ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 134 ACCEPTED FROM OPEN CALL TRITON: HIGH-SPEED MARITIME WIRELESS MESH NETWORK MING-TUO ZHOU, VINH DIEN HOANG, AND HIROSHI HARADA, NATIONAL INSTITUTE OF INFORMATION AND COMMUNICATIONS TECHNOLOGY JAYA SHANKAR PATHMASUNTHARAM, HAIGUANG WANG, PENG-YONG KONG, CHEE-WEI ANG, YU GE, AND SU WEN, INSTITUTE FOR INFOCOMM RESEARCH ABSTRACT This article presents the TRI-media Telematic Oceanographic Network (TRITON) project, which aims to develop a high-speed and low-cost maritime communication system The article includes information pertaining to background studies, high-level architecture, network feasibility, maritime communication environment, technology developments, prototype implementations and link performance measurements The motivation for this project stems from the fact that there is an increasing need for low-cost and high-speed maritime communication, with demands mainly coming from regulatory and crew welfare needs The system described in this article is a wireless mesh network based on the IEEE 802.16 An analysis of the node connectivity based on real ship traffic data shows that the network is feasible in shipping lanes with a high density of ships The system also considers the use of an intelligent middleware to allow communications to switch back to a satellite link in cases where neighboring ships are sparse or at locations far away from mesh base stations Protocol enhancements to both the Medium Access Control (MAC) and networking layers and a hardware design that features multiple transceivers and the implementation of antenna switching to counter sea wave reflection and rocking problems are presented Measurements of field trials show that the proposed wireless mesh network could be an effective solution for maritime communications INTRODUCTION Peng-Yong Kong is currently with Khalifa University of Science, Technology & Research (KUSTAR) The work was done when he was with the Institute for Infocomm Research 134 With the recent growth in the maritime industry stemming from an increase in global commerce, as well as the increase in the search for hydrocarbons in offshore locations, there has been an increase in the demand for affordable and highspeed maritime communications New demands from crew welfare solutions, offshore companies that need to coordinate their operations, and international maritime regulators such as the International Maritime Organization (IMO), 1536-1284/13/$25.00 © 2013 IEEE who are looking toward the timely dissemination of digital navigational charts, have also contributed to this increased demand Unfortunately, the current maritime networks, which are mainly based on conventional High Frequency (HF), Very High Frequency (VHF), and Ultra High Frequency (UHF) radios for communications near coasts and satellite systems for longrange ship-to-ship/shore communications are much slower and expensive than the wireless networks deployed on land In recent years, some new systems have emerged to provide maritime users with cheaper and faster communications, such as WISEPORT (WIrelessbroadband-access for SEaPORT) network in Singapore, digital VHF network in Norway, and maritime systems based on Very Small Aperture Terminal (VSAT) However, these systems are still limited in coverage distance (WISEPORT network — 15 km from coastline), speed (digital VHF system — 21 kb/s and 133 kb/s), or cost (VSAT system — expensive stabilized antennas) In order to meet the increasing needs and to alleviate the issues, a project TRI-media Telematic Oceanographic Network (TRITON) was launched recently, with the objective to develop a system for high-speed and low-cost maritime communications in narrow water channels and shipping lanes close to the shore This article serves to briefly introduce the analysis and developments we have done in TRITON The envisaged system is a wireless mesh network based on the IEEE 802.16 standard Analysis of ship node connectivity and route redundancy based on real ship traffic data shows that the envisaged network is feasible in regions with high ship density such as Malacca Strait In order to successfully develop such a maritime mesh network, a number of challenges such as sea surface movement and strong sea wave reflections, need to be overcome Therefore, TRITON firstly developed a maritime simulator based on derived/verified models (e.g., of sea surface movement, maritime radio channel, and ship mobility) and secondly used the simulator as a tool for development of Medium Access Control (MAC) and network layer protocols that are suitable for maritime IEEE Wireless Communications • October 2013 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 135 environments Thirdly, a middleware is developed for switching maritime mesh network back to legacy satellite system in case the ships are too sparse or too far away from the onshore base stations to provide mesh connection to land Finally, in hardware implementation, mesh node with antenna unit capable of countering sea wave movement and reflections, and 360˚reception/transmission has been developed Field trial shows that with the developed protocols and hardware, good link performance can be achieved on sea In literatures, it was reported in Japan a maritime mesh network operating in 27MHz/40MHz has been developed and the coverage range can be 70 km from shore [1]; however, the channel speed is only about kb/s, which is far from requirement of high-speed communications Other reported maritime mesh or ad hoc networks are now in phase of concept or preliminary study [2–5] This article is organized as follows First, high-level architecture and network feasibility study of the envisaged network are introduced We then present studies of the maritime communication environment Development of MAC and network layer protocols are described, and then details pertaining to the implementation of the prototype mesh node and field trial measurement results are presented Finally, our conclusion and some discussion are presented HIGH-LEVEL ARCHITECTURE AND NETWORK FEASIBILITY HIGH-LEVEL ARCHITECTURE Figure shows the high level architecture of TRITON The ships form a mesh network and are connected to terrestrial networks via onshore stations that are regularly placed along shore near shipping lanes Buoys are placed to relay traffic when necessary This network is ideal when there are sufficient ships to provide good mesh connectivity In locations where ships are sparse and difficult to connect with onshore stations via mesh links, the mesh radio installed onboard can fall back on a satellite communication link A mesh network is envisaged for maritime communications because: Firstly, the network coverage can be easily expanded as a mesh node can route and relay traffic for other nodes Secondly, a mesh network has resilient paths and is capable of self-healing This is important in maritime wireless networks since a routing path may fail often due to sea wave rocking and occlusions Thirdly, a mesh network can easily increase its capacity by simultaneous transmissions over nodes that are separated several hops away IEEE 802.16-2004 mesh MAC is adopted In comparison to other mesh technologies such as 802.11s and 802.15.5, 802.16-2004 mesh MAC is time-division-multiplexing (TDMA) based and can support a much longer transmission range It is easier to support guaranteed quality-of-service (QoS) with a TDMA-MAC Moreover, 802.16 supports much longer communication distance, IEEE Wireless Communications • October 2013 Land station Satellite Terrestrial network Land station Land station Land Satellite station Ship Bouy Bouy Ship Ship Oil platform Ship Ship Sea Ship Ship Figure The high-level architecture of a multihop wireless mesh network for maritime communications which is more suitable for maritime communications NETWORK FEASIBILITY STUDY The success of a maritime wireless mesh network will depend on the density of the ships since a mesh system requires cooperative relay transmissions Fortunately, there are potential areas in the world with high ship densities and shipping routes close to a shoreline, e.g., the Malacca Strait In order to verify the maritime mesh network concept, network feasibility has been studied based on actual ship mobility data and measured achievable 802.16 radio connection distance on sea The analysis includes ship node connectivity and route redundancy of a maritime mesh network The connection distance measurement was carried out in the East Coast of Singapore (a part of Malacca Strait), using a pair of 802.16 transceiver operating at 5.8 GHz (LIBRA 5800) Measured achievable connection distance (for BER < 10 –6) between two ships on sea is 8.66 km, and the achievable connection distance between an onshore station (Bedok Light House with height of 76 m) and a ship on sea is 14.2 km The measurement setup is BPSK modulation and the effective data rate is Mbps The sea region studied for ship node connectivity and mesh route redundancy is a 35.3-kmlong and 16.7-km-wide rectangular area close to the East Coast of Singapore Figure shows snapshot of ship locations within the region of interest The ship mobility trace is derived from Automatic Identification System (AIS) data provided by the Maritime and Port Authority of Singapore (MPA) The AIS data consists of instantaneous information of ships such as position and velocity, etc Based on the achievable link measurement, it is assumed that the maximum connection distance between an onshore station and a ship is 15 km, and the maximum connection distance between two ships is km This assumption is applicable since usually ships are restricted in shipping lanes which are certain far from shore and parallel to shore line A ship is said to be connected to the main network, if it has at least one one-hop, or one two-hop, or one three-hop route to an onshore station A route 135 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 136 Movement of sea surface causes ships and buoys, and thus antennas installed on them, to move in various ways: pitch and roll, and rise or fall It then leads to misalignment of the transmitter and receiver antennas, as well as changes of the effective antenna heights Figure A snapshot of ship locations within the region of interest; station is the onshore base station; others are ship stations on sea with four hops or more is not counted in as an effective connection It is found that all of the 44 ship nodes recorded in the studied region can be connected to the terrestrial network in 98.91 percent of the 4-hour observation time This is higher than the expected criteria for good mesh connectivity, in which 90 percent of the nodes are connected for 90 percent of the time A wireless link in maritime environment is unstable in some degree due to ship rocking, and may result in break of end-to-end route between a ship node and an onshore station Route redundancy is therefore important to realize a maritime mesh network When there is multiple routes connection to an onshore station, a ship node may switch to use another route when the serving one breaks, thus the mesh network is robust Analysis shows that in the studied sea area, with the same assumption of the 802.16 radio connection distances as before, averagely, a ship node has about 17 routes to the onshore station, meaning that plenty of optional routes available when the serving one fails Further study shows that in the sea area, the average hop number of the routes is 1.8, indicating usually a route from a ship node to the onshore station is no more than two hops Note that in above analysis, maximum route hop number is limited to three in order not to incur excessive end-to-end delay and overhead Details of network feasibility study can be found in [6, 7] published by the authors Similar analysis of maritime mesh ship connectivity in Taiwan water with real AIS data also shows feasibility of mesh networking in near-shore waters [8] Please note that the above analysis of network feasibility is based on measurements at 5.8 GHz At lower VHF/UHF frequencies, even with the same transmission power and antenna heights, signal transmission can be much further, e.g., tens kilometer [9], meaning that in areas with relatively more sparse ships a mesh network could be feasible, too 136 MARITIME COMMUNICATION ENVIRONMENT STUDY SEA SURFACE MOVEMENT Effect of Sea Surface Movement — In contrast to the land surface, the sea surface is rough and random Movement of sea surface causes ships and buoys, and thus antennas installed on them, to move in various ways: pitch and roll, and rise or fall It then leads to misalignment of the transmitter and receiver antennas, as well as changes of the effective antenna heights Misalignment of onboard antennas is challengeable as high-gain directional antennas with narrow beamwidth will be used for maritime communications to overcome large path loss due to long distance The signal strength variation due to misalignment of antennas can be quite significant Our measured data shows that the pitch angle and the roll angle of a ship on sea varies in a range of [–5˚, 5˚] and [–10˚, 10˚], respectively Dimension of the measurement ship is about 15 meter long, meter wide and meter high (to sea surface) The experimental site is in East Coast of Singapore, about km far from shore The wave height in this area is generally less than one meter The change of effective antenna heights, in general case, leads to not significant variations of receive signal strength due to relatively long radio distance compared to antenna height However, when two-ray channel model applies, at distances of path loss spikes, the radio propagation loss is very sensitive to changes of the effective antenna height A small change of one meter of effective antenna height may cause as high as 16 dB path loss change [10] Unfortunately, as shown by our and some other’s measurements, in open sea environment, the channel path loss behavior tallies very well a two-ray channel model [11], meaning that measures must be taken to overcome this challenge Moreover, sea surface movement can also lead to wave occlusions, which may break the IEEE Wireless Communications • October 2013 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 137 communication links The duration of these link breaks can be several seconds because of the long period of sea waves In this period, increasing the maximum number of retransmissions before discarding a packet is not an effective way to improve the performance, and other effective methods must be applied Effect of sea wave movement on BER performance has been studied by measurement with 802.16 radio (LIBRA 5800) in Singapore At the same receiver power (–84 dBm) where BER of 10–6 is achievable on land, BER of a ship-to-ship and an onshore station-to-ship communication connection is about 4.0E-5 and 1.0E-5, respectively, which becomes much worse Detailed research of sea surface movement effect on received signal strength can be found in the author’s publication of [10] Modeling Sea Surface Movement — In order to build an “ocean terrain” in simulation software Qualnet for network performance studies and new protocols development for the envisaged maritime mesh network, a model of sea surface movement has been developed in TRITON [12] The sea surface movement is modeled as Trochoid wave, which is a traveling wave created by tracing the path of a point inside a circle [8] The model parameters are derived from PiersonMoskowitz sea state [14], a method describes roughness of sea surface A sea state is characterized by three parameters: significant sea wave height (H), average sea wavelength (λ), and average sea wave period (T) Generally, there are 10 sea states, numbered from to The sea region close to East Coast of Singapore, where many of our experiments carried out, can be classified as sea state (H = 0.4~0.9 m, l = 6~12 m, and T = 3.5~4.5 s) Based on the above model, movement of sea surface can be simulated, so as the movement of the ships and the orientation of the onboard antennas CHANNEL PROPERTIES Radio channel properties on sea are closely related to the propagation environment In open sea and wide shipping lanes, when sea wave height is up to one meter and LOS dominates, radio channel at frequencies of interest for broadband communications (e.g., 1.9 GHz, 2.3 GHz, and 5.8 GHz) can be modeled as a two-ray channel with a direct LOS path and a sea surface-reflected path It is confirmed by our measurement at both 5.8 GHz and 2.43 GHz [6] and experiment at 2.45 GHz reported by others Theoretically, this is because sea water is a good conductor when the radio frequency is below GHz As presented in [12], we mainly use two-ray channel model in our Qualnet simulations of maritime mesh networks since our interested region is wide shipping lane, e.g., the region for studying maritime mesh network feasibility SHIP MOBILITY PATTERN The mobility pattern for ships is critical in planning a maritime wireless mesh network for a specific region This topic was investigated during the course of the TRITON project as almost IEEE Wireless Communications • October 2013 no such research has been reported A ship mobility pattern includes statistical distribution information about ship speeds and inter-arrival times of ships at a given reference border line on earth AIS information provided by the Singapore MPA is used in the analysis of ship mobility patterns in TRITON The region studied is the same as the one for network feasibility study By fitting the curves of the probability distributions in a way of non-linear regression, the mathematical models for ship speed and inter-arrival time are found The modeled probability distribution function (PDF) of ship speed is a factorized truncated normal function, and the PDF of inter-arrival time is in the form of p(x) = a × bx, where x is the inter-arrival time, and a and b are arbitrary constants derived from the actual mobility data Details of this study can be found in [6] MARITIME SIMULATOR A maritime simulator has been developed with Qualnet software based on the developed models of sea surface movement, radio channel, and ship mobility It is able to simulate effective antenna gain affected by ship rocking, path loss, and ship distribution and movement on sea Meanwhile, 802.16-2004 mesh MAC has been implemented using Qualnet Further developments on MAC and network layer protocols have been carried out based on the maritime simulator A maritime simulator has been developed with Qualnet software based on the developed models of sea surface movement, radio channel, and ship mobility It is able to simulate effective antenna gain affected by ship rocking, path loss, and ship distribution and movement on sea PROTOCOL ENHANCEMENT AND DEVELOPMENTS In TRITON, a list of protocols on MAC and network layer have been enhanced and developed with considerations of the unique characteristics of maritime mesh networks Extensive simulations with the developed “maritime simulator” show that the designs are effective for maritime mesh communications MAC LAYER 802.16-2004 Mesh MAC — Instead of coordinated centralized scheduling (CCS) scheme, TRITON employs coordinated distributed scheduling (CDS) scheme defined in 802.16-2004 Mesh, as CDS schedules resources among mesh nodes in a distributed manner and is more suitable for a dynamic maritime mesh network topology To implement CDS, Mesh-Election algorithm is used for scheduling transmission of control messages, since this algorithm minimizes transmission collisions of the control messages Moreover, a three-way handshake process: request–assignment–confirmation is used for resource application By this approach, two-hop neighbors are aware of resource allocation due to the broadcasted messages, and then resource collisions and interference can be minimized Distributed Adaptive Time Slot Allocation — The above three-way handshaking process defines a framework for data resource requests and grants However, it lacks a detailed algorithm for resource allocations TRITON developed a distributed adaptive time slot allocation (DATSA) 137 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 138 Antenna array Sector 4: Sector 3: Sector 2: 270°~360° 180°~270° 90°~180° Sector 1: 0°~90° Tx/Rx control PHY Tx/Rx Tx/Rx control Switching block PHY Tx/Rx Tx/Rx control PHY MAC Tx/Rx Tx/Rx control PHY Tx/Rx MCU Gyro Frontend antenna unit Backend processing unit Figure Block diagram of a mesh node implemented with antenna switching for maritime communications algorithm to fill this gap [13] With this method, a mesh node keeps track of the availability of its own and its one-hop neighbor’s data resource Only data resource in which a node is available and all of its one-hop neighbors are available can be granted or accepted This algorithm can also detect resource overlap induced by change of network topology due to nodes movement When this happens, it adaptively re-allocates data resource with an offset to the original conflicted allocation Multi-Channel Transmission — In a maritime mesh network, multi-channel transmission is necessary in a case where a single frequency channel cannot provide sufficient bandwidth This is especially helpful in an 802.16 mesh network since it only allows one transmission-reception in a single channel in a two-hop neighborhood TRITON developed a multi-channel transmission algorithm and simulation results show that it can obviously improve the network performance in terms of the packet delivery ratio and average packet delay [14] Fair Bandwidth Allocation — The 802.16 mesh MAC is reservation-based A new flow would have to reserve resource for transmission hop-by-hop along a route established by the network layer The data resource reserved is used exclusively for transmissions between two nodes in a twohop neighborhood If a flow reserves most of the capacity, some later flows may not have enough resource to use This issue is referred to as the inter-flow fairness problem Medium contention also possibly occurs within a flow — between hops in a path As the earlier hops are reserved prior to the later hops, it is possible that the later hops are starved of bandwidth by the earli- 138 er hops, leading to zero end-to-end throughput This is referred to as the intra-flow fairness problem In TRITON, mechanisms and protocols have been designed to address these fairness problems, including max-min fairness algorithm, auto-rate-control method, flow-based resource reservation, and equal-throughput allocation along a flow These designs have minimized changes to the standards specifications Furthermore, it curbs oscillations in reservations between nodes and flows, and converges quickly NETWORK LAYER Routing Protocols — In a maritime mesh network, a connection link may break often because of wave rocking and occlusion This may lead to frequent change of routes although a ship node may be only two-hop away from the onshore BS And then a routing protocol is very important to ensure connectivity and reliable packet delivery In [15], we evaluated a number of routing protocols but found that none of them is ideal for maritime mesh network Geographical routing protocols may be used due to availability of Global Positioning System (GPS); however, a geographical route selected based on locations may have poor signal quality due to ship and wave blocking TRITON developed and uses a new routing protocol called MAC-based Routing Protocol for TRITON (MRPT) MRPT is an optimization of the classical routing algorithms tailored to the features of an 802.16 mesh network and the requirements of maritime communications The design and features of the MRPT protocol include: • It is proactive • It delivers routing information by piggybacking on mesh MAC control messages • It maintains multiple routing paths Because MAC control messages are periodically transmitted in dedicated control slots in a collision-free manner, the overhead and delay for the spread of routing information can be greatly reduced As in a maritime mesh network, mesh nodes are mobile and the network performance is sensitive to routes, MRPT considers both routing cost and stability in selecting a route The hop count is set as the routing cost, and MRPT uses the Received Signal Strength (RSS) as the stability metric A link is only eligible to be chosen for routing if its RSS value is above an upper threshold A mesh node keeps several candidate links for next hops, and when a better route appears, it switches to it immediately If a link breaks, a mesh node consults its list of backup links and chooses the one with the least hop counts among the links with a sequence number larger than the broken one Simulation results show that MRPT is better than the routing protocols OLSR, AODV, and AOMDV in initial packet delay, average packet delay, and throughput in maritime mesh networks Details of this development are published in [16] Intelligent Network Switch Middleware — In order to maintain stability and backward compatibility with existing maritime communication facilities, intelligent middleware is developed in TRITON IEEE Wireless Communications • October 2013 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 139 to merge high speed maritime mesh networks with legacy satellite communications The design objective of the integrated system is quality-ofservice (QoS) provision and cost control for maritime users Consequently, the integrated system consists of a high speed maritime mesh network, satellite communication devices, and middleware components, and the equipment on board a ship includes a legacy satellite modem, intelligent middleware, and a maritime mesh node The intelligent middleware acts as an access router for local onboard applications that may have different bandwidths and service requirements When the end-to-end connectivity of the maritime mesh network can be established between a specific ship and a land base station, the application may be forwarded through the mesh network to provide a lower cost and higher bandwidth On the other hand, in case where the end-to-end connectivity cannot be established due to low node density or a bad channel condition, satellite communications may be used as an alternative method of communication for onboard users The prototype system developed shows that the intelligent system can provide service continuity in a harsh environment when the end-to-end connectivity of a mesh network cannot be guaranteed, while minimizing the cost incurred by satellite communications Reference [17] presents detail designs of the middleware PROTOTYPE IMPLEMENTATION AND FIELD TRIAL MEASUREMENTS A mesh node prototype has been implemented in TRITON and link performance has been tested in actual maritime environments A node prototype consists of a specially designed switchable antenna unit and transceiver cards embedded protocols enhanced and developed for maritime mesh communications MESH NODE IMPLEMENTATION As shown in Fig 3, an implemented mesh node mainly consists of two parts: a backend processing unit and a frontend antenna unit The former is a computing system running Linux operation system that supports four transceiver interface cards and loads MAC implementation and time synchronization module, and the later is a 360˚-reception/transmission antenna array that is capable of countering problems introduced by strong radio reflection, sea wave movement, and so on The time synchronization module makes use of pps pulse produced by GPS receiver and slot interrupts produced by a timing circuit to achieve high accuracy of sync between different nodes which is a core requirement for TDMA MAC to function correctly Below introduces the antenna unit in more details as it is a critical part to make the radio to suit the maritime mesh networks In a mesh network, where relay is performed, there is strong reliance on the use of broadcast messages This means the design of the mesh node has to have 360° reception In addition to this, the transmission of broadcast packets has to IEEE Wireless Communications • October 2013 Parameter Value Frequency 5.8 GHz Bandwidth 20 MHz Raw data rate Mb/s Modulation & coding QPSK 1/2 Transmitter power (EIRP) 4W Onshore station antenna height 8m Onboard antenna height 4.5 m Receiver antenna gain 16 dBi Receiver sensitivity –83 dBm Operation mode Half-duplex A mesh node prototype has been implemented in TRITON and link performance has been tested in actual maritime environments A node prototype consists of a specially designed switchable antenna unit and transceiver cards embedded protocols enhanced and developed for maritime mesh communications Table Parameters for measurements be done in all directions so that all of the neighbors are aware of the MAC states The simplest way to achieve 360˚ reception/transmission is to use omnidirectional antenna However, in this case, reflections by sea surface or metal ship bodies may introduce severe deconstructive interference, especially in open sea environment where exists a strong reflection wave due to tworay channel property In TRITON 360˚-reception/transmission is achieved by the designed frontend antenna unit shown in Fig An antenna unit consists of four sectors and each sector has a 90˚ horizontal beam width, which is able to cover one of the four directions When receives, each sector captures radio signal and the corresponding transceiver card demodulates the signal to baseband All four package copies are sent to the implemented common MAC where duplicated copies are drop off When transmits, only one transceiver card sends data but the radio signal is split into four copies for transmission by the four sector antennas In order to overcome the big challenge of sea surface movement for relatively stable link connection, each sector of the antenna unit consists of three antennas each of which has a 5˚ vertical beam width Antenna with narrow vertical beam width is used to achieve high antenna gain The three antennas (of each sector) are mounted in such a way that one antenna is mounted vertically (0˚) while the other two are mounted with a +5˚ and –5˚ tilt, respectively Then the effective vertical beam width of the antenna array is 15˚, which is good enough to counter ship tilt due to wave movement A switch is used to control the three antennas and at any moment, only one of the three (in each 90˚ sector) is activated — the one that tilts the least A gyro is used to detect roll and pitch angle of the ship node continuously Antenna switching is triggered when ship rocking angle (pitch or roll) is more than a threshold, which is ±5˚ in TRITON Experiment at East Cost of Singapore with above settings shows that overall antenna switching 139 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 140 CONCLUSION AND DISCUSSION Figure Picture of a mesh node installed on a ship The × antenna array is able to achieve 360°-reception/transmission probability is about 10 percent Note that the above angle settings are for operating frequency of 5.8 GHz TRITON also implemented prototype operating at 2.3 GHz At 2.3 GHz frequency, the antenna vertical beam width, mounting tilt angle, and the switching threshold angle are all 10˚ FIELD TRIAL MEASUREMENTS Onsite measurements have been carried out to evaluate the link performance of the developed prototypes Up to two-hop link performance is measured, because our analysis of network connectivity shows that in the interested region, the routes averagely consist of two hops The measurement location of one-hop link is the southern Singapore waters around St John’s Island One node is mounted on a tower with a height of meter, and the other is carried on a boat which is about 4.5 meter above water surface Figure shows a picture of a mesh node installed on a boat in field trial measurement The boat carrying the mesh node stopped at various distances from the onshore station The measurement parameters are listed in Table and “Ping” method is used in test Figure shows the measured packet delivery ratio at operation frequency of 5.8 GHz We compared performance with and without antenna switching As shown in Fig 5, by switching antenna to counter sea surface movement, package delivery ratio at 3.2 km and km are clearly improved At km, 99 percent of the packets could be delivered in both cases The round trip delay is measured about 3.1 ms, at all above distances, for cases with and without antenna switching Two-hop measurements were carried out for concept proof Distance of the first hop (the first boat to the base station) is about 1.3 km and the second one (the second boat to the first boat) is about km Stable two-hop package delivery between the base station and the second boat has been observed The overall package loss rate is between to percent, for different package size (40, 560, and 1316 bytes) transmission 140 Research and development of a wireless mesh network for high-speed and low-cost maritime communications have been carried out in TRITON project and are briefly introduced in this article Studies on network connectivity and route redundancy show that the envisaged network is feasible in area with high ship density Maritime environment for wireless communications have been studied and various protocols that are suitable for sea environment have been developed based on a maritime simulator The implemented mesh node is capable of overcoming challenges of sea surface movement and sea wave reflections Field trial shows that good link performance can be achieved A number of factors are critical to realize a maritime mesh network, including ship density, time synchronization, and operation frequency, etc Areas with sparse ships may require middleware to switch back to legacy satellite system and the cost of hardware may increase The developed mesh nodes use GPS to achieve time synchronization, which may be difficult for areas with poor GPS coverage, such as high north As ships may travel between countries and in international waters, the frequency usage must follow regulations of various countries and related international bodies such as IMO and ITU To allocate frequency band internationally for maritime mesh communications is hard work since most of bands are occupied A possible solution is to utilize television (TV) white space based on cognitive radios Besides the developed MACbased routing protocol, network coding may be used to increase the network redundancy to minimize possible package loss due to link broken induced by sea wave ACKNOWLEDGMENT The authors would like to thank the other TRITON team members for their work in hardware implementations, field trial measurements, and live demonstrations REFERENCES [1] T Yoshikawa et al., “Development of 27 MHz/40 MHz Bands Maritime Wireless Ad Hoc Networks,” 2nd Int’l Conf Ubiquitous and Future Networks, Jeju Island, Korea, June 16–18, 2010, pp 177–82 [2] S.-J Chang et al., “Cluster-Based Spatial Planning and Evaluation of Maritime Mesh Network Using Vessel Traffic Data,” 2012 Int’l Conf ITS Telecommun., Taipei, Taiwan, Nov 5–8, 2012, pp 471–75 [3] J H Laarhuis, “MaritimeManet – Mobile Ad Hoc Networking at Sea,” Int’l Waterside Security Conf 2010, Marina Di Carrara, Italy, Nov 3–5, 2010 [4] L Lambrinos and C Djouvas, “Creating a Maritime Wireless Mesh Infrastructure for Real-Time Applications,” 2011 GLOBECOM Wksp Mobile Computing and Emerging Commun Networks, Dallas, TX, Dec 5-9, 2011, pp 529–32 [5] S M Mun et al., “An Implementation of AIS-Based Ad Hoc Routing (AAR) Protocol for Maritime Data Communication Networks,” 8th Int’l Conf Natural Computation, Chongqing, China, May 29–31, 2012, pp 1007–10 [6] J S Pathmasuntharam et al., “TRITON: High-Speed Maritime Mesh Networks,” PIMRC 2008, pp 1–5 [7] Y Ge et al., “Route Analysis for a Maritime Communication Network,” 6th Int’l Conf Info., Commun & Sig Proc., 2007, pp 1–5 [8] http://whatonearth.olehnielsen.dk/oceanwaves.asp, valid on Feb 15, 2013 IEEE Wireless Communications • October 2013 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 141 BIOGRAPHIES M I N G -T U O ZHOU [S’01, M’04, SM’11] (mingtuo@nict.com.sg) received his B.E degree from Hunan University in 1997, his M.E degree from Chongqing University of Posts and Telecommunications in 2000, and his Ph.D degree from Asian Institute of Technology in 2003 During January to June 2004, he was a research specialist of the Finland Government Program, Asian Institute of Technology In July 2004 he joined NICT and now is with the Smart Wireless Laboratory of NICT (Singapore Office) He served as co-editor of two books, co-chair, committee member, or finance chair of a number of international conferences/workshops He is the technical coeditor of the IEEE 802.16n and the IEEE 802.16.1a, a voting member/technical contributor of the IEEE 802.11, 802.15 and 802.16, a member of the Test and Certification Working Group of the Wi-SUN Alliance He was the treasurer and now is the secretary of the IEEE Vehicular Technology Chapter at Singapore His research interests include TVWS communications, cognitive radio networks, wireless smart utility networks, broadband wireless access, and radio-over-fiber communications V INH D IEN H OANG (hvdien@nict.com.sg) was born in HaiHung, Vietnam on October 11, 1974 He graduated from Hanoi National University, Vietnam, with a B.Sc degree in mathematics and informatic in 1996, and a M.Sc degree in mathematics 1999 In 2002, he received the M.Sc in “Communication software and Networks” from School of EEE, Nanyang Technological University, Singapore He is currently working as a research scientist for the Smart Wireless Laboratory, National Institute of Information and Communications Technology (NICT), Singapore office His research interests include TCP, routing, sensor networks and cognitive radio H IROSHI H ARADA (harada@nict.go.jp) is director of Smart Wireless Laboratory at National Institute of Information and Communications Technology (NICT) He joined the Communications Research Laboratory, Ministry of Posts and Communications, in 1995 (currently NICT) Since 1995, he has researched Software Defined Radio (SDR), Cognitive Radio, Dynamic Spectrum Access Network, Smart Utility Network (SUN) and broadband wireless access systems on the VHF, TV white space, microwave and millimeter-wave band He also has joined many standardization committees and forums in United States as well as in Japan and has fulfilled important roles for them He has served currently IEEE Wireless Communications • October 2013 100 80 Delivery ratio (%) [9] Raptor X VHF/UHF Broadband Network Radio, http://www.safari-networks.com/wpcontent/uploads/2012/11/RaptorX-App-Note-001final.pdf, valid on Feb 15, 2013 [10] C.-W Ang and S Wen, “Signal Strength Sensitivity and Its Effects on Routing in Maritime Wireless Networks,” IEEE LCN 2008, pp 192–99 [11] J S Pathmasuntharamet al., “High Speed Maritime Ship-to-Ship/Shore Mesh Networks,” 7th Int’l Conf ITS Telecommun., Sophia Antipolis, France, June 6–8, 2007, pp 460–65 [12] S Wen et al., “A Novel Framework to Simulate Maritime Wireless Communication Networks,” MTS/IEEE OCEANS 2007, Vancouver, Canada, Oct 2007, pp 1–6 [13] http://www.syqwestinc.com/support/Sea percent20State percent20Table.htm, valid on Feb 15, 2013 [14] J Joe et al., “Path Loss Measurements in Seaport for WiMAX,” IEEE WCNC ’07, Hong Kong, China, Mar 11–15, 2007, pp 1871–76 [15] P.-Y Kong, “Distributed Adaptive Time Slot Allocation for WiMAX Based Maritime Wireless Mesh Networks,” IEEE WCNC ’09, Budapest, Hungary, Apr 5–8, 2009 [16] M.-T Zhou et al., “Multi-Channel Transmission with Efficient Delivery of Routing Information in Maritime WiMAX Mesh Networks,” 5th Int’l Wireless Commun and Mobile Computing Conf., Leipzig, Germany, June 21–24, 2009 [17] P.-Y Kong et al., “A Performance Comparison of Routing Protocols for Maritime Wireless Mesh Networks,” IEEE WCNC, Mar 2008 [18] P.-Y Kong et al., “A Routing Protocol for WiMAX Based Maritime Wireless Mesh Networks,” IEEE VTC ’09 Spring, Barcelona, Spain, Apr 26–29, 2009 [19] R Boreli et al., “Intelligent Middleware for High Speed Maritime Mesh Networks with Satellite Communications,” 9th Int’l Conf on ITS Telecommun., Lille, France, Oct 20–22, 2009 60 40 20 With antenna switching Without antenna switching km 3.2 km km Link distance Figure Measured packet delivery ratio at different link distances Operation frequency is 5.8 GHz on the board of directors of Wireless Innovation Forum (formerly SDR Forum), White Space Alliance, and Wi-SUN alliance, and also the chair of IEEE Dyspan Standards Committee (formerly, IEEE SCC41 and IEEE 1900) since 2009 and the vice chair of IEEE P1900.4, IEEE P802.15.4g, TIA TR-51, and IEEE P802.15.4m since 2008, 2009, 2011, and 2011, respectively He moreover was the chair of the IEICE Technical Committee on Software Radio (TCSR) in 20052007 and has been the chair of Public Broadband Mobile Communication Development Committee, ARIB since in 2010 He is also involved in many other activities related to telecommunications He has been a visiting professor of the University of Electro-Communications, Tokyo, Japan, since 2005 and is the author of Simulation and Software Radio for Mobile Communications (Artech House, 2002) He received the achievement award and fellow of IEICE in 2006 and 2009, respectively and the achievement award of ARIB and Funai Prize for Science in 2009 and 2010, respectively, on the topic of cognitive radio research and development JAYA SHANKAR PATHMASUNTHARAM (jshankar@i2r.a-star.edu.sg) received his B.Eng, M.Eng and Ph.D degrees in Electrical and Electronics Engineering and Computer Engineering from the Nanyang Technological University of Singapore in 1994, 1996 and 2006, respectively He began his career at I2R, A*STAR in 1996 as an R&D Engineer working in the area of mobile computing His earlier research works covered areas related to mobile middleware and handover in heterogeneous wireless access networks From 1997 to 2002, he participated in several European Community funded research projects such as ACTS OnTheMove, ACTS Cameleon, IST Moby Dick and IST Diadalos by contributing to research areas related to mobile middleware, mobile IP, IPv6 and software agents From 2008 to 2010, he held the position of Programme Manager for High Speed Maritime Mesh Networks Programme He is currently heading the Intelligent Transport Systems Programme in I2R and manages the A*STAR Capability for Automotive Research (A*CAR) Programme His current research interest includes optimization of routing or MAC protocols for efficient mesh networks and localization in challenging indoor environments HAIGUANG WANG [M’09, SM’11] (hwang@i2r.a-star.edu.sg) received his B.Sc degrees from Peking University, China in 1996; and M Eng Degrees from Beijing Univ of Posts and Telecom., Beijing, China in 1999; and M.Sc and Ph.D from National University of Singapore in 2002 and 2010 respectively Currently he is a Scientist in Institute for Infocomm Research (I2R), Singapore Prior to that, he served Institute for Infocomm Research, Singapore as Senior Research Officer (July 2002–June 2008), and then was appointed as Senior Research Engineer at Hisilicon Technology, Huawei Technology Co Ltd (June 2008–August 2009) He has published 20 research papers and filed more than 40 patents, covering a wide range of research areas such as Header Compression, TCP Congestion Control, routing protocols for wireless Ad Hoc/Sensor networks, and MAC layer protocols for IEEE 802.11/16/22 standards He is actively involved in the IEEE 802.11/16/22 standard development 141 ZHOU_LAYOUT_Layout 10/21/13 3:47 PM Page 142 P ENG -Y ONG K ONG [S’99, M’03, SM’12] (pengyong.kong@ kustar.ac.ae) is currently an Assistant Professor at the Department of Electrical & Computer Engineering, Khalifa University of Science, Technology & Research (KUSTAR), Abu Dhabi, United Arab Emirates He was previously an adjunct Assistant Professor at the Electrical & Computer Engineering Department, National University of Singapore, concurrent to the appointment of Research Scientist at the Institute for Infocomm Research (I2R), Agency for Science, Technology & Research (A*STAR), Singapore Prior to the Ph.D study, he was an Engineer with Intel Malaysia He received the B.Eng degree in electrical & electronic engineering (first-class honors) from the Universiti Sains Malaysia, and the Ph.D degree in wireless networking from the National University of Singapore His research interests are in medium access control protocols, transmission scheduling, network traffic control and quality of service provisioning for wireless networks His work in TRITON was done when he was with I2R, Singapore CHEE-WEI ANG [SM] (angcw@i2r.a-star.edu.sg) is currently working for Institute for Infocomm Research, A*STAR as a scientist He has been serving as principal investigator for industry-funded projects He joined the company in 1996 as a research engineer when the company was a research centre (CWC) in NUS He pursued his postgraduate studies in NUS and was awarded the Ph.D degree in 2009 His research interests are in the area of mesh net- 142 working, focusing on MAC and routing layers Machine learning is his other interest area, where the techniques are applied in the optimization of networking parameters Currently he holds two international patents on cognitive radio MAC and has published in IEEE Transactions and Elsevier journals, and IEEE conferences He was a recipient of the IES Prestigious Engineering Achievement Awards in 2007 Y U G E (geyu@i2r.a-star.edu.sg) received her M.Eng and Ph.D degrees from National University of Singapore and Nanyang Technological University, all in wireless communication networks area She joined Institute for Infocomm Research (I2R), A-Star, Singapore in 2001 as Scientist She worked in the research areas of VoIP in heterogeneous wireless networks, wireless mesh networks, and wireless sensor networks She is currently leading a research team in the area of wireless body sensor networks for healthcare Her current research interests are transmission and sensing technologies in wireless communication networks for end-to-end service provisioning SU WEN received her Ph.D in Computer Science at the University of Kentucky She was an Assistant Professor at the Naval Postgraduate School in Monterey, CA She currently resides in Singapore and worked as a Senior Research Fellow at the Institute for Infocomm Research and the National University of Singapore IEEE Wireless Communications • October 2013 ... of a multihop wireless mesh network for maritime communications which is more suitable for maritime communications NETWORK FEASIBILITY STUDY The success of a maritime wireless mesh network will... for Maritime Data Communication Networks,” 8th Int’l Conf Natural Computation, Chongqing, China, May 29–31, 2012, pp 1007–10 [6] J S Pathmasuntharam et al., TRITON: High-Speed Maritime Mesh Networks,”... considerations of the unique characteristics of maritime mesh networks Extensive simulations with the developed maritime simulator” show that the designs are effective for maritime mesh communications