연구의 선진화와 생산성 향상에
필요한 최적의 소프트웨어와 컨설팅을
공급하도록 노력하겠습니다.

Automatic Identification of Ships (AIS)

The plot on the right was derived from the Shipplotter AIS software; this type of result was a part of a research program at the Royal Military College, Kingston. The concept was to fuse AIS detections both from a receiver at the college and with AIS data from other locations with RADARSAT-2 data. There are five ships in this plot.

  • AIS is a VHF short range transponder system designed for ship collision avoidance. The range is determined by line of sight. The transmissions include position, course and the ship's Maritime Mobile Service Identity (MMSI) number, which in principle is unique to each ship. There are databases relating the MMSI number to the ship name. In addition information is transmitted about the destination and the cargo. Thus this information is invaluable for maritime surveillance. For example, the economy of Panama is very dependent on the Canal as described in "Panama Canal". The Canal is also very important for the world economy.
    The line of sight requirement (some 30 km or more) renders AIS difficult for wide area surveillance and long coastlines require many terrestrial receiving stations. Buoys are an option but, for anything approaching complete coverage of maritime approaches to North America, a terrestrial system would be very expensive.
    AIS is fundamentally a self-reporting system and the data is sometimes in error by accident or design. For example the MMSI number may have a default value or have been entered incorrectly. Therefore AIS data is unreliable for maritime security purposes and it should be verified using data from other sensors, such as satellite-borne SAR.
    There are two classes of AIS transponder. Class A is designed for ships with a gross displacement of 300 tons or greater, while Class B is for small craft. The Class A transponders transmit with a nominal power of 12.5 W but the Class B power is only 2 W. Surveillance of the open oceans normally involves just Class A signals, because Class B transmissions are typically confined to the coasts.
  • Satellites can be fitted with AIS receivers and ORBCOMM, COMDEV (through its subsidary company exactEarth), SpaceQuest and LuxSpace are currently flying these.
    ORBCOMM specializes in short messaging sytems and has numerous satellites in orbit, while exactEarth has several with AIS receivers on board. To achieve global AIS coverage with a latency of about 10 minutes about 30 satellites are required. A number of consortia have been formed to control costs and their membership appears to be broadening. For example ORBCOMM is partnered with LuxSpace and SpaceQuest with exactEarth. COMDEV Europe (UK) and LuxSpace participated in the Pasta Mare Project.
    Unfortunately satellite receivers are not fully compatible with AIS because transmissions from ships located far from each other can arrive at the receiver simultaneously. This is not a problem for terrestrial receivers because the system uses Self Organizing Time Division Multiple Access (SOTDMA) to implement a cellular architecture in which signals from within an individual cell do not interfere unless the shipping density is extraordinarily high.
    Various strategies are possible to mitigate the interference. Advanced signal processing relying on the separation of signals in the frequency, time and other domains can each yield significant improvements that are important in areas where the shipping density is very high. Such areas exist off the Atlantic and Gulf coasts of the US. In these areas a very significant proportion of AIS signals are likely to be missed due to mutual interference effects. Alternatively, the space-borne receiver antenna pattern can be made more directive so as to reduce the number of transmitting ships in the beam.
    Other problems are associated with timing problems because of the large variation in the range of an AIS transponder when viewed from space. This causes signals to fall partially out of their TDMA slot on the satellite receiver. Finally the present VHF allocation is not totally dedicated to AIS maritime use and terrestrial interference is likely from some geographic areas, mainly outside of North America.
  • ORBCOMM has recently provided LRDC with 30 days of space-based AIS product from their new satellites and this has been analyzed to provide estimates of performance. The full report must be requested directly from ORBCOMM (Flessate.Greg@orbcomm.com), but a summary version is available here: J.K.E. Tunaley, "Analysis of ORBCOMM S-AIS Product", Summary of LRDC report to ORBCOMM, August 2012.
  • An approach to the signal collision problem is to drastically reduce the number of transmitted signals. This can be accomplished by extending the interval between transmissions from a few seconds to minutes: current AIS transmissions from a ship can take place every two seconds. To avoid compromising collision avoidance, a new AIS channel is needed and this is the approach favored by the USCG.
    The interval cannot be extended beyond the time that a ship is in the space-borne receiver beam, otherwise the ship is likely to be missed. This limits the absolute maximum interval to about 15 minutes.
  • The theory of AIS signal collisions can be described in terms of Poisson processes (e.g. J.K.E. Tunaley, "An Analysis of AIS Signal Collisions", August 8th, 2005 and J.K.E. Tunaley, "A Stochastic Model for Space-Borne AIS", August 14th, 2005). The second of these has been cited in www.fcc.gov and www.ntia.doc.gov. A theoretical model of signal collisions that takes into account the ability of a space-based receiver system to resolve collisions is described in J.K.E. Tunaley, "Space-Based AIS Performance", May 28th, 2011. This has recently been extended to include the effects of thermal noise and interference on the probability of receiving an uncorrupted message. The theory has been compared with some data from COMDEV presented at the ASAR-2011 conference at the Canadian Space Agency in June. This confirms the predicted functional dependence of the probability that a message will be correctly deciphered on the number of ships in the field of view. See J.K.E. Tunaley, The Performance of a Space-Based AIS System, August 8th, 2011.
    The performance of a S-AIS system can be expressed in terms of the probability that a transmitted AIS message is correctly received together with the bit error rate or the packet error rate, depending on the application. This information can be regarded as one point on the Receiver Operating Curve (ROC). It is determined by the quality of the system; the operating parameters are chosen by the data provider. In general, as the parameters of the system are changed so that the probability of detection rises, the error rate also rises. A compromise is chosen by the provider possibly to satisfy the principal customer requirements. As noted, S-AIS is not fully compatible with terrestrial AIS because failure of the self-organizing protocols allow signal collisions. The probability that an AIS message is received in spite of signal collisions depends on the number of slots used by the message according to the TDMA specification. Type 1 position reports occupy one slot but Type 5 static reports occupy two consecutive slots. Therefore the Type 1 and Type 5 messages are affected differently by collisions. The ratio of the number of these messages provides information on the probability that either will be received. The bit error rate can be estimated by detecting unreasonably large jumps in ship position in a sequence of AIS messages from a single ship. When these occur, the most significant bits in the latitude and longitude are usually incorrect. This permits bit error and packet error rates to be estimated. Some results are described in J.K.E. Tunaley, S-AIS Performance Analysis, March 2012.
    A presentation on the cross-validation of of AIS signals with RADARSAT ship detections by data fusion is available: J.K.E. Tunaley, "Polar-Epsilon and Canadian MDA", June 2006. This was given at an NDIA conference organized by the USCG, "Maritime Domain Awareness Requirements, Capabilities and Technology Forum, Tampa, Florida. When comparing AIS reports with the positions of ships in a Synthetic Aperture Radar (SAR) image, it is important to correct the apparent ship position in the image for "Velocity Bunching". A simple derivation for this is provided in J.K.E. Tunaley, "SAR Velocity Bunching Relationships", LRDC Report, April, 2009.