Using Ship Movement in the Irish Sea for MANET Evaluation

Second UKSIM European Symposium on Computer Modeling and Simulation Using Ship Movement in the Irish Sea for MANET Evaluation Adrian J Pullin School of Computing Liverpool Hope University Liverpool, UK [email protected] Steve Presland School of Computing Liverpool Hope University Liverpool, UK [email protected] relief and vehicle ad-hoc networks e.g. [1], [2], [3]. Since the cost and complexity involved in building real MANETs is considerable, most current research is conducted using simulations, with the most popular software being ns-2 [4]. Within these simulations, the mobility of nodes is usually based on either the Random Waypoint (RWP) model [5] or variations thereof [6]. VANET research generally makes use of Manhattan or similar mobility models [7]. The size of world modelled and the number of nodes within the simulation are often arbitrary [4]. The possible use of MANET by shipping is discussed in this paper. In particular, the focus is on the movement of commercial shipping (including passenger and freight) within the north eastern part of the Irish Sea. Real data is gathered using the Automatic Identification System (AIS) [8] that is carried by all commercial and passenger vessels. This is processed into the format required by ns-2 for its movement file. Experiments are run using this world model with various different transmission ranges to identify the range required to establish a viable MANET in this situation. Planned future work in this area is discussed. Abstract A Mobile Ad Hoc Network (MANET) is a selfconfiguring multi-hop network that does not rely on infrastructure being available. Vehicular Ad-Hoc Networks (VANET) are an application of MANET, with most research focusing on road traffic. The possibility of constructing a VANET for shipping at sea is presented in this paper. Most research into MANET technologies is done using simulations that contain models of the physical space in which the MANET operates, the number of nodes deployed and the movement patterns and communications requirements of the nodes. A method of producing a realistic model for a specific application is presented. Data from the international standard Automatic Identification System (AIS) used by commercial shipping is gathered and processed to produce a realistic movement model. This model is used in ns-2 simulations to evaluate the application of MANET technologies to shipping. Experiments are performed using different transmission ranges with 802.11 Wireless Ethernet as the MAC layer and marine VHF radio as the physical layer. It is demonstrated that the model produces results that are in line with known transmission ranges of marine radio and further concluded that it is possible to construct a MANET for ships in the north Irish Sea using marine VHF as the physical layer. 2. Related work 2.1. VANET Models Extensive work has been done on VANET, including five ACM workshops and for example [9], [3], most of which focuses on road traffic. It is common to use standard vehicle simulation models such as Manhattan [10]. There have also been some studies using real data gathered from traffic monitoring. For example in [11] a model is proposed using data gathered from the Seattle bus system. 1. Introduction A Mobile Ad Hoc Network (MANET) is a selfconfiguring multi-hop network that does not rely on infrastructure being available. Typical applications cited in the literature include battlefield, disaster 978-0-7695-3325-4/08 $25.00 © 2008 IEEE DOI 10.1109/EMS.2008.50 Colin Pattinson Innovation North Leeds Metropolitan University Leeds, UK [email protected] 394 (such as just Liverpool Docks) or a completely different area (for example the English Channel). The particular location was chosen pragmatically, being a local area for which data is available, but it does provide an example of a fairly busy sea area with sufficient nodes of different types, such as fixed Coastguard stations, ships at anchor, regular traffic made up mostly of scheduled ferry services and ad hoc traffic. The data used was taken from one hour periods within single days. This provides a reasonable snapshot of the traffic in this area but can be easily modified to cover different periods if required. Studies on Urban Mesh networks and similar systems where the nodes are carried by pedestrians are also reported, for example [12]. 3. Automatic Identification System The Automatic Identification System is a global maritime safety system by which ships over 300 tons are required to broadcast their identification, position, heading and speed every 10 seconds if moving or every three minutes if at anchor. This is transmitted on the marine VHF band in the range 161.975 and 162.025 MHz and is usually received, processed and displayed by the ships’ navigation systems. AIS signals can be picked up by a standard marine VHF radio receiver, decoded via a commercially available interface such as the NASA AIS Radar Receiver [13] and displayed on a computer using ShipPlotter software [14]. A web site called ShipAIS [15] collates information gathered from a number of receivers around the Irish Sea and from other parts of the UK and produces a real time display of ship movements. The data is also logged so that historical records are available. The web site owner has kindly supplied the data used in this research. To the best of our knowledge, no one has produced a simulation model of shipping based on data gathered from the AIS and no one has produced a MANET simulation based on shipping at sea. 4.2. Node types There are effectively two types of node in the model. A small number of fixed nodes, identified under AIS as Base Stations, comprise the Coastguard Vessel Traffic Service (VTS) stations in the area, located at Snaefell, Isle of Man (node 1), South Stack, Anglesey (node 2) and Crosby (node 3). [17] The remainder of the nodes are classed by AIS as vessels. A significant number of them are permanently in one location. For example the Elsam Traffic node 0, which is classified as a manned VTS by the Maritime and Coastguard Agency (MCGA), is physically a ship but stays in one place for periods of months at a time and has access to fixed, land based infrastructure. As far as the simulation is concerned, this is a Base Station. Ships can remain at anchor or moored for the duration of the simulation. These vessels do not have infrastructure data connections and can move; therefore they are mobile nodes as far as MANET modelling is concerned. 4. Description of AIS based model A model of shipping in the north eastern Irish Sea is presented in this paper. The location used and the data on ships and their movements are all real world. The communications modelled are simplistic, designed to establish whether connectivity can be achieved is this situation. The simulator used is ns-2 version 2.31 [16] which is used extensively in MANET research [4]. 4.3. Radio range The antenna heights for the VTS significantly affect the radio range. Marine VHF radio operates in the 160MHz band and has line of sight range. This gives approximately 5km range if the antenna is 2m high. Raising the antenna can increase this significantly, with a mast height of 70m being required for 30km range. The Elsam Traffic ship based VTS and the Crosby Coastguard station are both based at sea level, so their mast heights will only give a range of about 10 km. South Stack mast is on top of a hill at 130m, giving a range of approximately 40km and Snaefell is 600m high, giving about 87km range. 4.1. The world modelled The area covered in this example runs from N52° (Strumble Head, Wales) to N55° (the Northern end of the Rhins of Galloway, Scotland) and W2° (AshtonUnder-Lyne, England) and W7° (Waterford, Ireland), an area of approximately 105,000 km2. The area is bounded to prevent occasional anomalies within the data from interfering with the model. The data available covers a rather smaller area within these bounds. The bounds are straightforward to adjust, enabling the focus to be on a different size of world 395 produces movements that are realistic in terms of ship speed but can also result in ships apparently sailing across land. Checking for significant time or location gaps in a ship’s AIS record enables the ship to be removed from the model when it disappears from the record. However, this ship is still in the real world and could potentially contribute to the MANET. The results presented here are based on ignoring the anomalies and the model produced has been verified as having no significant errors. Ns-2 presents some reliability issues when the model is physically large. The area covered, approximately 300km x 350km, and the radio communication ranges used, up to 30km, were found to be beyond the reliable limits of ns-2. In order to produce usable models, the entire world has been scaled down by a factor of 1000. Thus, distances in kilometres are represented in the model as metres. This applies to the distances between nodes, the radio communication ranges and the speeds of movement. A snapshot of the nodes’ locations is shown in figure 1. 5. Producing the movement model Data files have been obtained from ShipAIS [15], which logs and displays ship movements for the area. These files contain a record of all AIS traffic for a 24hour period. Relevant records are processed to extract vessel Maritime Mobile Service Identity (MMSI), location and speed. This data is converted into the format required by ns-2. The area covered and time of simulation are hard coded, whereas the number of nodes is extracted from the AIS data. The AIS data is not entirely reliable. Since ships can switch their AIS on and off, they can appear or disappear at any time, although this usually indicates that they are starting or stopping. Due to the limitations of the marine VHF radio used and the limited number of receivers supplying the data, there are also black spots where data is not available. This means that ships can seem to jump from one location to another. Two solutions have been tried for this problem. Ignoring the anomalies when producing the ns-2 movement file results in ns-2 extrapolating the paths of ships that disappear and then reappear. This Figure 1. Location of nodes at start of a sample run including VTS communications and this is build in to ns-2. The physical layer of 802.11, operating at 2.4GHz, has a free space range of only a few hundred metres, which is insufficient to connect the network. All ships in the 6. Communications Models A large proportion of the MANET research uses 802.11 Wireless Ethernet as its MAC layer 396 model carry marine VHF radio and this is used for AIS data transmission and voice communication. Therefore, this radio model has been used at the physical layer. A number of runs have been performed using different ranges to determine whether it is possible to construct a MANET using VHF radio in this scenario. The connectivity required was for ships to communicate with one of the four manned VTS stations in the area. These are identified as node 0, the Elsam Traffic, which is a mobile VTS that was moored at Mostyn Quay in the River Dee (N53.32°, W3.268°), and the Coastguard stations at Snaefell, Isle of Man (N54.26°, W4.461°) node 1, South Stack, Anglesey (N53.31°, W4.69°) node 2 and Crosby, Merseyside (N53.50°, W3.058°) node 3. As fixed stations with land based, wired connections, these nodes could provide connections to the Internet and other communication systems as well as the usual Coastguard services. The data traffic in the model is constant bit rate over UDP. This allows the testing of connectivity whilst reducing the simulation overheads that can arise in more complex communications. The MANET routing protocol used was DSDV [18], for the same reason. Simulations were run using the DSDV protocol implementations in ns-2 supplied by the CMU Monarch Project [19]. as can the routes taken by ships. This has been further confirmed by matching with the displays from the ShipAIS web site for the period simulated. 8.2. MANET performance metric Data packet delivery ratio (PDR) is used to evaluate the results. PDR = received data packets / sent data packets. This provides a measure of how effectively the network is at providing a connection. Since the purpose of this simulation is to evaluate the feasibility of establishing a MANET in this scenario, this metric provides a clear indication of the connectivity provided. Normalised routing load (NRL): NRL= routing packets sent / data packets received This shows how much routing traffic is generated per packet delivered and is a commonly used measure of the overheads of routing protocols in MANETs. 8.3. Required range for connectivity Table 1 shows the PDR results from the simulations. TABLE 1. Packet delivery ratio for each VTS node for different transmission ranges 7. Experiments performed 5km 10km 15km 20km 25km 30km The purpose of the simulation model is to allow the evaluation of certain aspects of MANETs as applied to this scenario. In order to demonstrate that the model is suitable for this purpose, a set of experiments has been performed using this model to find the minimum communication range required for a MANET to be established that would connect ships to each of the four VTS. Ranges from 5m in the scale model, representing 5km real world, to 30m (30km) in 5m (5km) steps were used to attempt to connect to each of the VTS in turn. Elsam Traffic node 0 3% 97% 98% 99% 99% 99% Snaefell node 1 0 0 0 0 0 99% South Stack node 2 0 0 97% 99% 99% 99% Crosby node 3 77% 97% 99% 99% 99% 99% At 5km transmission range, a small amount of data can reach the Elsam Traffic at Mostyn (node 0) but this is derived only from local ships and is insufficient to provide a reliable connection for the rest of the MANET. This range also provides local connectivity for Crosby (node 3) but again is insufficient to set up the whole MANET. Establishing a reliable connection to these VTS requires a transmission range of 10km, which is within that available to these stations using marine VHF. Looking at the node locations on the map, it is clear that there are a significant number of nodes within 10km of these VTS and each other. 8. Results 8.1. Verification of model The location and movement of the nodes has been checked using the Network Animator (NAM) tool that comes with ns-2 [16]. By overlaying NAM screen shots on a map of the area (figure 1), the positioning of the four fixed nodes can be confirmed 397 There are fewer nodes in proximity to South Stack (node 2) and none within 10km. If the range is increased to 15km, it becomes possible to connect to this VTS. There is a traffic separation scheme in operation off the north west coast of Anglesey which results in most ships heading to and from Liverpool passing with 15km of South Stack VTS, so the real world would provide the required vessel movement pattern. The mast height at South Stack would provide the required range. Snaefell, being the most remote VTS in this experiment, requires a range of 30km in order for ships to connect. Given the mast height at these two stations, again this would be feasible. These results match the expectation that the relatively high node density in the Mersey approaches would allow a MANET to be established in that area even with the low antenna heights of the Crosby and Elsam Traffic VTS. They also show that even the low node density near South Stack and Snaefell, it would be possible for a network to operate, as these stations have sufficient altitude to provide the range required. 9. Conclusions It has been shown that publicly available data gathered from marine AIS can be used to build a model of ship movement for the evaluation of MANET technology using ns-2. The construction of this model is flexible enough to be applied anywhere that AIS data is available. The model has been verified in terms of node location and movement against maps of the area and the ship location and movement information on the ShipAIS web site. This model has been used to show that a MANET could be established that covers the north eastern portion of the Irish Sea to provide connectivity from ships to land based stations in the area but that the routing overhead using DSDV is high. 10. Future developments 10.1. Geographical area covered This model has been developed to demonstrate the modelling of a realistic scenario for MANET evaluation using AIS data. Although the current model covers one specific area, it would be simple to use data from other regions to produce models covering any area. For example, the busy shipping lanes of the English Channel would provide a useful comparison, whereas the more sparsely populated areas of the Hebrides would present a significant challenge in building a MANET. 8.4. Cost of connectivity Table 2 shows the NRL results from the simulations. TABLE 2. Normalised Routing Load for each VTS node for different transmission ranges 5km 10km 15km 20km 25km 30km Elsam Traffic node 0 3913% 153% 174% 144% 135% 128% Snaefell node 1 0 0 0 0 0 128% South Stack node 2 0 0 174% 146% 132% 128% Crosby node 3 10.2. Mast location 172% 153% 168% 145% 133% 128% The locations of the VTS are identified in their AIS transmissions. The MCGA also operates a number of relay stations to provide radio coverage in areas either out of range or in radio shadow from the base stations. There is almost complete coverage of UK waters, with the exception of small black spots. These relay stations are not identified separately by AIS. The inclusion of these masts in the model would provide a better representation of connectivity but would require the hand coding of their locations. It is planned to add these relay stations to the model manually in the future to provide more realistic radio coverage. With nearly forty routing packets for each delivered data packet, the cost of establishing even a very poor connection with node 0 using a 5km transmission range is prohibitive. Using transmission ranges that permit a MANET to be established, the routing overheads are still high. It costs between one and two routing packets for every data packet delivered using the DSDV routing protocol. The routing overhead is almost entirely dependent on the transmission range, with no significant variation between the target nodes. 10.3. Use of the model This set of experiments only looked at one routing protocol and one data traffic pattern. Now that the 398 [5] Johnson, D.B. and Maltz, D.A. “Dynamic Source Routing in Ad Hoc Networks”, Mobile Computing (Imielinski & Korth eds.), Kluwer Academic Publishers, 1996. [6] Yoon, J., Liu, M., Nobel, B. “Random Waypoints Considered Harmful”, IEEE Infocom 2003. [7] F. Bai, N. Sadagopan, A. Helmy, "IMPORTANT: A framework to systematically analyze the Impact of Mobility on Performance of RouTing protocols for Adhoc NeTworks", The 22nd Annual Joint Conference of the IEEE Computer and Communications Societies INFOCOM2003, San Francisco, March/April 2003. [8] US Coastguard , “Universal Shipborne Automatic Identification System (AIS) Transponder” Available at http://www.navcen.uscg.gov/marcomms/ais.htm (access checked 28/04/08) [9] W.K.G. Seah, F.W.H. Lee, K.W.L. Mock, E.K.W. Ng, M.Q. Kwek, “Mobility modeling of rush hour traffic for multihop routing in mobile wireless networks”, IEEE Vehicular Technology Society Fall 2006, Montreal, 2006. [10] Cano, J.-C. Calafate, C. Manzoni, P. Toh, C.K. “Modeling of mobility and groups in inter-vehicular MANET-based networks”, 2nd International Symposium on Wireless Pervasive Computing, ISWPC '07, San Juan, 2007. [11] J.G. Jetcheva, Yih-Chun Hu Santashil PalChaudhuri Amit Kumar Saha David B. Johnson, "Design and Evaluation of a Metropolitan Area Multitier Wireless Ad Hoc Network Architecture," Proc. 5th IEEE Workshop on Mobile Computing Syst. & Applications (WMCSA 2003), Monterey, CA, October 2003. [12] Kumiko Maeda, Kazuki Sato, Kazuki Konishi, Akiko Yamasaki, Akira Uchiyama, Hirozumi Yamaguchi, Keiichi Yasumoto, and Teruo Higashino. “Getting Urban Pedestrian Flow from Simple Observation: Realistic Mobility Generation in Wireless Network Simulation.”, Proceedings of the 8th ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM2005), 2005. [13] NASA Marine web site http://www.nasamarine.com/ (access checked 28/04/08). [14] Shipplotter web site http://www.coaa.co.uk/shipplotter.htm (access checked 28/04/08) [15] ShipAIS web site http://www.shipais.com/ (access checked 28/04/08). [16] Ns-2 web site http://www.isi.edu/nsnam/ns/ (Access checked 28/04/08) [17] Maritime and Coastguard Agency web site http://www.mcga.gov.uk/ (access checked 28/04/08). [18] Perkins, Charles E. and Bhagwat, Pravin, "Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computers”, sigcom94, 1994. [19] CMU Monarch Project, “The CMU Monarch Project’s Wireless and Mobility Extensions to ns”, Computer Science Department, Carnegie Mellon University, Pittsburg, 1999. model has been established, it can be used for the evaluation of routing protocols and the development of applications to be carried over MANET at sea. Investigations into the routing overhead of other protocols in this scenario need to be done, particularly comparisons between pro-active, reactive and location based protocols are required to determine whether a more cost effective solution is available. 10.4. Reliable AIS data More recent data sets have been obtained and these are gathered from a larger number of receivers, significantly reducing the gaps in the data. These will be used in the future to build models that are more robust. 11. Summary A method of generating a simulation model of ships at sea using AIS data has been presented. A model using this method has been developed and verified against the records of ships’ movements at the time. This model has been used to show that a ship based MANET could be constructed to provide connections to shore based stations in the north eastern Irish Sea. The required transmission range for such a network to work has been established and this matches the results expected, based on the known transmission range of marine VHF and node density in the area. The routing overhead of such a MANET using DSDV has been determined. 12. Acknowledgments The authors would like to thank Mr. Ian McConnell, author of the ShipAIS web site [15] for supplying the AIS data for this project. 13. References [1] R. Chellappa Doss, A. Jennings, N. Shenoy, “A review of current mobility prediction techniques for ad hoc networks”, Proceedings of the 4th IASTED International multi-conference Wireless and Optical Communications, Banff, Canada, July 2004. [2] F. Bai, A. Helmy, “A survey of mobility models”, Wireless Ad Hoc and Sensor Networks, Kluwer Press Publishers, 2004. [4] Kurkowski, S., Camp, T., Colagrosso, M. “MANET Simulation Studies: The Incredibles”, ACM Mobile Computing and Communications Review, volume 9 number 4, October 2005 399