User terminals with multipaneled phased arrays and omnidirectional antennas applied in an elliptical medium Earth orbit satellite communication system for the Arctic

1 Introduction

It is increasingly important for the northern countries to have ubiquitous maritime communications coverage of the vast Arctic Ocean areas north of Canada, Alaska, and Scandinavia, including the Svalbard archipelago, the Arctic areas up to the North Pole, and Greenland. The communication needs include remote sensing, Internet of Things (IoT), climate surveillance and observations, maritime activities like fishery, tourism, and military activities. The applications include reporting data from buoys, unmanned submarine/surface vehicles, sensor units, and human activities on the iceshelf, iceshelf map updates, and iceberg navigation.

While global mega communication constellations have become common over the last 10–15 years (Chougrani et al., 2024; Leyva-Mayorga et al., 2022; Del Portillo et al., 2019; Homssi et al., 2022), their deployment remains a high-cost endeavor. Recently, the demand for enhanced security has driven a shift toward national and regional networks (EU Space Policy, 2025). For smaller countries, like the Nordics, to participate in this trend, developing more affordable low-cost solutions is essential.

Accordingly, we will focus on using an elliptical medium Earth orbit (eMEO) satellite system with three satellites in a single plane, giving circumpolar coverage down to the southernmost point in Norway at latitude 58°. This low-cost eMEO orbital system may be an advantageous system compared to a low Earth orbit (LEO) system that needs about 20–30 satellites for continuous coverage of the same area, depending on orbital altitude and required elevation. The main disadvantage of an eMEO system compared to LEO is the increased path loss and delay. The needs mentioned above could be covered with voice, medium-rate data communications up to 9.6 kbit/sec for a VHF link using handheld user terminals with omnidirectional antennas. For a smaller coverage area, like the Nordic countries, it is possible to increase the bitrate by increasing the satellite antenna gain and repointing the satellite’s on-board antenna. For an S-band link, the bitrate varies between 128 kbit/sec and 1,024 kbit/sec, depending on the terminal antenna gain.

eMEO constellations covering the northern hemisphere have been reported in several references (Ulybyshev and Doniants, 2002; Ulybyshev et al., 2005; Ulybyshev, 2009; Ulybyshev, 2012), describing different orbital options, including an eMEO orbit with a period of 1/6 of a sidereal day (∼4 h). An experimental satellite (TacSat-4) has also been operating in a similar orbit (Anderson et al., 2013).

A crucial technology for the user terminal (UT) is recent advances in the electronically steered antenna solutions that can track moving satellites and perform handover from a satellite going below the horizon to a rising satellite (Agmon, 2024; He et al., 2016). Furthermore, antenna systems that have several panels to which a steerable array of antenna elements is attached may be an attractive solution for mounting on a vessel. An optimal design of such a system can significantly enhance the user terminal’s gain-to-noise temperature (G/T) and equivalent isotropically radiated power (EIRP) and contribute to increased capacity of the communications system.

Onboard satellite technology has evolved rapidly during the last 10–15 years. Advances in radio frequency integrated circuit (RFIC) technology and digitization at radio frequency (RF) have enabled powerful solutions for adaptive multibeam-steering of satellite antennas that have enhanced capacity and spectrum efficiency in satellite communications considerably (Liang et al., 2009; Bazzi et al., 2017). However, we have not pursued this theme here and instead focused on the user terminals and propagation issues.

The organization of the article is described in the following. In Sections 2.2 and 2.3, we present different LEO and MEO systems, discuss their coverage and performance, and select an elliptic medium Earth orbit (eMEO) satellite system with three satellites in a single plane for coverage of the Arctic. We present orbital parameters such as elevation, azimuth, and distances for some selected places (latitude and longitude) on the Earth. In Section 2.4, we present an analysis of the orbits defining the distribution in elevation and azimuth seen from three typical sites within the coverage area. The hemisphere above the user terminal (UT) positioned at a specific latitude and longitude point on Earth is divided into grids of elevation (0°–90^°) and azimuth (0°–360^°) pairs. A histogram comprising all the grids will show the distribution of elevation and azimuth for a particular site on the Earth’s surface.

In Section 3, we choose the frequency bands and the associated user scenarios to be subject to further assessment and analysis. In Section 4, we discuss antenna solutions for the different frequencies, concluding that omnidirectional antennas are best suited for VHF frequencies (Slade, 2015). The electronically steered antennas with multiple panels of antenna element arrays are particularly suited for the S-band (2–4 GHz). We assess and analyze an antenna configuration from He et al. (2016), denoted a frustum, with five flat antenna panels and electronically steered antenna arrays on each panel. This antenna concept can be used for stationary or semi-stationary use on land or on a vessel. We have proposed and analyzed ways to optimize the performance of this antenna configuration. In Section 5, different propagation mechanisms in satellite communications are addressed. These are diffraction, atmospheric loss and reflections, and scattering from the sea surface. We discuss the possible impact of these mechanisms on the chosen frequency bands. We find that diffraction loss will occur at very low elevation angles, and reflections and scattering from the sea at these angles will have a significant impact on the VHF band.

In Section 6, we present typical link budgets for up- and downlinks for VHF and S-band links. We end this section with a discussion of the potential information rate for the systems. In Section 7, we sum up and conclude our proposed eMEO satellite system.

2 Satellite systems

2.1 Orbital requirements

To provide coverage of the Arctic region, it is possible to utilize multiple satellites in various orbital configurations, such as LEO, medium Earth orbit (MEO), elliptical MEO, and high Earth orbit (HEO).

The geostationary Earth orbit (GEO) has no coverage above approximately 75°–78° north, depending on the location of the satellite. The high altitude of HEO orbits complicates the use of small terminals with low-gain antennas. Furthermore, both GEO and HEO missions require expensive launch vehicles and generally large and expensive platforms. Accordingly, the most suitable orbital regimes to be discussed are LEO (500–1,000 km altitude) and MEO (5000–10,000 km altitude), including eMEO configurations.

Depending on the services, the requirements for orbits covering the Arctic include the following:

• Continuous coverage down to a specified latitude
• Avoidance of the radiation belts (Van Allen belts) to increase satellite lifetime
• Minimum fuel consumption for orbital control
• Overall cost minimization of the system, that is, the use of small commercial spacecraft and a common launcher for all spacecraft in the system, which are launched into one single orbital plane

2.2 Feasible orbital constellations

In order to meet the coverage requirement, the number of satellites and orbital planes will increase as orbital altitude and eccentricity decrease. The required number of satellites in a LEO or circular MEO polar system meeting the coverage requirement is calculated (see Table 1 using the method described by Beste (1978).

TABLE 1. Number of LEO satellites, orbital planes, and altitudes for circumpolar coverage down to the specified latitude.

As shown in Table 1, a LEO constellation with altitudes ranging from 800 km to 1100 km requires between 21 and 30 satellites distributed across three orbital planes to achieve full circumpolar coverage above 58^° north. An MEO constellation requires six satellites in two different planes. Consequently, neither LEO nor circular MEO constellations will meet the cost requirements due to the high system costs associated with a large number of satellites and multiple launches.

In this article, we have analyzed an elliptical MEO with three satellites in one orbital plane for the Arctic region and compared it with a LEO alternative; see Table 2.

TABLE 2. Arctic LEO and eMEO comparison.

Conservative number, which depends on technology and depth of discharge, example specifications from Berlin Space Technology.

A very important advantage of eMEO over LEO is the reduction in the number of satellites required for Arctic coverage, coupled with more than twice the expected lifetime due to fewer temperature and battery charging cycles. The longer the distance, the more complex antennas and RF circuitry, improved shielding, and propulsion are needed to move from a LEO to an eMEO system. This will, of course, result in heavier and costlier satellites but is well within the capabilities of high-end LEO bus manufacturers. The eMEO solution would still be very attractive even if the satellite cost becomes twice the cost of a small LEO.

An attractive alternative is an eccentric MEO orbit with an argument of perigee ω = 270^°, which provides excellent coverage of the northern hemisphere with a limited number of satellites in a single orbital plane. To minimize fuel consumption for orbital control, the inclination should be i = 63.4^° (the critical inclination). It is also desirable to minimize exposure to the Van Allen radiation belts. One approach is to position the orbit between the inner and outer radiation belts, within the so-called “safe zone,” by selecting an appropriate orbital eccentricity. To simplify frequency coordination, repeating ground tracks are also recommended. Interference and frequency coordination is a comprehensive task that is not prioritized in this article. The importance is unquestionable, and we refer to relevant ITU documents and recent publications on the theme (Rec. ITU-R S.1503-4, 2023; Rec. ITU-R S.1528, 2001; Ortiz et al., 2026; Zhu et al., 2025).

In the following, we are considering a constellation with parameters defined in Table 3.

TABLE 3. Orbital parameters for the three satellites in an eMEO constellation.

The constellation consists of three satellites with an orbital period equal to one-sixth of a sidereal day (3^h 59^m, or approximately 240 min). The time separation between adjacent satellites is 240/3 = 80 min, corresponding to true anomalies of 0°, 144.3°, and 215.7°, respectively. The orbital eccentricity is set to 0.3. The equatorial altitude of this orbit is 5,260 km, which positions the orbit within the region between the Van Allen belts (Fortescue et al., 2011).

Increasing the eccentricity above 0.3 will lower the equatorial altitude (e.g., to 2358 km for an eccentricity of 0.4), thus placing the orbit well within the first Van Allen Belt. While eccentricities below 0.3 keep the orbit within the “safe zone,” satellite visibility time will be reduced (e.g., by approximately 15% at an eccentricity of 0.2). In this case, we must add more satellites to the constellation to maintain continuous coverage down to 58° north. This is not an option.

Figure 1 shows the orbit ground track for three satellites over six periods.

2.3 Coverage of an eMEO constellation

We have simulated elevation, azimuth, and distance for three locations in the coverage area (i.e., from the North Pole to latitude 58^° N). The locations are given in Table 4.

TABLE 4. Locations used in the analysis.

The elevations above the horizon for all three sites and three satellites in an eMEO orbit for a time span of 24 h are shown in Figure 2.

We see from the diagrams that the minimum elevation for the best of the three satellites is 10^° at Kirkenes, 6^° at Nuuk, and 2^° at Lindesnes. Hence, Lindesnes is at the rim of the coverage area.

The longest pass (from e = 0 to e = 0) for Lindesnes is 135 min, and the shortest is 70 min.

Increasing the number of satellites to four in the constellation will increase the minimum elevation angle at Lindesnes to about 6^°.

The distance to the satellite changes with time, as shown in Figure 2. The figure shows the distance from Lindesnes to the satellites for six orbital periods. The maximum difference in distance is 5700 km or a delay difference of 19 msec. For the Lindesnes site, the maximum distance is 14,600 km, and the minimum is 8900 km. This will give a difference in free space loss of 4.3 dB for the two extremes.

2.4 eMEO orbit elevation and azimuth distribution

The MEO orbit system described in Section 2.2 has been simulated. In the simulation, 3000 elevation and azimuth data points are recorded every 30 s over a 1,500-min period for all three satellites. These data are collected, analyzed, and presented in diagrams below.

In Figure 3, the x, y, and z-axes define the topocentric-horizon frame with its origin at point O.

A user terminal (UT) has its position given by its longitude and latitude U_lon, U_lat at point O on the Earth’s surface. In the topocentric-horizon axis, the x-axis points east, the y-axis points north, and the z-axis points local vertically up at the UT position (i.e., zenith pointing out into space).

The whole semi-sphere seen from the UT at the origin O is divided into J grids numbered j ∈ 1 , 2 , 3 , … . J . Outcomes of the elevation/azimuth analysis of satellites in an orbital system can be associated with grid areas A_j and the central point c_j. The central point has an elevation and azimuth pair (ϕ_j, θ_j). The grid area of grid j is A_j = Δϕ_j Δθ_j. The elevation grid increment is constant Δϕ for all ϕ ∈ [0 → 90^°], and we choose Δϕ = 5^°. The grid sizes should be close to equal for all grids. Hence, the increment Δθ_j for azimuth must be changed with increasing elevation to obtain close-to-equal grid sizes over the whole semi-sphere (or equivalently, the same spherical angle in steradians). The azimuth covers θ 0 ° → 360 o. The azimuth increment, Δθ, at the lowest elevation grid central point 2.5^°, is 20^°. Figure 3 shows the result of this strategy. The total number of grid points is J = 206.

The total number of all pairs of elevation and azimuth data from orbit simulations of all three satellites is denoted M. The pairs of elevation and azimuth data from the orbit simulation must be a time series with a constant sampling interval T (uniform sampling with T = 30 s). The number of outcomes from simulations for all three satellites falling into grid j is M_j. All M_j values, j ∈ [1,2 … J], are collected. A two-dimensional discrete distributionis defined in Equation 1, meaning the number of outcomes of any three satellites in the orbiting system falling into grid j with area A_j = Δϕ_j Δθ_j, at central point c_j = [ϕ_j, θ_j] seen from the UT.

D j A j , c j = D j ϕ j , θ j (1)

We have analyzed the simulations of three eMEO satellites seen from the Lindesnes site in Norway (latitude 58^°, longitude 7^° E). We assume that the user terminal will, at any time, choose the satellite with the highest elevation angle. Hence, we define the occurrence of outcomes by counting the satellite with the highest elevation angle and discarding the other two satellites. The result is the histogram displaying the number of outcomes M_j, j ∈ [1,2 … J] shown in Figure 4. The histograms show that we have many observations in the lowest elevation interval 0° → 5^° at this site.

The marginal distribution of outcomes over all elevation intervals ϕ_j is important for assessing link performance. To evaluate the marginal distribution, we define the grids c_j = [ϕ_j, θ_j] by their central points, as shown in Figure 4. The elevation has K intervals k ∈ 1 , 2 , 3 , … . K

As stated earlier, the azimuth increment must change with the elevation to have a nearly equal grid size. The number m of azimuth grids at elevation k is denoted as L_k:

m ∈ L 1 , L 2 … L k , … . … . , L K − 1 , L K .

At elevation k = 1 (the lowest elevation grid point), the number of azimuth grids is L₁, which is the highest number 18, decreasing to L_K = 1 for the highest elevation ϕ_K = 90^°-Δϕ/2.

The marginal distribution of outcomes in the 18 elevation grids is given by adding outcomes in azimuth for each elevation, Equation 2:

D k ϕ k , a l l a z i m u t h a n g l e s = D k ϕ k = ∑ m = 1 L k D k ϕ k , θ m , k ∈ 1 , 2 , 3 , … K . (2)

The result for the Lindesnes site is shown in Figure 4, upper right.

The mean value of the elevation angle is given by Equation 3

E ϕ k = ∑ k = 1 K ϕ k · D k ϕ k ∑ k = 1 K D k ϕ k . (3)

The mean elevation angle is E ϕ = 34.9 ^° for Lindesnes, 39.1^° for Kirkenes and 38.4^° for Nuuk.

The 2D histogram for Lindesnes shows a spread of outcomes over most elevation/azimuth combinations. The elevation histograms clearly show that Kirkenes has a favorable distribution in elevation compared to the other sites. Lindesnes, located at the rim of coverage, has a considerable number of outcomes at low elevation.

3 User scenarios and choice of frequency bands

Some example user situations include: a) Maritime satellite communications such as telephony with a low or moderate data rate, b) communications to buoys or observation posts on ice or land, c) human activities on ice and tracking icebergs, d) communications to activities on Arctic islands. Other applications are data harvest from observations by radar stations or other sensor stations. Yet another application is the use of the UT for navigation as backup for GPS, Glonass, Galileo, and others in jamming situations of these systems.

We have selected the following two frequency bands assigned to satellite communication for assessment of the link budgets and antenna discussions.

• VHF: Uplink 200 MHz and downlink 300 MHz.
• S-band, 2 GHz communications for up- and downlink

Another important user condition is the availability of power. On a ship, the availability of power is usually not a problem. Power may be restricted on buoys, unmanned observation platforms, icebergs, or ice shelves.

4 Antenna solutions

The system performance strongly depends on the UT antenna solution. In the following, we discuss aspects of UT antenna solutions for fixed and handheld UTs and aspects of a moving antenna platform on a ship at sea.

Modern maritime satellite terminals increasingly use steerable phased-array antennas. These antennas can track satellites in different orbits, like low Earth orbit (LEO) and medium Earth orbit (MEO), without the need for mechanical movement. The electronic steering maintains satellite lock and can also compensate for the vessel’s pitch, roll, and yaw. The phased-array antennas are best suited for high-frequency applications, and their size may become unacceptably large at low frequencies.

Detailed antenna design is outside the scope of this article. However, we will discuss the antenna solution and performance from a system point of view, keeping the following in mind:

• The user terminal (UT) tracks a satellite as it moves and performs seamless handover to a new rising satellite when the elevation angle of the rising satellite becomes larger than that of the currently tracked satellite.
• One drawback of an electronically steered antenna is the scan loss.
• The UT antenna on a vessel must compensate for the 3-axis angular movements, pitch, roll, yaw, and the ship’s heading.
• The UT on a stationary or semi-stationary site can be made simpler and at a lower cost than that on a ship. However, a UT in remote areas may have limited power available that may restrict the EIRP and G/T of the UT.
• The antenna must operate at all azimuth angles (0°–360^°) and from a minimum system specified elevation angle to 90^°. This requirement is valid for both a stationary UT and a UT on a vessel. The scan loss using an electronically steered antenna will be unacceptably high, covering the above azimuth and elevation requirements with only a single antenna panel. Hence, there is a need for a multipaneled antenna system. The antenna steering algorithm must be able to utilize all antenna elements on all antenna panels, irrespective of the position of the satellite and the movement of the vessel.
• The antenna on a vessel should be circularly polarized to avoid the 3 dB loss of a linearly polarized antenna. It is assumed that the radio waves from the satellites are circularly polarized.

Our discussion includes implementation feasibility but not detailed design, depending on the user scenarios and carrier frequencies. Antenna arrays are impractically large and unfeasible at VHF. In this case, an omnidirectional antenna must be used.

4.1 Omnidirectional antenna for VHF

We have simulated an example linear polarized vertical dipole omnidirectional VHF antenna shown in Figure 5. The VHF dipole measures 540 mm in length and has a diameter of 18 mm. Its peak gain is 2.1 dBi. The antenna data are used in link budgets for the VHF presented in Section 6.

4.2 Electronically steered antennas

An array of antenna elements on a flat panel can operate as an electronically steered antenna (Agmon, 2024; Delos et al., 2020). By monitoring the delay to each antenna element, the array antenna boresight can be steered in a direction θ_s, being the angle between boresight and a vector that is perpendicular to the flat antenna panel. The angle θ_s is the scan angle. In practice, the delay is implemented using a phase shifter. In our case, with a narrow band system, the error using a phase shifter instead of a delay shifter is insignificant.

The composite antenna pattern consists of an array term and an element term. If the distance d <0.5λ between antenna elements on the array, an unwanted side lobe (grating lobe) will appear. Usually, a distance d ≥ 0.5λ is used. A disadvantage of beam steering using a flat panel is that the effective antenna area changes with the scan angle θ_s, giving a reduced effective area, A e θ s = A e 0 · ⁡ cos θ s, where A_e (0) is the effective antenna area when θ_s is zero. When θ_s = 60^°, the scan loss is 3 dB caused by the reduced effective antenna area.

For digital control of the antenna elements, we assume the following:

• A modern RF front-end unit consists of a duplexer, a high-power amplifier (HPA), a low-noise amplifier (LNA), sampling, and analog-to-digital (A/D) and digital-to-analog (D/A) converters directly on the RF signal. A modern RFIC (e.g., Texas Instruments AFE7950 (Texas Instruments data sheet, 2025)) can sample, perform A/D and D/A conversion, and handle analog input/output units for four antenna elements in an array. See Supplementary Appendix SA. With N_a antenna elements, there are N_a/4 numbers of RFIC front-end units of this kind.
• A multipaneled antenna system must cover azimuth 0° → 360^° and elevation 0^° → 90^°. Figure 6 shows a multipaneled frustum configuration with five faces.
• We assume digital phase control and level control of each antenna element, both transmitting and receiving.

Controlling the multipaneled antenna system to achieve optimum performance is a complex task. In the following, we will discuss some basic functionalities and requirements:

• The multipaneled frustum antenna configuration (He et al., 2016) shown in Figure 6 has five panels. This configuration will be discussed and analyzed. It illustrates the advantages of a multipaneled antenna configuration.
• In the following, we discuss solutions in which i) one panel, the most favorable, participates in reception (RX) and transmission (TX) or ii) all panels can be continuously active and able to participate in RX and TX. The latter is a step up in complexity.
• The changeover from one satellite to a new satellite can be made seamlessly. The panel that maintains communication with a satellite nearing a position below a predefined elevation angle continues its connection, while another panel initiates and establishes communication with the ascending satellite. When communication with the ascending satellite is established, communication with the descending satellite is released.
• Once the connection to the new satellite is established, the electronically steered antenna multipaneled system must track the orbital movement of that satellite and, in addition, when on a vessel, compensate for the angular movement caused by the sea waves. This compensation is implemented by a control system based on a system of sensors in which the angular rate sensors (gyrators) are the most important. The angular movements may be significant in harsh weather, and the use of the most favorable antenna elements in the panels may continuously change with the vessel’s angular movement. Furthermore, the digital phase control of the direction of the beam from the active elements in the panels must be continuously updated.
• The main performance criterion for the antenna is the effective antenna gain. With the five-paneled frustum antenna configuration shown in Figure 6, the performance changes with the scan angle. This performance loss will change with a vessel’s angular movement on the sea. However, using the frustum geometry with multiple panels overcomes this problem.

4.2.1 Pyramidal frustum antenna array configuration

Figure 6 shows a multipaneled electronically steered antenna (He et al., 2016) comprising five flat panels.

Each panel consists of a 2D antenna array that can be steered. A multipaneled antenna system is required to reach satellites at all azimuth angles 0 ° < θ ≤ 360 ° and elevation angles 0 <ϕ < 90^° There are two important angles describing the frustum; see Figure 6. These are: a) α_s the side angle DOH (see the capital letters on the figure) of the frustum, where O is the origin of the 3D x, y, z topocentric-horizon axis system, and b) γ is the side face elevation angle. In the following, we describe and assess the trade-offs and compromises between antenna complexity, physical size, and performance.

4.2.2 Stationary or semi-stationary UT on a site

In this case, the UT is on a fixed site determined by the site’s latitude and longitude (U_lon, U_lat) on a rotating Earth. The base plane EFGH of the frustum is in the horizontal x-y plane.

He et al. (2016) analyzed and optimized the five-sided pyramidal frustum with respect to the two angles α_s and γ shown in Figure 6. Notice the (x, y, z) coordinate system showing the orientation of the frustum in the topocentric-horizon frame. For practical reasons, we want all five panels to be the same size. Hence, we choose the height l 4 of the trapezoidal side panels to be equal to the side l 1 of the top panel. Furthermore, we choose γ = 41°. Analyzing the frustum geometry in Figure 6, we get the following Equations 4,5 defining l 3, h, α s expressed by γ and l 1, the latter is set to unity in the expressions.

l 3 = 2 · ⁡ cos γ + 1 , h = sin γ , α s = atan h · 2 . (4)

The maximum scan loss angle θ_max is given by the following expression (He et al., 2016):

cos θ max = sin γ cos π 4 sin ϕ min + cos γ cos ϕ min . (5)

The max scan angle dependence on the side face angle, with ϕ min as the parameter, is shown in Figure 7. We choose ϕ min = 0 in the following analyses. The data for the frustum geometry are: γ = 41 ° giving a maximum scan angle for the side panels θ_max = 41°, the size of the five panels is 0.20 m × 0.20 m, the height of the antenna system is h ≈ 0.15 m, and the bottom side l 3 ≈ 0.55 m. This corresponds to a maximum scan loss of 1.2 dB for both the side panel and the top panel. We chose this antenna for the S-band at 2 GHz. With a panel size of 0.2 × 0.2 m, we can have four antenna elements on each panel.

In the following two subsections, we discuss two concepts: a) Use of one active communicating panel having the most favorable scan angle at any time, and b) optimized use of three to five panels taking place in the communication.

4.2.3 Optimized use of each antenna panel separately

The top panel has the lowest scan angle compared to the side panels when the elevation angle ϕ> α_s. The side panels experience maximum scan loss when the communicating satellite has azimuth angles of 0°, 90°, 180°, or 270°, that is, along one of the vertices of the pyramidal frustum, and ϕ < α_s. The vector pointing from the UT starting at O toward a satellite will be either in or close to the x-z plane or the y-z plane for these four maximum scan loss cases. For any of the four side panels to have the most advantageous scan angle, the elevation angle to the satellite must be 0 < ϕ ≤ α_s. If, for example, the vector pointing toward a satellite has an azimuth angle 0 < θ ≤ 90 ° and an elevation angle 0< ϕ ≤ α_s, the side panel HDCG is the most favorable antenna array panel of the five panels with respect to scan angle. When ϕ> α_s, the top panel is the most favorable.

With a stationary or semi-stationary UT, the steering principle of the electronically steered beam can be summarized as follows.

• The control algorithms always choose the most favorable satellite, that is, the satellite with the highest elevation angle.
• The concept of one panel active needs a handover between panels.
• The phased steered array of different antenna panels can generate separate beams, which can ensure no link interruption when switching between two satellites.

The antenna gain of a flat panel array antenna increases with the number of elements according to 10·log₁₀(N_a), where N_a is the number of radiating elements in the array. For a uniformly excited, equally spaced planar array, the antenna gain is:where G_element (ϕ, θ) is the gain of a single antenna element at angles (ϕ, θ), and L_losses includes ohmic loss factors.

Antenna gain dB = 10 · ⁡ log 10 N a + G element ϕ , θ − L losses ,

Using G_element = 6 dBi (at boresight), N_a = 4, and L_losses = 0.8 dB, the maximum frustum antenna gain is (Equation 6)at zero scan angle.

A n t e n n a G a i n = 10 · ⁡ log 10 N a + 6 − 0.8 d B i = 11.2 d B i (6)

An element antenna with a 90° half-power beamwidth gives

Antenna gain = 10 log₁₀(N_a) + (5.2-3) – (scan angle loss at 45°) = 10 log₁₀(N_a) + 0.7 dBi = 6.7 dBi at a scan angle = 45°

4.2.4 Optimized use with all panels active simultaneously

The top panel ABCD and two or more side panels will receive power and may add to the total received power, depending on the azimuth and elevation angles to the satellite. This opens the option for optimizing the use of the frustum geometry antenna, having three to five panels being active simultaneously, and adding power from them. The spherical antenna configuration described by He et al. (2016) addresses this principle. We here describe the application of this scheme on the frustum configuration.

In Equation 7 we define the term Area Gain:

A r e a G a i n = T o t a l r e c e i v e d p o w e r f r o m a l l a c t i v e p a n e l s R e c e i v e d p o w e r f r o m o n e p a n e l w i t h n o s c a n l o s s . (7)

This principle is characterized by the following:

• Three to five antenna panels are continuously active. This strategy is increasingly interesting because of new RFICs available (Texas Instruments data sheet, 2025. AFE7950).
• Using the principle with all receive, transmit, and antenna elements active, there is no need for handover between antenna panels.
• There are four antenna elements in each panel. Along with the antenna elements, there are a) four hybrid circuits to split RX and TX, b) four power amplifiers, c) four low-noise amplifiers, and d) four A/D and D/A converters performing direct RF to digital conversion in both directions.
• State-of-the-art technology facilitates extensive digital signal processing both in the receive and transmit directions for a) digital phase control of steered antenna arrays, b) digital signal processing (DSP) algorithms for estimating the signal-to-noise ratio E_s/N₀, and c) DSP algorithms for adaptive co-phasing and optimum combining of radio signals from multiple panels.

The following examples show the potential antenna system gain where three panels of the frustum configuration are active. We refer to Figure 6. When the satellite azimuth is in the x-z plane, the azimuth angle is θ = 0° (along the x-axis), the elevation is ϕ = α_s, and the top panel and the side panels HDAE and HDCG can receive and transmit power. The scan angle, θ_s, to all these three panels is close to equal (41°), and we can add power from each of the three panels corresponding to an area gain = 10∙log₁₀ (3∙cos (41°)) = 3.5 dB, which is the gain compared to the concept with one active panel. A notable example occurs when the satellite is directly overhead at the zenith. The scan angle loss is 0 dB for the top panel and cos(γ) = cos (41^°) = 0.76 for each of the four side panels. Adding the power from all panels, there will be an increase in receive and transmit power of 6.05 dB compared to the power from the top panel only. In this case, modifying Equation 6, the max gain is (11.2 + 6) dBi = 17.2 dBi.

The practical implications of the optimized use of the frustum configuration are not trivial and are addressed in the following:

• Total UT power consumption and temperature problems on electronic boards
• The need for adaptive co-phasing and combining of digitized radio signals from multiple antennas
• The need for calibration caused by variations in the delay in the analog part of electronic boards

Solutions for co-phasing and combining radio signals are:

• Forward control: The geometry of the frustum configuration and placement of antennas is known, and the azimuth and elevation to the satellite are deterministic and known. Hence, it is possible to calculate the phase differences and correct phases digitally before combining them.
• Feedback control: In satellite systems, at the link layer, known symbols are inserted in the burst frame format for maximum likelihood- (ML) based carrier frequency, phase, and time synchronization needed when the E_s/N₀ is low (Mengalia and D’Andrea, 1997). Carrier phase estimation can be used for co-phasing radio signals from multiple panels before they are combined.

In Supplementary Appendix SA, we have presented an example implementation block diagram of the frustum antenna configuration in which up to five panels are combined at RX and simultaneously transmitted at TX.

4.2.5 A user terminal on a moving platform, such as a ship

This case is a step up in complexity in terms of sensor technology, control algorithms, and computer resources. The worst case of maximum antenna tilt in any angular direction is Ψ_m = 10^° (Reference to max pitch and roll: IMO MSC.192 (79) revised performance standard for radar equipment). This worst case can occur in any azimuth direction. The procedure to calculate the scan angle to any of the five panels of the frustum is according to well-known procedures in linear algebra (Lang, 1979). We refer to Figure 6.

• The [x, y, z] coordinates of all points A, B, C, D, E, F, G, and H of the frustum geometry can be calculated by the parameters γ, α_s, and l 1, choosing l 1 = 1.
• We define a unity vector O P →, starting from the antenna origin O toward the satellite, ending at point P. The elevation/azimuth angles of O P → are chosen to go through a vertex in the frustum that is the worst-case azimuth angle. Hence, we define the unity vector O P → to be in the y-z plane with azimuth θ = 90^° and elevation ϕ. Furthermore, worst-case sea movement implies a clockwise rotation of the frustum along the x-axis by an angle Ψ_m (ref Figure 6). The new points of the rotated frustum are denoted A′, B′, C′, D′, E′, F, G′, and H. The [x, y, z] coordinates of the new points can be calculated using the rotation angle Ψ_m.
• The plane B′, C′, G′, and F in 3D is defined by the two vectors F G ′ → and F C ′ →. The other four planes are defined in the same way. Taking the cross product of the two vectors F G ′ → and F C ′ →, we find a new vector that is perpendicular to the plane of the side face B′, C′, G′, F of the frustum. This vector is denoted as V → = F G ′ → × F C ′ →,
• The scan angle is calculated using the definition of the dot product, Equation 8.

cos θ s = V → · O P → V → O P → (8)

We have analyzed with the “worst-case” azimuth angle being along the y-axis (see Figure 6) with respect to the two side panels D′C′G′H and B′C′G′F versus the elevation angle ϕ ∈ 2.5 o → 87.5 o. Furthermore, we have analyzed the gain with the two side panels and the top panel being active compared to a single panel with no scan loss. The results are presented in Figure 8.

From Figure 8, we see that at elevation angle ϕ = 10°, the scan angle is 61.5^°, and the corresponding scan gain for this panel is G_s = 10 log₁₀ (cos 61.5^°) = −3.2 dB.

4.2.6 Area-gain analysis over azimuth–elevation grids

We analyzed the distribution of area gain over all 206 grid points (see Figure 3) and present the results in Figure 9. In the analysis, we have discarded contributions from a panel if its scan angle is below 60^°. The histogram in Figure 9 clearly shows the reduced performance at low elevation. The tilt in this analysis is Ψ = 0^°.

4.2.7 Impact of frequency bands on antenna design

Table 5 displays the following parameters for the frequency bands defined in Section 2:

• Antenna gain of an antenna panel, and the number of antenna elements in the panel.
• The impact of restrictions on the antenna size on a vessel.
• Choice between electronically steered antennas and omnidirectional antennas.
• Number of antenna elements.

TABLE 5. Frequency bands, antenna size, and performance.

5 Propagation mechanisms

In this section, we focus on propagation issues to/from a vessel on the open sea. The propagation mechanisms can be divided as follows: 1) tropospheric losses due to atmospheric gases, water vapor, and precipitation (rain, snow), and 2) ionospheric losses due to the large amount of electrons and ions in the upper atmosphere, and scintillations due to the ionosphere. An excellent survey of these impairments is given by Rogers and Bouchard (2012). Furthermore, we have diffraction loss, reflection, and scattering from the sea surface.

5.1 Ionospheric scintillations

Scintillation is caused by solar radiation and ionization in the ionosphere. The impact of scintillation on the propagation is mainly based on models described in the ITU-R M.531-14 document (Recommendation ITU-R P.531-14, 2021).

The scintillations may be described as a series of ionospheric scintillation events. During one such event, it may be meaningful to define a short-time statistic.

During a scintillation event, the Nakagami distribution (It is referred to equation 7 in ITU-R P.531) is accepted as an adequate description of the intensity distribution.

Temporal behavior, given an event, may be characterized by the autocorrelation function. A Fourier transform in ITU-R P.531 indicates that the peak spectrum is at 0.1–1 Hz, corresponding to a coherence time of 1–10 s of the scintillation intensity stochastic process.

The ionospheric scintillations have large regional and seasonal variations. It is referred to in Figure 6 in ITU-R P.531. In the Arctic, within our coverage area of interest, there is a zone at high latitudes with a fading depth of 2 dB peak-to-peak at 1.5 GHz in years of solar maximum. In solar minimum years, it is far less.

When assessing the link margin that should be used for the scintillation effect, it is worthwhile citing a paragraph in the ITU-R 531 document:

Due to the complex nature of ionospheric physics, system parameters affected by ionospheric effects, as noted above, cannot always be succinctly summarized in simple analytic formulae. Relevant data edited in terms of tables and/or graphs, supplemented with further descriptive or qualifying statements, are for all practical purposes the best way to present the effects.

From Allnutt (2011), we recall the following: “The concept of annual statistics is of dubious merit for ionospheric phenomena.”

The above citations indicate that there is no straightforward way to estimate a link margin for a specific application. Hence, the reader must be aware that the margins presented here and used in the link budgets in the section are subject to uncertainties. It is emphasized that most mechanisms affecting the margin in link budgets are complex stochastic processes. Using the “worst case” of all these mechanisms will lead to margins that are too high and have an unfortunate impact on the cost of a system.

We have used the following margins in the link budgets caused by scintillations: 1 dB at 200 MHz, 1 dB at 300 MHz, and 0.5 dB at 2 GHz.

5.2 Fading and diffraction

5.2.1 Diffraction loss

For all practical antenna heights on a vessel, the D06 Fresnel zone clearance (Rappaport, 2002) to the round Earth of radio waves from a satellite will be at an elevation angle below the specified minimum elevation angle for the MEO orbital system. Accordingly, in our system scenarios, the diffraction loss will not have an impact on the link budget. However, it should be mentioned that the diffraction loss from the round Earth is significant when the satellite starts to be below the horizon (Yang et al., 2019). Another mechanism is atmospheric refraction. This has an impact at very low elevation angles. Hence, we discard this mechanism in our assessment.

5.2.2 Scattering and shadowing

We refer to Yang et al. (2019) and Smith (1967) on rough sea scattering and shadowing phenomena from the sea surface. The reflection from the rough sea surface is represented by an effective reflection coefficient R_rough according to the Kirchhoff theory (Molisch, 2006). R_rough is given by the expression in Equation 9

R r o u g h = R · ⁡ exp − 2 2 π σ h ⁡ sin ⁡ ϕ λ 2 . (9)

The variables in the above equation are defined in Table 6. A varying electromagnetic (EM) field at the antenna, superimposed on this average, will contribute to fading and is part of the fading statistics.

TABLE 6. Overview of symbols and definitions used in the text.

The definitions of symbols used in the text are shown in Table 6.

According to Yang et al. (2019), shadowing is caused by sea waves where the incident ray has an elevation angle that will shadow reflection or scattering points from the sea surface. The reflection from the sea surface will always be reduced by this effect. Smith (1967) proposes a shadowing coefficient given by Equation 10 Here, erfc is the complementary error function, and the expression in Equation 11 defines

S f u n = 1 − 0.5 e r f c cot ⁡ θ i 2 β 0 Λ cot ⁡ θ i + 1 . (10)

Λ cot ⁡ θ i = 1 2 2 π β 0 cot ⁡ θ i exp − cot ⁡ θ i 2 β 0 2 − e r f c cot ⁡ θ i 2 β 0 . (11)

The other variables are defined in Table 6.

The difference in distance Δ d = q − p (see Figure 10) between the reflected radio wave and the direct wave in the flat sea situation, is given by Equation 12:

Δ d = h s sin ϕ 1 − cos 2 ϕ . (12)

In the above expression, h s is the antenna height, and Δ d is needed to calculate the phase difference k · Δ d in Equation 14 between the direct and reflected radio waves from the flat sea. The composite gain G_s that comprises the shadowing and reflection from the rough sea is given by Equation 13:where

G s = 20 · l o g 10 η . (13)

η = 1 + S f u n · R r o u g h · ⁡ exp j k Δ d (14)

When there is no reflection from the sea, the variable η = 1. The variable k is the wave number. Table 7 shows the parameters used in the analysis.

TABLE 7. Parameters used in the analysis.

The results of the analysis are shown in Figure 11.

Figure 11 shows that at the long wavelengths in VHF, specular reflection from the sea at low elevation may occur. There may be deep fades when the sea is calm. At 200 MHz and 300 MHz, in a sea with rms height 0.7 m and slope 0.1 m/m, we see there is a fade of 11 dB when the elevation angle is 7°. One may regard the ship’s antenna movement as a semi-stationary (weather-dependent) stochastic process with a coherence time in the order of seconds. Forward error-correcting coding (FEC) combined with bit interleaving of data in a data frame may not be effective in fades of duration in the order of seconds. For VHF frequencies, it is recommended to have a margin against these mechanisms. For the reflected radio waves from the sea, the vertical polarization is suppressed. Hence, circularly polarized radio waves from the satellite are an advantage.

6 Link budget

Typical up- and downlink budgets for VHF (handheld terminal on land) and S-band (maritime terminal and a terminal with an omnidirectional antenna) are shown in Tables 8, 9. We assume ½ rate coding and quadrature phase shift keying (QPSK) modulation. The distance from Lindesnes to the satellite is chosen as one of the handover points (after about 650 min). The distance is approximately 11,000 km, and the elevation is 35°. Because the link budgets are for a 35° elevation angle and not the worst case, we have not included margins for sea reflections. The VHF link budget is for a land-based terminal. The scintillation loss is set to 1 dB and 0.5 dB, respectively, for VHF and S-band.

TABLE 8. Typical up- and downlink budget for a VHF link to and from Lindesnes and not the edge of coverage. Required E_s/N₀ is set to 3 dB, including a 2 dB implementation margin (Maral and Bousquet, 2009).

TABLE 9. Typical up- and downlink budget for the S-band link with an omni- and a frustum-shaped ship’s antenna user terminal. The antenna gain is according to the concept of a single active panel. The link budgets are to and from Lindesnes and not the edge of coverage. The required E_s/N₀ is set to 3 dB, including a 2 dB implementation margin (Maral and Bousquet, 2009).

The coverage of the VHF link is circumpolar, down to 58° north. The peak gain of the satellite antenna is assumed to be 6 dBi. The coverage for the S-band link at the 18 apogee passages for three satellites optimized for the Nordic countries is shown in Figure 12. In this case, the satellite antenna has a peak gain of 21.2 dBi. The edge of coverage in this case is set to −4.2 dB contour.

Due to the satellite’s movement, the distance from Lindesnes will exceed 11,000 km. The maximum distance is about 14,600 km, giving an additional free space loss of 2.5 dB. As can be seen from Tables 8, 9, both up- and downlinks have a margin above 4 dB with an information bitrate of 9.6 kbit/s at VHF. In the S-band, the bitrates are 128 kbit/s and 1024 kbit/s with omni- and frustum-antennas, respectively. The link margins include variation in distances and interference.

7 Conclusion

In this article, we have studied and analyzed the use of the eMEO orbit using an inclination of 63° for coverage in the Arctic. We have described orbital parameters and simulated the orbits over a period of 24 h. The simulations have been analyzed, and 2D histogram plots show the outcomes of elevation and azimuth angles for three sites in the Arctic. These places are Lindesnes and Kirkenes, Norway, and Nuuk, Greenland. The large spread in elevation and azimuth seen from the 2D histograms clearly shows a need for an antenna system that can “see” in almost all azimuth/elevation directions. The elevation plots show the worst-case elevations of 3°, 5°, and 15° at Lindesnes (Norway), Nuuk (Greenland), and Kirkenes (Norway), respectively. Hence, coverage in southern Norway and Greenland in a worst-case satellite position is poor. The maximum and minimum distances from any satellite to, for example, Lindesnes, are 14,600 km and 8,900 km, respectively, representing a delay difference of 20 msec. We have chosen three frequency bands: VHF 200–300 MHz and 2 GHz (S-band). We discuss and analyze the antennas, propagation, and link budgets for these frequencies.

At VHF, we use omnidirectional antennas because the size of electronically steered antennas is impractical. Some examples are presented. However, at the S-band, the electronically steered antenna is perfect due to the rapid technological advances in recent years. An antenna configuration denoted as a frustum presented by He et al. (2016) has been analyzed, and the performance of this solution in different situations has been given. By exploiting transmission/reception simultaneously from panels in the frustum not in the shadow, it is possible to increase the equivalent antenna gain by 4–6 dB compared to the use of only one panel.

The atmospheric losses due to precipitation and atmosphere are insignificant for the frequencies considered here. The loss due to scintillations can be significant for VHF. They are, in most cases, insignificant for the S-band. At VHF slant angles, scattering is close to specular reflection. Significant fades are likely to occur at slant angles.

We have presented typical link budgets at the three frequencies of 200 MHz, 300 MHz, and 2 GHz. At VHF, a link budget for a handheld user terminal is presented. It is concluded that voice and low-data-rate services to the Nordic region with a handheld terminal used on land are feasible. At S-band, we have presented a link budget for services to a ship as two alternatives. In alternative 1, the link budget for an omnidirectional circularly polarized antenna is presented, showing reasonable margins at an information bitrate of 128 kbit/sec. Using the electronically steered antenna with the frustum geometry having five panels, each 20 cm × 20 cm, the information bitrate is increased to 1024 kbit/sec.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fanpr.2026.1764792/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Ahmad Hoirul Basori, King Abdulaziz University, Saudi Arabia

Reviewed by: Joshua A. Abolarinwa, Namibia University of Science and Technology, Namibia

Ahmad Bazzi, New York University Abu Dhabi, United Arab Emirates

Footnotes

nD, n-dimensional; DPD, discrete probability distribution; EIRP, equivalent isotropic radiated power; E/S, Earth Station; FEC, forward error correction; GEO, geostationary Earth orbit; HEO, high elliptic orbit; HPBW, Half-power beam width; IoT, Internet of Things; LEO, low Earth orbit; LNA, low-noise amplifier; MEO, medium Earth orbit; ML, maximum likelihood; PDF, probability density function; RFIC, radio frequency integrated circuit; RX, receive; TX, transmit; TEC, total electron content; UT, user terminal; VHF, very high frequency.