV link becomes he bo leneck o he o e all sys em and he RSS alues become lowe han hose in a ea A. Hence, i is necessa y o keep he d2 dis ance o be on he le o he ed line. The ed line ep esen s he op imal RSS while he GC–UAV–USV dis ances a e g owing. To ind he ela ion be ween d1 and d2 on ha line, each (d1 , d2) pai on he ed line was collec ed and pa sed o a cu e i ing unc ion in Ma lab. The unc ion poly i () in Ma lab is used o his pu pose o calcula e he coe icien s o a polynomial ha bes i s he inpu da a. Then, he line can be es ima ed as d2= 0.6309 d1− 0.1954, wi h d1 and d2 measu ed in km. Thus, wi h he cu en an enna se up and ansmi e powe se ings, he UAV should keep i s dis ance o he USV o be equal o o sho e han 0.6309 d1− 0.1954 o op imize he o e all RSS o he sys em. The ela ion o op imal d1 and d2 can be exp essed as d2⩽0.6309d1−0.1954. Figu e 19. Bo leneck RSS o he o e all adio link. Senso s 2021,21, 3871 18 o 20 Using he es ima ed op imal posi ion o he UAV agains he USV, he simula ed bo leneck RSS o he elay link was compa ed o he measu ed RSS. The x-axis ea u es he ho izon al dis ance l be ween he GC and he USV, which means l=l1+l2 . Using ha , along wi h he heigh and he ela ion d2= 0.6309 d1− 0.1954, a pai o (l1 , l2) and (d1 , d2) can be ound o each dis ance l . F om hose alues, he pa h loss alues o GC–UAV and UAV–USV we e calcula ed and he lowe ones we e collec ed as a basis o RSS o be plo ed. Figu e 20 p esen s he men ioned compa ison. The USV loses he LOS a a dis ance beyond 3.5 km; he elay link ( ed cu e) wo ks be e han he di ec link (blue cu e) and becomes a solu ion o main ain he connec ion be ween he GC and he USV. Howe e , he di ec link pe o ms be e han he elay link a dis ances below 3.5 km. The eason o his is he lowe ecep ion gain o he USV and UAV an ennas used in he elay communica ion mode. In he LOS communica ion mode wi h di ec GC–USV connec ion, he an enna on USV is di ec ional and he gain o he main lobe is 23 dBi while in elay mode, he ecep ion gain is 16 dBi o he di ec ional an enna on he UAV and 12 dBi o he omnidi ec ional an enna on USV. The use o 12 dBi omnidi ec ional an enna on USV also limi s he possible UAV–USV dis ance, which c ea es imbalances be ween op imal dis ance alues o he elay mode componen s. While he UAV has o keep i s p oximi y o he au onomous essel, he ange be ween he GC and he UAV becomes signi ican ly highe han he UAV–USV dis ance. The 16 dBi di ec ional an enna on he UAV canno compensa e ha dis ance o p o ide an adequa e signal s eng h compa ed o he 23 dBi one in he LOS mode, which leads o an un ai compa ison o he LOS mode and elay mode in e ms o equipmen se ings. Figu e 20. Compa ison o measu ed RSS and simula ed elay RSS. The e o e, ano he compa ison be ween LOS mode and elay mode was made wi h equal condi ions o GC and USV. Assuming ha he e ec i e adia ed powe om he ansmi e (GC) a e he same o bo h cases, he USV will now be assumed o use he 23-dBi di ec ional an enna on a UAV–USV link. Using he same app oach men ioned p e iously, he simula ed elay RSS is illus a ed wi h a yellow cu e. I is clea ha he elay mode sys em wi h he 23-dBi di ec ional an enna on USV pe o ms be e han he LOS mode sys em when he essel is 2-km a he han he GC. In his case, wi h he highe ecei ed gain in USV, he ange o UAV agains he USV is ex ended and he GC–UAV and UAV–USV links in elay mode a e now mo e balanced in e ms o op imal ange. Knowing he op imal posi ion o he UAV in e e ence o he essel p o ides ideas on how he UAV should be used in an au onomous collabo a i e o sho e ne wo k. To p e en he signal loss, o session in e up ion in he NLOS case, he essel should p edic ha he e will be obs acles blocking he channel based on he map calcula ions and hen deploy he UAV in ad ance o main ain he connec ion be ween he GC and he USV. Senso s 2021,21, 3871 19 o 20 A e being launched, he UAV ollows he essel wi hin a ange ha is calcula ed based on he dis ance be ween he USV and he GC using he op imal posi ion in es iga ed in his sec ion. Ano he usage o he UAV is o ex end he co e age o he wi eless sys em. Wi h highe USV an enna gain, he elay mode sys em can pe o m be e han he LOS mode link in a he dis ances. The e o e, he essel can deploy he UAV o imp o e he connec ion quali y when he RSS d ops below a de ined alue. 7. Conclusions In his a icle, we p esen ed he esul s o ou esea ch wo k, which aimed o design, implemen , and es a wi eless au onomous collabo a i e communica ion sys em o eme g- ing ma i ime use cases. The co e componen s o he p oposed sys em a e GC, USV, and UAV. To suppo o sho e moni o ing and unmanned ope a ions, he designed sys em is based on high-speed, long- ange communica ions be ween hese componen s. Taking in o accoun he LOS/NLOS scena ios be ween he GC and USV, wo communica ion modes we e de ined: an LOS communica ion mode, which implies a di ec LOS link be ween he GC and USV; and a elay communica ion mode, whe e a UAV is deployed o ac as a elay in NLOS scena ios. Each unmanned ehicle in he p oposed au onomous collabo a i e sys em is equipped wi h communica ion modules ha inco po a e wi eless in e aces, di ec ional an ennas, o a ion mechanisms o mechanical beam-s ee ing, and con olle s o da a collec ion and p ocessing. Fu he de ails on he design and implemen a ion o ou communica ion sys em we e p o ided in his a icle. We also assessed he pe o mance o he designed and implemen ed links using measu emen campaigns. Addi ionally, simula ions we e u ilized— i s , o compa e he measu emen da a o analy ical esul s based on he e- iewed p opaga ion models; second, o de e mine he op imal UAV posi ion in he elay communica ion mode. Po en ial enhancemen s o ou communica ion sys em include conduc ing he mea- su emen s on he elay communica ion mode in a ma i ime en i onmen , and de eloping a hando e mechanism be ween he wo de ined communica ion modes o main ain con- inuous connec ions be ween he unmanned ehicles and he con ol s a ion. A nex s ep in he implemen a ion o he o e all au oma ed o sho e sys em depic ed in Figu e 1can be he design and implemen a ion o communica ion links wi h he AUV. Including he unde wa e componen will enable a ully in eg a ed collabo a i e sys em wi h mo e au onomous ehicles and u he suppo o eme ging ma i ime use cases. Au ho Con ibu ions: Concep ualiza ion, J.P., K.M., M.G.; me hodology design, S.S., J.F.; p o o ype implemen a ion, J.P., K.M., J.F., J.V., M.G.; expe imen al e alua ions, K.M., J.V., M.G.; esul s ali- da ion, J.P., S.S., J.H.; w i ing—o iginal d a p epa a ion, J.P., K.M., S.S., J.F.; w i ing— e iew and edi ing, J.V., M.G., J.H.; supe ision, M.G., Y.K., J.H.; p ojec adminis a ion, Y.K.; unding acquisi ion, M.G., Y.K. All au ho s ha e ead and ag eed o he published e sion o he manusc ip . Funding: This pape is based on he Au onomous and Collabo a i e O sho e Robo ics (aCOLOR) p ojec unded by he Technology Indus ies o Finland Cen ennial and Jane & Aa os E kko Foun- da ions unde Fu u e Make s Funding P og am 2017. The au ho s g a e ully acknowledge he con ibu ions o he company Alama in-Je Oy. The desc ibed esea ch was inanced by he Minis y o Indus y and T ade o Czech Republic p ojec No. FV40309. The au ho s also acknowledge unding om he Eu opean Union’s Ho izon 2020 Resea ch unde he Ma ie Sklodowska Cu ie g an ag eemen No. 813278 (A-WEAR p ojec ). Ins i u ional Re iew Boa d S a emen : No applicable. In o med Consen S a emen : No applicable. Da a A ailabili y S a emen : No applicable. Con lic s o In e es : The au ho s decla e no con lic o in e es . Senso s 2021,21, 3871 20 o 20 Re e ences 1. Sulli an, B.P.; A ias Na a, E.; Desai, S.; Sole, J.; Rossi, M.; Ramundo, L.; Te zi, S. De ining Ma i ime 4.0: Reconciling p inciples, elemen s and cha ac e is ics o suppo ma i ime essel digi alisa ion. IET Collab. In ell. Manu . 2021. [C ossRe ] 2. Sanchez-Gonzalez, P.L.; Díaz-Gu ié ez, D.; Leo, T.J.; Núñez-Ri as, L.R. Towa d digi aliza ion o ma i ime anspo ? Senso s 2019,19, 926. [C ossRe ] 3. Zolich, A.; Palma, D.; Kansanen, K.; Fjø o , K.; Sousa, J.; Johansson, K.H.; Jiang, Y.; Dong, H.; Johansen, T.A. Su ey on communica ion and ne wo ks o au onomous ma ine sys ems. J. In ell. Robo . Sys . 2019,95, 789–813. [C ossRe ] 4. Khos a i, Z.; Ge asimenko, M.; U ama, J.; Pya ae , A.; Escusol, J.V.; Hosek, J.; And ee , S.; Kouche ya y, Y. Designing high-speed di ec ional communica ion capabili ies o unmanned su ace ehicles. In P oceedings o he 2019 16 h In e na ional Symposium on Wi eless Communica ion Sys ems (ISWCS), Oulu, Finland, 27–30 Augus 2019; pp. 651–655. 5. Ma, K. Implemen a ion and E alua ion o Communica ion Sys em o Au onomous O sho e Vehicles. Mas e ’s Thesis, Tampe e Uni e si y, Tampe e, Finland, 2020. 6. Vala anis, K.P.; Vach se anos, G.J. Handbook o Unmanned Ae ial Vehicles; Sp inge : Be lin, Ge many, 2015; Volume 1. 7. Salhaoui, M.; Gue e o-González, A.; A ioua, M.; O iz, F.J.; El Oualkadi, A.; To eg osa, C.L. Sma indus ial IoT moni o ing and con ol sys em based on UAV and cloud compu ing applied o a conc e e plan . Senso s 2019 ,19, 3316. [C ossRe ] [PubMed] 8. Dela a pou , N.; Kopa an, C.; Nowa zki, J.; Bajwa, S.; Sun, X. A Technical S udy on UAV Cha ac e is ics o P ecision Ag icul u e Applica ions and Associa ed P ac ical Challenges. Remo e Sens. 2021,13, 1204. [C ossRe ] 9. Mehallegue, N.; Djellab, M.; Loukhaoukha, K. E icien Use o UAVs o Public Sa e y in Disas e and C isis Managemen . Wi el. Pe s. Commun. 2021,116, 369–380. [C ossRe ] 10. Global Comme cial D one Ma ke Size in 2018 and 2024. A ailable online: h ps://www.s a is a.com/s a is ics/878018/global- comme cial-d one-ma ke -size/ (accessed on 28 May 2021). 11. Moza a i, M.; Saad, W.; Bennis, M.; Nam, Y.H.; Debbah, M. A u o ial on UAVs o wi eless ne wo ks: Applica ions, challenges, and open p oblems. IEEE Commun. Su . Tu o . 2019,21, 2334–2360. [C ossRe ] 12. Ca illo, D.; Mikhaylo , K.; Na delli, P.J.; And ee , S.; da Cos a, D.B. Unde s anding UAV-Based WPCN-Aided Capabili ies o O sho e Moni o ing Applica ions. IEEE Wi el. Commun. 2021. [C ossRe ] 13. Alzenad, M.; El-Keyi, A.; Lagum, F.; Yanikome oglu, H. 3-D placemen o an unmanned ae ial ehicle base s a ion (UAV-BS) o ene gy-e icien maximal co e age. IEEE Wi el. Commun. Le . 2017,6, 434–437. [C ossRe ] 14. Lai, C.C.; Chen, C.T.; Wang, L.C. On-demand densi y-awa e ua base s a ion 3d placemen o a bi a ily dis ibu ed use s wi h gua an eed da a a es. IEEE Wi el. Commun. Le . 2019,8, 913–916. [C ossRe ] 15. Li, M.; Yu, F.R.; Si, P.; Yang, R.; Wang, Z.; Zhang, Y. UAV-Assis ed Da a T ansmission in Blockchain-Enabled M2M Communica ions wi h Mobile Edge Compu ing. IEEE Ne w. 2020,34, 242–249. [C ossRe ] 16. Zeng, Y.; Zhang, R.; Lim, T.J. Wi eless communica ions wi h unmanned ae ial ehicles: Oppo uni ies and challenges. IEEE Commun. Mag. 2016,54, 36–42. [C ossRe ] 17. Khawaja, W.; Gu enc, I.; Ma olak, D.W.; Fiebig, U.C.; Schneckenbu ge , N. A su ey o ai - o-g ound p opaga ion channel modeling o unmanned ae ial ehicles. IEEE Commun. Su . Tu o . 2019,21, 2361–2391. [C ossRe ] 18. Sánchez-Ga cía, J.; Reina, D.; To al, S. A dis ibu ed PSO-based explo a ion algo i hm o a UAV ne wo k assis ing a disas e scena io. Fu u e Gene . Compu . Sys . 2019,90, 129–148. [C ossRe ] 19. Magán-Ca ión, R.; Camacho, J.; Ga cía-Teodo o, P.; Flushing, E.F.; Di Ca o, G.A. A Dynamical Relay node placemen solu ion o MANETs. Compu . Commun. 2017,114, 36–50. [C ossRe ] 20. Zhan, C.; Zeng, Y.; Zhang, R. Ene gy-e icien da a collec ion in UAV enabled wi eless senso ne wo k. IEEE Wi el. Commun. Le . 2017,7, 328–331. [C ossRe ] 21. Xu, J.; Zeng, Y.; Zhang, R. UAV-enabled wi eless powe ans e : T ajec o y design and ene gy op imiza ion. IEEE T ans. Wi el. Commun. 2018,17, 5092–5106. [C ossRe ] 22. E Sil a, T.D.; de Melo, C.F.E.; Cumino, P.; Rosá io, D.; Ce quei a, E.; De F ei as, E.P. STFANET: SDN-based opology managemen o lying ad hoc ne wo k. IEEE Access 2019,7, 173499–173514. [C ossRe ] 23. Zhao, Z.; Cumino, P.; Souza, A.; Rosa io, D.; B aun, T.; Ce quei a, E.; Ge la, M. So wa e-de ined unmanned ae ial ehicles ne wo king o ideo dissemina ion se ices. Ad Hoc Ne w. 2019,83, 68–77. [C ossRe ] 24. How To Calcula e Dis ances, Azimu hs and Ele a ion Angles O Peaks. A ailable online: h p:// ches e .o g/sgm/analysis/ peaks/how_ o_ge _ iew_pa ams.h ml (accessed on 28 May 2021). 25. Pa sons, J.D. The Mobile Radio P opaga ion Channel; Wiley: Hoboken, NJ, USA, 2000. 26. Cui, Z.; B iso, C.; Guan, K.; Ma olak, D.W.; Cal o-Ramí ez, C.; Ai, B.; Zhong, Z. Low-al i ude UAV ai -g ound p opaga ion channel measu emen and analysis in a subu ban en i onmen a 3.9 GHz. IET Mic ow. An ennas P opag. 2019 ,13, 1503–1508. [C ossRe ] 27. Yee Hui, L.; Dong, F.; Meng, Y.S. Nea sea-su ace mobile adiowa e p opaga ion a 5 GHz: Measu emen s and modeling. Radioenginee ing 2014,23, 824–830. 28. Mo ón Alguacil, C. Design and Analysis o Di ec ional An enna S uc u e o Unmanned Su ace Vessel. Mas e ’s Thesis, Tampe e Uni e si y, Tampe e, Finland, 2019.