Viability and Performance Analysis, at a Physical Level, of a Maritime Communication Network on Medium Frequency

Abstract

This paper analysed the potentialities that have not yet been exploited of the Automatic Vessel Identification System on Medium Frequency (AVISOMEF) that was previously presented to the scientific community. This system expanded the spectrum of possibilities of the Global Maritime Distress and Safety System (GMDSS). By the time AVISOMEF was presented, one of the unexpected benefits was that it could constitute a platform capable of supporting a Maritime Communication Network on Medium Frequency; all remotely located vessels could communicate with each other by using AVISOMEF, making use of the new established network, within the geographic area of implementation, without needing to use satellite techniques. Medium frequency propagation is stable and is not subject to coverage restrictions. To conduct this experiment, we collaborated with the Spanish Maritime Safety Agency (SASEMAR). It was thanks to them that we gathered real-time information about the nationwide maritime traffic that navigated through the Spanish Search and Rescue jurisdiction waters from 2018 to 2020.

1. Introduction

As a preliminary step in the relative analysis regarding the viability of the new Maritime Communication Network, we first provide a brief description of the Automatic Vessel Identification System on Medium Frequency (AVISOMEF) on which it is based [1,2].

The Global Maritime Distress and Safety System is a technology that surpassed its intended use when it was implemented on vessels [3]. Its real success consisted in ensuring its presence at a global level. However, it represented a starting point and a door opener for the system’s gradual implementation of functionalities that initially had not been contemplated [4,5].

AVISOMEF is one of those new functionalities not initially envisioned by the Global Maritime Distress and Safety System (GMDSS). Nevertheless, its viability has already been revealed to the scientific community [6]. It is still not operational, but there is an increasing number of expert voices who are encouraging authors to present it to the International Maritime Organization (IMO), when the possibility to do so arises.

AVISOMEF basically consists of a system capable of accomplishing a remote identification of vessels that navigate across a certain implantation area, which must meet specific requirements regarding the minimum density of maritime traffic within it.

A digital mapping of the area to be controlled must be implemented, which is made up of a grid based on square cells measuring 44 nautical miles per side (Figure 1). Within each of these cells, a series of listening and calling channels must be established and the ships located within the controlled areas must operate in the specified channels following the protocol established by AVISOMEF. At the end of each grid there would be antennas that emit signals from the governing Traffic Control Centre (TCC).

The vessels’ identification will be conducted on medium frequency (MF). Nevertheless, the system will not be limited by a maximum distance coverage of 100 miles, typical for an MF, as the process is propagated on radio-links made from a cell to the following contiguous one and so on [7,8,9].

By implementing AVISOMEF, which incorporates Digital Selective Calling (DSC) as a technology already proven in GMDSS, a vessel will automatically know whether or not one of the radio packets sent by the Traffic Control Centre (TCC) is directed to it or not.

When the call is directed to her, she will react by responding with a message back to the Control Centre. This message will not only include information concerning the vessel that is expected to be identified, but it will also provide information related to the environment of that vessel. This additional information will be important for the TCC to establish a relative understanding of the maritime traffic around the ship, including details about new additions to the cell from areas outside the grid.

If the message is not addressed to any of the ships located in the cell, one of them will work as a repeater, thus establishing a control protocol and advancing the message to the next cell, until it reaches the ship that is expected to be identified.

The calls will be launched to the various ships in the area periodically by the TCC in order to establish real-time tracking of all the vessels located there. The periodicity in which AVISOMEF must proceed to identify each vessel located in the controlled area will depend on its speed. The system, with the information received, will regulate itself. It must carry out at least one identification of each ship per cell.

The messages, broadcasted from the TCC to the ships to be located, will be launched in an open collective message. The return messages from the ships to the TCC are part of a peer-to-peer system, as the initial message stores the callsigns (MMSI) of the stations acting as repeaters, therefore establishing a map of the communication path.

The requirements demanded by AVISOMEF, as far as hardware is concerned, are the use of four receivers (A, B, C, and D) and a transmitter, all of them managed by a computer control [10]; a block diagram is shown in Figure 2.

When the message meant to identify a certain vessel starts from the TCC, it will first arrive at the vessels located in the first cells of the grid (Figure 1) and be captured by their receivers, named “A” (Figure 2), through channel “1”. These receivers will be equipped with DSC technology.

If said message is addressed to one of the vessels, it will be noticed thanks to the implemented DSC technology. Automatically, said ship will proceed to manifest its presence in the cell by sending a message back to the TCC on the channel that appears in Figure 1 entitled “80”. The rest of the ships located in this first cell of the grid listen to channel “80” through their receiver “C” (Figure 2). When the ship in question is identified, there is no need for any of the vessels to rebroadcast the message to subsequent cells. As can be seen, this “C” receiver does not need to be equipped with DSC technology as it is only a listening channel receiver; it never needs to send a signal of self-identification.

If the message sent by the TCC to the first cells of the grid is not directed to any of the ships located there, none of them would send a message back to the TCC, since none would be alerted by their receiver “A” (Figure 2). Therefore, one of them would rebroadcast the message to the second cell of the grid, on channel “11” (Figure 1), making use of its transmitter. This process (SITOR-B) would be repeated from cell to cell until the message reaches the cell in which the ship being called is located, and that ship sends back the appropriate message.

The return message from the ship to be identified to the TCC, passing through the different repeaters that have contributed to this message evolving from each cell to the next one, is no longer an open message. The return response will follow the inverse path to that of the initial message, passing through each of the vessels that had acted as a repeater. It is a point-to-point message (SITOR-A); therefore, listening to this message will require a receiver equipped with DSC. Since each ship will be called directly, it requires receiver “D” (Figure 2).

In the operation of this system, there are two main factors that play a decisive role: wait times and repeater-ship determination. Wait times are important since a certain message calling a ship is heard until it acts as a repeater if there is no response. It is also important to select the best ship to act as the repeater. As we have already commented in previously presented literature, current technology can provide solutions to these problems. A delay protocol was proposed when acting as a repeater, in which the geographical distance to the central point of each cell minimizes response times. After a previously established time, if no ship responds as the recipient of the message from the TCC, those located in the cell in which the message has been broadcast can act as repeaters, but the closer a ship is to the geographic point indicated above, the shorter the time of response imposed in the protocol. This is perfectly feasible as the system incorporates data from geographic information systems installed on board.

To maintain safety at sea [11] when vessels are far away from the coast, it is mandatory to establish a communication system capable of overcoming distance limitations of the frequency band used. AVISOMEF is a system that allows for the identification of vessels without using satellite communications, overcoming the range limitations of medium frequency communication.

Communication between vessels and shore using satellites has been a very useful system to communicate with vessels far away from the coast [8,12,13] but it is associated with high costs. Some studies have proposed the integration of satellite systems with non-satellite systems that have less coverage but that are cheaper or free [14,15,16]. These hybrid systems can overcome the coverage limits of non-satellite systems thanks to the use of satellite systems.

Other proposals only use non-satellite technics. In [17], a new net layer protocol in an OSI model based on AIS was proposed. This new protocol uses VHF in its physical layer, so it needs less distance between vessels than do MF systems such as AVISOMEF. In other studies, WiMAX was proposed as a physical layer of the system [18,19]. WiMAX is a non-satellite communication system with a lower associated cost than satellite systems with a maximum range between 40 and 70 km: a range that can be overcome with AVISOMEF.

This paper studies the possibility of using AVISOMEF as a physical layer in a non-satellite maritime communication network. It is necessary to check if the maritime traffic is sufficient to cover the needs of AVISOMEF and, therefore, of the new maritime communications network.

The rest of the paper is organized as follows: in Section 2, the hardware necessary for the new network is described; in Section 3, a description of the simulator developed to check the viability and performance of the new system is given; in Section 4, an implementation of the system on the Spanish coasts is proposed; in Section 5, the results obtained in the simulation executed with the implementation proposed in the previous section are presented and, in Section 6, the conclusions are extracted from the obtained results.

2. Descriptive Analysis of the Maritime Communication Network on Medium Frequency

The leap of AVISOMEF from conducting remote identification of vessels to constituting a Maritime Communication Network on Medium Frequency will be due to the achievement of the objective of the already stated Automatic Vessel Identification System on Medium Frequency: the verification of the existence of a continuous linkage along the maritime traffic routes in which it has been implemented. The forward phase will consist in mounting a new “chain” on those routes that allows communication to be established between any two of the previously identified vessels.

The information will leap forward from ship to ship from origin to destination, so that the communication can flow in this new chain in a similar way to the process used in AVISOMEF.

A new channel that is different for each cell, called the “listening/calling (channel L/C)”, will have to be implemented in AVISOMEF. Therefore, each ship that aims to establish a communication process with another ship must express said intention by making use of this new channel. In this request, it must also include the MMSI of both the call recipient and itself.

The rest of the ships located in the same AVISOMEF cell as her will take note of her intention, so that the first of them that is required by AVISOMEF to make one of the periodic identifications will include with the information to be transmitted to the TCC the request of initiating a communication process on the part of her cellmate.

Once the corresponding TCC is aware of the mentioned request, it will do a search in a communal database of all the TCCs involved in the Maritime Communication Network on Medium Frequency. In this way, it will verify if the contacted ship is within it. If so, it will determine the proper routing [20].

During the routing phase, once a TCC involves a certain vessel in a communication process, this will be visible to all the remaining TCCs. Consequently, if a vessel takes part in a communication process, as a repeater of a chain for communication to flow or as an interlocutor with a ship, and it is requested by another vessel to initiate a communication, the TCC in charge of the process will take notes of the fact and will manage the communication the moment it becomes available [21].

The chain will begin to materialize once the responsible TCC of a communication process has determined the necessary chain that is able to establish a channel between two vessels. The channel will be registered it in a common TCC database so that no one else can make use of the resources that are intended to be employed, placing it within the grid belonging to the Maritime Communication Network on Medium Frequency (elevated with respect to AVISOMEF, Figure 3). The initial step will consist in the promotion of a call made by the corresponding TCC to the first linkage, using the listening/calling (channel L/C) of its cell, providing the vessel with the constitutive list of the chain. This first member of the chain will proceed to communicate its membership to the second member, again making use of the listening/calling channel in which this second member operates, and so on, until all the members are notified of their belonging to this communication chain. Subsequently, each one of them will begin to operate on the different operating frequencies that the Maritime Communication Network on Medium Frequency imposes to each cell of its grid (upper grid of Figure 3), leaving the listening/calling channel free as soon as possible.

In order to materialize this process, it is necessary to modify the hardware architecture of the primitive AVISOMEF as shown in Figure 4. Two new receivers with digital selective calling (E and F) will be implemented in the system.

The new receivers will not only allow for the new communication process that is being presented (implemented in Figure 3 within the upper grid), by means of listening only to the channels of the Maritime Communication Network on Medium Frequency, making the information flow in the appropriate direction, but they will also be in charge of monitoring the listening/calling channel (channel L/C), responsible for initiating the communication process and creating the link between AVISOMEF and the Maritime Communication Network on Medium Frequency.

The communication that is established will be a duplex communication. Both receivers (E and F) will listen to the transmitting channels corresponding to the cells before and after the cell in which the ship is located within the Maritime Communication Network on Medium Frequency, helping communication flow in one direction or the other.

Multiplexing technologies can be used to reduce the cost of the reception process of the new network. Therefore, the total number of receivers involved can be reduced. However, this hardware architecture model has been chosen to avoid diminishing the commitment of the receivers involved in AVISOMEF, as the process of remote identification of the different vessels in the area where the Communications Network on Medium Frequency is intended to be established is considered essential.

3. Description of the Employed Simulator

The viability and performance of the system were checked with a simulation of its operation using AIS data on maritime traffic from the years 2018, 2019, and 2020. In this way, we started with a sufficiently representative and updated sample.

These data were processed and stored in a database to permit working with them. AIS data included a timestamp indicating when the vessel was identified. The frequency of identification per vessel was 30 min. Therefore, in order to carry out the simulation, the AIS data were grouped into 30-min time slots. This entailed small deviations on the real situation of the vessels at a specific moment, but due to their speed and the size of the squares proposed for the system, these deviations are negligible.

For example, a vessel with an AIS positioning with a timestamp of 12 August 2020 12:03:14 and another one of 12 August 2020 12:14:17 were in the same time slot during the simulation. As we did not have the localization of the two ships at the same moment, we placed them, approximately, in a range of 30 min in a specific cell of the system.

It is important to highlight that this is the only way to work with this type of data, since it is not possible to have AIS data of the same moment of all the vessels.

To carry out the simulation [22], two complementary programmes were developed: (1) a programme with a graphical interface that allows for the design of the Communication System, locating all of its elements geographically and visualizing the maritime traffic in a specific time slot and (2) another programme that takes as input the system proposed by the first program and performs its simulation. The second program does not have a graphical interface and was optimized to be able to work with the large amount of data available.

The first programme was implemented as a Geographic Information System (GIS). It allows you to load AIS data from the database of two specific dates [23,24,25]. A grouped display of all these data can be completed in specific time slots of 30 min on the OpenStreetMap cartography [26,27]. Once the maritime traffic data has been loaded, it is possible to properly locate the necessary elements for the operation of the System: Traffic Control Centres and grids. The program allows for working with these elements in various ways: add, delete, associate Traffic Control Centres and grids, change their colour to facilitate the interpretation of the system approach, merge two grids to allow the change of direction, etc. Once the design is finished, the System can be exported to a file that can be opened with this first program or can be used as input to the second.

The second programme receives as input a file with the representation of a system to be simulated and is in charge of obtaining the number of communication channels and the length of each of them every 30 min. To obtain these data, the proposed Algorithm 1 is as follows:

Algorithm 1 Simulation Algorithm: Number and Length of Channels.
1:	for each Traffic Control Centre
2:	for each Grid
3:	Cell_init ← Cell_last
4:	while Cell_index different Cell_first + 1
5:	Channel ← Empty
6:	from Cell_index ← Cell_init to Cell_first
7:	if there are vessels in Cell_index
8:	Channel ← add Cell_index
9:	Remove vessel from Cell_index
10:	else
11:	Cell_init ← Cell_index − 1
12:	if Channel is complete
13:	Num_channels + 1
14:	Channel_length ← Number of vessels in Channel
15:	Generate output data for Grid

The algorithm (Algorithm 1) tries to form far-reaching communication channels within the corresponding grid. When we start with the farthest cell of the Traffic Control Centre, the algorithm selects a ship randomly until it reaches the second closest cell of the corresponding TCC. The vessels that are involved are eliminated from the different cells so that they will not be used to establish other communication channels. The first cell, the one closest to the TCC, is not reached, since it has direct communication with it, and it is not necessary to build a communication channel for the ships that are in it.

Once the itinerary through the cells is finished, the completed channel and its length are stored to generate at a later time the output file of the program for each grid.

If a channel cannot be concluded due to an empty cell, the algorithm starts again from the next closest cell to the TCC after the empty cell to search for complete channels of shorter length (Figure 5).

4. Methodology Used and Description of the Experiment

We had to verify the possibility of implementing a Maritime Communication Network on Medium Frequency that would be capable of performing based on the coverage provided by the existence of a maritime traffic route with a certain density of ships that was suitable to withstand AVISOMEF. Therefore, we proceeded with the simulation as follows.

We started the process by choosing a specific maritime area to conduct the experiment.

With the use of other previously conducted studies, we determined that both the maritime area of Spanish management and the Portuguese coastal area located in the Iberian Peninsula could be adequate for this study. In the aforementioned research works, it was shown that these were optimal areas for AVISOMEF implantation.

Subsequently, four Traffic Control Centres (TCCs) were established. It was determined, thanks to previous studies, that four centres were sufficient to perform a remote identification of all the maritime traffic that navigates through the system’s implementation area [28,29]. These TCCs would be located in Finisterre, Tarifa, Valencia, and Las Palmas de Gran Canaria (Canary Islands) (Figure 6). We selected these locations to take advantage of the infrastructures corresponding to the Medium-Frequency Wireless Telegraph centres, belonging to the Telefónica, S.A, a Spanish multinational telecommunications company.

Coverage areas surrounding each of these four TCCs can be drawn that show the areas covered by the radio signal of the TCCs in the worst propagation circumstances. As can be seen in Figure 6, these coverage areas cover more than enough the first cells of each grid associated with them. The ships located in the first cells of each grid are the only ones that must connect directly with the TCC, so in this way good coverage is ensured.

The following step consisted in the establishment of the grids linked to each of the TCCs, achieving in this way a remote identification of all maritime traffic that was navigating through the areas in which the Maritime Communications Network on Medium Frequency was intended to be installed. As we can see in Figure 6, the TCC of Finisterre (yellow) links 4 grids of 4, 10, 5, and 7 cells each and that of Valencia (green) has 5 grids of 5, 3, 7, 6, 6, and 7 cells each. The one in Tarifa (purple) has 4 associated grids of 6, 7, 5, and 5 cells each and the one in Las Palmas (blue) has 3 grids of 10, 10, and 7 cells each.

Once everything concerning the digital cartography of the area under analysis had been implemented in the simulator, we proceeded to supply the data corresponding to the ships that sailed through it. The data processed corresponded to the years 2018, 2019, and 2020. The data renewal frequency was every 30 min. Therefore, worked with all the existing maritime traffic data in the area, with a refreshment frequency of 30 min, 24 h a day, 365 days a year, over a period of three years. We considered that this sufficiently representative sample was able to establish conclusive results that could be extrapolated to maritime traffic routes of similar densities [30].

With the simulator working with all the data, it was asked to provide results on the degree of implementation that the Maritime Communication Network on Medium Frequency could have on the Automatic Remote Identification System of Vessels on Medium Frequency. This requirement from the simulator was the object of study here presented.

5. Results

With the purpose of verifying the possibility, or the impossibility, of establishing a Maritime Communication Network on Medium Frequency supported on an Automatic System of Remote Identification of Ships on Medium Frequency for the proposed area, the simulator was programmed to provide the following results, based on which we could draw conclusions.

The first result provided is the number of “maximum links” that can be made, starting from the vessels located in the furthest cells from the corresponding TCC and ending in the second of the cells closest to said TCC. This involves, in the passage through each cell, one ship located within it. It has already been indicated that the communication does not end in the cell closest to the TCC, considering that here the communication links are guaranteed, without having to require any vessel to act as an element in the chain to be able to establish a connection between the ship and the TCC.

Once this possibility was fully completed, the simulator began to constitute all the possible links that could be formed, starting from the penultimate cell furthest from the corresponding TCC.

We continued working in succession, until single-cell links were established, with vessels that were not yet involved in any of the links previously mentioned.

We opted to request this type of information from the simulator since it is based on the most restrictive links that can be proposed, due to the fact that they are the ones that require the highest amount of resources for the establishment of a communication channel between two ships. We preferred to quantify the transmission capacity of information of the network based on the most restrictive conditions, even though we know that the reality will be much less rigorous, since the network occupancy needs will not always start at the ships furthest from the TCC.

In the presented graphs, these communicative links are referenced as “Number of Channels”.

The second result provided is the “average length of the channel”, understood as the average cell number of the totality of the communication links (channels) that the system was able to establish in each grid.

We want to state the obvious beforehand so that it will not go unnoticed: a certain number relative to the average length of a channel will imply the existence of a sufficient number of ships, as far as possible from the TCC, to counteract the length of the links that could form the closest links. For this reason, when it is seen in a grid of five cells, for example, that the average channel length is 3, this will imply that there will be as many communication links that start from the fourth and fifth cells as channels that start from the second and first cells.

That said, we will present in the form of graphs the results obtained for each year analysed from January to December.

First, a general graph of the system, corresponding to each of the years under study, is provided for the entire geographical area in which the Maritime Communication Network on Medium Frequency is intended to be located. Here, the data provided are the number of communicative links (channels) established at each moment (Figure 7).

Next, a detailed analysis corresponding to each grid associated with each of the four TCCs is presented for each of the three years analysed (Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14 and Figure 15). Here the data presented, by grid and year, are those related to the number of communication links (channels) formed at each moment (Figure 8, Figure 10, Figure 12, and Figure 14) in addition to the average length of the channels constituted throughout the year in real time (Figure 9, Figure 11, Figure 13, and Figure 15).

6. Conclusions and Future Research

From the experimental results, the following conclusions can be drawn:

The total number of channels constituted over different years and displayed in the Figure 8, Figure 10, Figure 12, and Figure 14 brings to light a clearly sufficient ability to establish a fluid traffic of radio-packets within the Maritime Communication Network on Medium Frequency, the viability of which is the object of this study. It can be stated that the communication would take place without the existence of delays that slow down the correct functioning of the same.

It was confirmed that there was an availability that was repeated throughout the years that ranged from a minimum of about 400 channels to a maximum of 800 to be used simultaneously. Taking into account that a typical transmission in digital selective calls in decametric/hectometric waves (MF/HF) lasts between 6 and 7 s [2], we can get an idea of the enormous communication absorption capacity of the proposed network.

The communication system is based on maritime traffic due to the geographical position of the Iberian Peninsula. This makes the system viable even in different situations that cause a decrease in traffic, such as the one experienced in 2020 due to COVID-19.

The number of channels available in each of the different geographical areas analysed individually reveals an availability capable of absorbing more than enough of the communication generated there, without any queues due to delays.

The areas with the worst data are those known as “Blue area 1” and “Blue area 2” (Figure 8 and Figure 9), ranging between extremes going from 4 to 15 possible channels simultaneously. The same situation was repeated throughout the three years, so it cannot be attributed to the pandemic effect due to COVID-19. There will be a number of stable channels in these two areas, which would provide permanent communication support to the ships underway through them without producing, at any time, any shadow areas that would be subject of interruption of the potential communication capacity.

In areas such as “Green area 1” there is a simultaneity of channels that varies between an availability that goes from 80 to 160, although remaining almost always well above 100 (Figure 10).

In some areas, the density of traffic and therefore the number of channels and their length increase during the summer months. This phenomenon can be clearly observed in the green area grids 2 and 4 in the case of the number of channels (Figure 10) and in the green area grids 4 and 5 in terms of channel length (Figure 11), which during the summer months reaches the maximum possible. This variation can even be seen in Figure 7 of the total channels of the system.

In areas with a lower density of maritime traffic, such as the yellow grid 2 area, there is a significant variation between the channels available during the week and those available at the weekend, due to the presence of a large number of fishing boats during the week (Figure 14). As can be seen, for example, in Figure 14 and Figure 15, the system is capable of working correctly in these areas during the weekend.

If we look at the areas that are already separated from the coast, as in the case of “Yellow area 4”, an average of 15 channels can be ensured (Figure 14), with no shadow areas generated, as we saw previously for other geographical areas.

The data provided by the experiment with reference to the average length of the channels was excellent. We must bear in mind the restriction we imposed on ourselves, by which the different communication channels began to be formed from the furthest TCC cell, with the aim of getting closer to it. Undoubtedly, this means that there is a decrease of resources as channels are formed. However, the averages are satisfactory as we obtained highly centred figures within their ranges. Therefore, the experiment was applied in an appropriate maritime zone in terms of traffic density, satisfying the requirements imposed by the Automatic Vessel Identification System on Medium Frequency (AVISOMEF) on which the proposed Maritime Communication Network on Medium Frequency is based.

To conclude, based on the results obtained in the experiment described above, we can state that it is feasible to consider a Maritime Communication Network on Medium Frequency for maritime routes that meet minimum requirements of traffic density, taking advantage of the unexpended resources of the Global Maritime Distress and Safety System (GMDSS).

Future research should implement the programmes designed in this study to study other maritime routes that are capable of supporting the system and the interconnection between them, making it feasible to consider a Maritime Communications Network in Medium Wave that is capable of bringing together a good part of the global maritime traffic.

It is also within our objective to communicate to the International Maritime Organization (IMO), when the possibility arises, the presented potentialities in this work that have not yet been tapped by GMDSS.