Multipath propagation can always be considered as a case of non-ideal behavior of navigation systems. An undisturbed navigation signal – regardless of the particular navigation system – is associated with only one single propagation path between emitter and receiver. There are several examples of multipath disturbances of navigation systems with strong relevance in flight safety operations. Commonly known are landing course disturbances of the instrument landing system (ILS) due to scattering at large airport buildings or taxiing aircraft near the runway. Avoiding such disturbances by means of so-called ILS protection areas – areas that are kept free of any potential scattering object – has strong impact on an airport’s capacity. Moreover, there have been severe incidences, where multipath propagation led to crucial distortions of the ILS localizer course signal, e.g. .
Another navigation system the performance of which is said to be considerably influenced by multipath propagation is the VHF omni directional radio range (VOR).
In this case scattering objects are single rotating wind turbines or even wind farms. Especially in Germany, in the frame of the so-called “Energiewende”, currently 1,700 MW of wind power investments are blocked because of expected intolerable disturbances of the VOR . However, the authors do not know of any reliable analysis demonstrating a quantitative connection to a VOR disturbance scenario, neither by simulations nor by measurements.
Fundamental difficulties in accurately investigating multipath propagation of navigation systems are at least twofold. On the one hand simulation techniques can hardly handle complex boundary conditions in a largely variable parameter space. Geometries both in the range of possible resonances as well as much larger than the wavelength do occur with mostly a-priori unidentified critical ones which is one key issue for a general validation of numerical tools as described in .
On the other hand, real world measurements are so cost-intensive and logistically demanding that it is impossible to cover all relevant parameters, not to mention finding worst case scenarios with respect to flight safety. In  a demanding measurement campaign is given with large taxiing aircraft disturbing the ILS localizer at airports. Nevertheless the limitations of such a measurement approach are clearly demonstrated, e.g. a missing sensitivity and reproducibility analysis, a limited number of scenarios as well as the lack of any well-directed variation within the parameter space to improve the situation.
The common approach presented here is the scaling of scattering objects where object dimensions are shortened by the same factor by which the measurement frequency is increased. This technique results in a well-controlled environment where reproducible measurement conditions are established that can hardly be realized in a real environment, such as an airport. Scaling itself requires dispersion-free or metal-like objects. Due to lightning protection the aircraft hull can be considered to have a high metal-like conductivity as is the case for most other scattering objects like cars, hangars or also wind turbine blades.
The idea of scaled measurements itself is not new and commonly established, e.g. . Common environments for such scaled measurements are anechoic chambers and compact ranges. But the fundamental restriction to the mentioned current state of scaled measurements is that they are always realized using plane wave excitation. In fact, this is the particular condition that nearly all analysis methods start with when determining the radar cross section (RCS) of objects, both experimentally as well as using simulations. Or vice versa, the RCS of an object is always associated with the particular excitation of a plane wave . Whereas such a single source, plane wave excitation approach is well-suited for a variety of RCS-related situations, it does not contain any particular navigation information. The latter must a-priori be a space-modulated measure and consequently a navigation distortion analysis must measure the distortions of space-modulation properties of the signal’s propagation.
The inadequateness of the plane wave approach can also be seen from the ILS localizer illumination of a large taxiing aircraft at a realistic distance in the range of airport dimensions: the so-called “ILS antenna footprint” gives a 10 dB field strength variation over the whole aircraft’s dimension. Such an illumination does not occur using plane waves .
For the ILS case the initial approach of scaling and including the actual navigation information is presented in . Investigations on scaled wind turbines are presented in ,  but they are performed under the mentioned plane wave restriction that does not resemble realistic scenarios. The actual navigation information that provides a spatially differential signal is not included. These particular properties, i.e. a realistic illumination and the scattering of space modulated signals, are addressed in this paper.
It is organized as follows. In Section 2 the functional principles of the ILS and VOR are recalled very briefly and merged to a general idea on how to scale arbitrary navigation systems with a universal reconfigurable hardware architecture presented in Section 3. In Section 4 the scaled environment of the ILS localizer and taxiing aircraft is described. Measurement results are presented both for taxiing aircraft as well as for a metallic sphere, the latter being also analytically treatable and thus giving an out-of-doubt validation for the first time. This section also comprises necessary considerations of antenna pattern synthesis with slotted waveguides. The next section applies the proposed universal architecture to the VOR case and gives an overview on first results of measured Doppler-shift spectra of a rotating scaled wind turbine. Though this contribution focuses on navigation systems at VHF frequency (ILS localizer and VOR), ongoing considerations for applicability at higher frequencies, e.g. radar or satellite-based navigation, are discussed in the conclusion.
2 General consideration for scaling navigation systems
Developing a universal approach for scaling arbitrary navigation systems requires to focus on their common function principles irrespective of their individual specifications. Though those are quite different for ILS, VOR, primary radar or the microwave landing system (MLS) with respect to frequency, modulation and radiation characteristics, their fundamental property is to provide an electromagnetic signal that varies in space. It is generated by a variety of distinct antenna radiation patterns, each one giving a unique space angle distribution of field strength or field orientations around points or lines in space. Here they are called “antenna radiation states” or simply antenna states S1, S2,..., SN the plurality of which translate into space-dependent variations with respect to amplitude, frequency, phase, polarization, etc. These antenna radiation states are excited simultaneously, one after the other or in more complex timing schemes, but the navigation information is always extracted from the superposition of all of them under specified time constraints, i.e. continuously like in the ILS-case or within one mechanical rotation of a radar’s antenna. However, for analyzing possible disturbances of the navigation signal, the disturbance analysis of each individual antenna radiation state is not restricted under this specified time constraint: each can be analyzed individually, and the total distortion analysis of the whole navigation signal is still given by the superposition of the disturbances of each individual antenna state. The exploitation of this simple principle will facilitate the design of the scaled system and the measurements drastically.
Considering the ILS localizer, the particular navigation information of the horizontal angle to the runway axis is provided by a two-tone amplitude-modulated antenna array, where varying amplitude ratios are the measure for an aircraft’s lateral position angle to the axis. A detailed description of the ILS function principle is given in  and further discussed in Section 4.
The VOR, as a second example, provides varying phases proportional to the angle between the direction into magnetic north and the direction from VOR to the aircraft. Its particular function principle is briefly described in Section 5. As probably the most familiar navigation system the rotating primary radar receives its navigational information from time varying spatial directions, where the navigational information further comprises both amplitude and frequency of reflected signals.
Consequently, a universal approach in scaling navigation systems is to split the space coding of the actual navigation information of the original system into distinct radiation patterns of the scaled antennas which are realized separately and alternating within well-defined time slots.
With corresponding technical effort it might be possible to scale navigation systems with a complete functional modelling, thus keeping the original coding of the navigation information without the before mentioned step of abstraction into individual antenna states. However, this would require much more demanding specifications of high frequency components of the emitter and receiver. On the other hand, once the universal scaling concept is validated for one single original navigation system – in this case the ILS – its application to other, respectively arbitrary navigation systems can already rely on a substantially validated approach. Another important argument against a complete functional scaling is that it simply limits the analysis of the distortion scenarios much more than the proposed use of distinct antenna states. This will be further discussed in Section 5 in the context of VOR and rotating wind turbines.
Summarizing, the fundamental approach of scaling navigation systems and measuring their performance degradation due to multipath propagation is to provide time varying antenna states, respectively radiating signals and distinguish them at arbitrary locations of observation.
3 Hardware architecture and coding of navigation information
Original navigation systems are transferred into a universal hardware architecture at corresponding higher frequency. In , in the context of the ILS, a scaling factor of 1:144 is chosen. This choice is based on the availability of plastic model aircraft, additionally galvanized with copper, as scattering objects. Consequently, the original VHF frequency at about 110 MHz corresponds to a scaled frequency around 16 GHz. In that Ku-band a wide range of HF-components having adequate specifications is commercially available. The emitting part of the hardware architecture radiates constituting parts of a navigation signal into different spatial directions with an unmodulated fixed carrier frequency. The receiving part unambiguously identifies the respective components of the navigation information, but without the necessity of synchronization between emitter and receiver. Not only would this imply additional technical demands, but also would have limitations on the mobility of the receiver. The number of different spatial directions, respectively emitting antennas is variable, depending on which particular real navigation system is to be scaled. For example, the highly directive ILS localizer only needs two such antenna states whereas the omni-directional VOR requires more antenna states – for the presented VOR demonstrator twelve. The actual idea of coding the navigational information is to associate them with characteristic preceding pauses of emitting and/or characteristic emitting durations with respectively only one antenna state active as formulated initially in . A scheme of the proposed architecture is shown in Figure 1. The activation of respective antenna states and the realization of the pause durations are carried out with a switching matrix, consisting of 1 to 4 switches that can be cascaded to increase the number of different antenna states. The characteristic pauses for coding antenna states are realized by switching the carrier signal to a 50 Ohms match, in order to avoid transient oscillations leading to unwanted sidebands.
On the receiving side the signal is down converted to an intermediate frequency (IF) level that can either be measured with a power detector (e.g. Analog Devices AD8362) or an oscilloscope. Whereas a power detector is suitable for measuring amplitude variations, an oscilloscope allows measuring frequency variations in the IF additionally, e.g. due to rotating wind turbines causing Doppler shifts. In this particular case the IF is 200 MHz when using the power detector and about 100 kHz when measuring with an oscilloscope to achieve a reasonable sample rate. The processing of the measurement data, thus the identification of the received signals as their respective antenna states and their superposition and calculation of the final navigation signal and distortion, is done with a PC. A robust algorithm for arbitrary numbers of navigation signal components has already been realized and is published in .
Figure 2 shows an oscilloscope measurement example for a scaled navigation system consisting of two antenna radiation states S1 and S2 as for the ILS localizer. The durations of the navigation signal components are both 10 ms. The characteristic preceding pause corresponding to S1 is 20 ms, for S2 10 ms. Thus, a period of the whole navigation signal takes 50 ms. The signal sequence “long pause – signal – short pause” is interpreted as S1, the sequence “short pause – signal – long pause” corresponds to S2.
In Figure 2 an additional inset is shown as zoomed from S2. Clearly visible are the oscillations at the intermediate frequency. Its variations would indicate probable Doppler shifts due to moving scatterers. This important feature is further discussed in Section 5 in the context of the VOR and rotating wind blades.
Depending on the respective navigation system and the measuring quantity the timing of the switching matrix is adjusted. The presented measurement example is the particular coding for a scaled ILS localizer system, a detailed description of which is given in the following section.
4 Scaled measurements of ILS localizer disturbances
The instrument landing system (ILS) comprises a navigation information for the glide angle (glide slope) and for the horizontal angle to the middle of the runway (localizer). The latter is known to be sensitive to reflections on taxiing aircraft near the runway. The navigation signal of the ILS localizer consists of a carrier (110 MHz, scaled 16 GHz) that is amplitude modulated (AM) with the frequencies 90 Hz and 150 Hz with an antenna array. The resulting space modulation is such that the (AM) sidebands have their maximum at small horizontal angles (1°–2°) to the runway middle, respectively to the left and to the right with equal intensity at 0° to the runway middle which is the information of an ideal landing approach.
In the scaled version this is realized in hardware by a slotted waveguide radiator which allows a flexible pattern synthesis as described in detail in the following section. It is essential that the scaled ILS keeps the same radiation characteristics as the original one. Reflecting taxiing aircraft on ground, such as an Airbus A380, are that large (80 m length) and that close to the ILS localizer transmitter (closer than 5,000 m) that the antenna footprint on the object is quite different from the ideal plane wave illumination and has a strong influence on the scattering characteristics. To the knowledge of the authors this particular antenna footprint feature is one of several problems not taken into account by any simulation tools.
4.1 ILS Antenna pattern synthesis with slotted waveguides
The antenna patterns of the scaled ILS are formed with slotted waveguides as they have proven to allow an accurate and flexible synthesis of the two needed antenna radiation states of the ILS localizer. This idea was presented initially in  in the context of ILS and extended to the idea of reconfigurable antenna arrays with slotted waveguides more generally in . Necessary amplitude tapers on the array of waveguide slots are realized by partially covering radiating slots with blends of different sizes.
The CW 16 GHz is fed into the respective ends of the slotted waveguide a photograph of which can be seen in Figure 4. This frequency is set such that the wavelength in the waveguide is slightly smaller than the spacing of the radiating slotted elements. Consequently, the radiating elements of the slotted waveguide have a constant phase shift to each other adjusting the main lobe of the array slightly off the direction perpendicular to the waveguide. Depending on the feeding RF at the left or right waveguide’s port, the main radiation lobe will be to the left or right, i.e. realizing S1 or S2. Thus the waveguide provides two different antenna states depending on which end is fed.
The antenna pattern synthesis for a given original ILS pattern is basically an inverse Fourier transform . A necessary precondition for its application is that the antenna array elements are decoupled. The larger the coupling of the array elements is, the more inaccurate are the synthesis results. In order to avoid calibration routines for such inverse and ill-conditioned problems of mutual coupling beforehand, the radiating elements of the slotted waveguide are designed to radiate only weakly. Another advantage of using slotted waveguides is that the routing of the antenna array itself is carried out with the fundamental waveguide mode. Thus, no feeding network is required at all.
This results in an excellent synthesis quality as Figure 3 exemplarily shows. The pattern to be synthesized is calculated from real data of a CATIII-ILS at London Heathrow airport. Measurements of the scaled antenna are done in an anechoic chamber at a measurement distance of 1.7 m what is clearly not in the farfield. For measurements the architecture proposed in Figure 1 is used. The 90 Hz and 150 Hz components are normalized to the respective maxima. In fact, a corresponding adjustment of emitted power is also done in the later measurements at an open area test site (OATS). The details of this procedure additionally comprising the antenna alignment are described in , .
4.2 Scattering scenarios with a scaled ILS
Figure 4 shows the measurement environment for ILS-scattering scenarios at the OATS of the national metrology institute of Germany (PTB) in Braunschweig. It provides an area of 60 m × 50 m with a conductive coating of the ground. The landing approach of an aircraft is realized with a wagon on a guiding rail to ensure autonomic movement on an ideal course. The orientation of the rail and the position of the scaled ILS are adjusted and measured with geodetical means.
Depending on the distance to the scaled ILS the height of the receiving unit is automatically adjusted to a vertical glide angle of 2.5° to touchdown. As receiving antenna a horn worked well as it receives the whole scattering scenario and is not influenced by the wagon behind it. Exemplarily, the ILS sensitive areas on the taxiway near the touchdown zone are highlighted in the photograph. The waveguide of the scaled ILS and the scattering aircraft are displayed in a zoomed inset. For the later application an airport layout – in this case Frankfurt Main, Germany’s largest airport – is laid on the ground to ease the positioning of the scattering aircraft. Such a layout can easily be created from satellite photographs provided by Google Earth Professional.
One single landing approach from a scaled distance (60 m) to touchdown (30 m) takes about 1.5 min.
A reference measurement is done without any scattering object. By subtracting such a reference from measurement data of other landing approaches the influence of non-idealities of the measurement environment are calibrated out. Such non-idealities could arise from the boundaries of the OATS.
Figure 5 presents measurement examples that already have been calibrated with such reference. An ideal landing approach corresponds to an equivalent ddm of 0 dB which is the ratio of the 90 Hz and the 150 Hz component, respectively the difference between them in logarithmic scale. The equivalence of these logarithmic values and the actual course indicator current in an aircraft is explained in . Also indicated are the allowed ICAO tolerances for landing course disturbances as they are specified in the categories CAT III (0.67 dB, 5 µA) and CAT I (2 dB, 15 µA). The aircraft B747 and A380 are positioned 2.29 m from the ILS on the runway with a heading of 150° clockwise to the runway (corresponding to position 1a at Frankfurt airport in the mentioned ICAO study ).
Disturbances as shown in Figure 5 can be considered as worst-cases given by typical runway-crossing scenarios. The different influences of a Boeing B747 and an Airbus A380 aircraft are that large only for this scenario. As a comparison a calibrated reference measurement is shown which indicates the minimum sensitivity of the setup which is 0.1 dB, well below allowed ICAO CAT III-tolerances. Such scenarios are crucial not only because of the amplitude of the disturbances but also because the disturbances have a nearly constant offset during the whole landing approach.
Figure 6 presents other measurement results with a scattering aircraft as positioned originally at Toulouse airport as specified in Table 1. Distances are measured from the center of the aircraft where the main landing gear is.
It can be seen that an aircraft at position P4 has negligible influence on the landing course and resembles a reference measurement. Whereas Figures 5 and 6 show measured results for an in-flight landing approach, thus a 3D scattering scenario, Figure 7 presents measured data of a 2D scenario which corresponds to a landed aircraft rolling on the runway to exit over a taxiway. In such a 2D scenario the aircraft’s receiving antenna is at its minimum height of 18 cm above ground. Naturally, such 2D scenarios can only be performed if the scattering object is between the scaled ILS and the receiving antenna where the measurement wagon has no influence.
Figure 7 also gives an idea on the reproducibility of the scaled measurements as it shows two landing approaches at two different days after a complete reassembly and alignment of the measurement setup. Both with respect to maximum amplitude and the characteristic of the disturbances the measurement data are in good agreement. Only the locations of the maximum disturbance levels vary slightly. Such variations depend mainly on the positioning accuracy of the aircraft as discussed further in the next section. However, for an assessment if the disturbances exceed allowed tolerances this is not relevant.
In the scaled ILS environment all scenarios are measured as the 1:1 measurement campaigns at the real airports Frankfurt, Heathrow and Toulouse . It turned out that measurement results in the scaled setup led to the same classification with respect to CAT III and CAT I, but that the scaled set-up yields much more detailed information. This particular comparison is quantitatively described in detail in  but it does not comprise the engineering and coding aspects given in this paper.
However, in order to assess the overall accuracy of the scaled measurement setup and thus validate the total procedure it might be more convincing to measure objects the scattering behavior of which is known. Consequently, additional validation measurements are conducted with a metallic sphere as reference object as described in the following section.
4.3 Validation with a metallic sphere
There are two fundamental reasons why a sphere is quite common as reference test object for scattering measurements. On the one hand the sphere is insensitive for angular misalignments with respect to its axes. For measurements with aircraft at least the heading of the aircraft is likely to suffer from inaccuracies, both in 1:1 real-scale measurements as well as in the scaled measurement setup. Some of such alignment issues and in particular their influence on measurement results are addressed in  in the context of radar cross section measurements at THz frequencies. On the other hand, unlike any other object, the scattering of a sphere can be calculated with an exact analytical expression given in , thus it does not depend on any discretization and neither faces numerical difficulties due to its electrical size. Of course, the conducting ground is an additional boundary condition that is taken into account with image theory keeping the advantage of purely analytical expressions. A detailed description on how to analytically calculate a sphere over ground is beyond the scope of this contribution and will be published separately. However, for the sake of analytical expressions, following approximation and assumptions are made when analytically calculating the scattering of the sphere.
Calculations are done with the original ILS pattern to be synthesized. Thus, a comparison between measurement and calculations comprises both the design quality of the scaled ILS as well as its alignment on the OATS.
The receiving antenna characteristic is not taken into account and assumed to be omni-directional.
The coupling between the sphere and the ground is weak. This is a precondition for applying image theory. Consequently, the higher the sphere is above ground the better the analytical expressions for the scattering of the sphere will be.
The illumination of the sphere is supposed to be plane wave as the analytical expression for the RCS of a sphere is only valid for this case. In contrast to the geometry of large airplanes, the problem of antenna footprint is of negligible influence for the smaller sphere geometry used here.
Figures 9–11 obviously show how sensitive such ILS disturbance scenarios are with respect to positioning accuracy of a scattering object, respectively aircraft. Moreover, placing an aircraft instead of a sphere includes the degree of freedom of the aircraft’s heading. Measurements with aircraft as well as measurements with a sphere indicate that the closer the scattering object is to the ILS the less oscillations are visible in a disturbed ddm. However, validation measurements with a sphere clearly show that the scaled measurement setup and its architecture reproduces the analytical calculations and are in very good agreement with the maximum amplitude of ILS disturbances, which is most important for evaluating disturbances quantitatively and getting confident with the scaled approach.
It should be emphasized that this validation with analytic and exact calculations is fundamentally different from numerical simulation techniques as it is traceable and independent from particular scenarios. The accuracies of numerical simulations are known to be sensitive to scattering geometries. For example, the common physical optics method (often used there) yields reasonable results for broadside incidence on a metallic plate (or tail fin) but becomes inaccurate with a scenario of grazing or end-on incidence .
For the actual application of the proposed scaled ILS setup, that is the optimization of ILS protection areas at real airports, it becomes evident, that always a set of several measurements is necessary to take into account positioning accuracy. In fact, this positioning accuracy of aircraft at real airports is very likely not better than in the scaled setup. However, with nearly unlimited availability of the scaled ILS environment and the short time one scaled landing approach takes, such parameter variations can easily be studied now.
5 VOR and rotating wind turbines
Scattering scenarios with the VOR and wind turbines are even more complex than for the ILS. The scattering objects are not only single wind turbines but maybe whole wind farms the electrical sizes of which are far beyond any capabilities of simulation tools nowadays. Moreover, simulation tools like the method of moments or raytracers are so-called frequency-domain-solvers and therefore always consider the stationary case of a scattering object. In particular, they do not include frequency shifts due to moving objects, a restriction which is crucial for numerical treatment of disturbances of the Doppler-VOR as its navigation signal itself is based on frequency modulation. The fundamental problem with 1:1 real-scale measurements is further that measurement results can hardly be related to the actual state of a wind farm with respect to wind direction, rotational speed, and synchronicity of wind turbines. These parameters can not artificially be adjusted for a corresponding parameter study, thus neither for identifying of worst case scenarios.
The function principle of the VOR is only briefly discussed in the following, for details see . The Doppler-VOR is composed of two signals. The first is radiated by an amplitude modulated center antenna and can be considered as a reference signal. A second signal is generated by switching antennas arranged circularly around the center antenna to resemble an antenna moving around the center antenna with a high speed that cannot be mechanically realized. The rotational speed is 30 Hz on a radius of about 6 m leading to a relative velocity of about 1100 m/s. Due to the rotating motion of the antenna an observer receives different sequences of a Doppler-shift, respectively phases with respect to the reference antenna depending on its angle to north direction. Figure 12 gives a sketch of the described function principle. The time diagram shows the instantaneous sinusoidal Doppler-Shift within one period T of the rotating antenna. Depending on the receiver’s angle to north these Doppler-functions are shifted in time with respect to a starting time t0 that is provided by the modulation of the reference antenna. A corresponding phase difference is the measure for the aircraft’s angle to north direction.
Though the integrity of the VOR course signal is known to suffer from static multipath propagation as can analytically be shown for very simple cases , , it is obvious that it is even more sensitive to time variant multipath propagation including a frequency modulation by moving boundaries.
5.1 Realization of the scaled VOR
As each antenna state of the original VOR corresponds to one position of the rotating antenna, the scaled VOR simply consists of discrete antennas. Since the identification, which antenna or spatial direction is currently active, is coded as described in Section 3, the scaled VOR does not need a reference center antenna.
The scaled VOR comprises twelve horizontally polarized patch antennas the radiation sequences of which are controlled with four cascaded HF-switches. Figure 13 shows the antennas on the circumference of a circle, but the antennas can be reconfigurably arranged, for example within only an angle sector to increase the angular resolution or even solid angles. Besides the antenna frontend the most important part of the scaled VOR is the timing sequence of corresponding antenna states. Unlike the instrument landing system where only amplitudes are of interest, the measurement quantities for the VOR are both amplitudes and frequencies. It is important to know the intensity of probable multipath propagation and the associated Doppler shift depending on the motion speed of scattering objects or their interactions. It is essential that these measurement quantities require different timing sequences. Whereas fast measurements of varying amplitudes should be done with short transmitting durations and pauses of 10 ms or even shorter (cp. Figure 2), measurements of frequency shifts with high resolution require long durations of transmitted signals. Both timing sequences of the rotating antenna states are easily realized by simply reprogramming the switching matrices.
Figure 14 shows a measurement example with all twelve antennas sequentially activated with an individual timing scheme. As a receiver the broadband power detector is used, thus no IF-oscillations are visible in the individual signal blocks. The patch antenna representing navigational signal S10 is directed towards the receiving antenna.
The unique identification characteristic, which navigation signal component is active, is provided by different signal durations that are respectively increasing by 10 ms starting with a duration of 10 ms for S1. Pause durations are kept constant with 10 ms. Even if a signal component is below the noise threshold the decoding algorithm is able to uniquely identify the other signals. Clearly obvious is the influence of the individual patch antenna orientations arranged on a circle in 30° sectors. The relative amplitudes of the signal components decrease the more antennas are oriented away from the direction to the receiver. The durations of individual signal components can be chosen much smaller than presented here, because the switches allow switching sequences smaller than 100 ns. An actual restriction is the baud rate from the broadband power detector to the recording PC. But the PC could easily be replaced with an oscilloscope which allows a higher recording sampling rate. In the following, measurement examples are shown for a generic wind turbine in motion.
5.2 Measurement examples of rotating wind turbines
Figure 15 shows a photograph and a sketch of the measurement geometry with a rotating wind turbine. This particular wind turbine can be considered as only a generic one which resembles real ones – manufacturing real scaled models is a present task. However, its wing span of 0.44 m corresponds to nowadays very large wind turbines with a blade length of about 60 m. Like the aircraft models the surface of the scaled wind turbine is made conductive with copper foil and a conductive silver varnish. With this generic wind turbine a reference scattering object in motion is provided that demonstrates the function principle of the proposed channel sounder architecture to measure both amplitude and frequency modulation in a multipath propagation environment.
The distance between receiver and scaled VOR is 12 m, the distance between scaled VOR and scaled wind turbine 6 m.
For assessing disturbances of the actual navigation information it is mandatory to evaluate the ratio of the scattered path and the intentional, respectively ideal propagation path.
Clearly visible are eight periodic maximum amplitude peaks in a time interval of 20 s. Peaks occur if wing blades are in horizontal orientation corresponding to the horizontal polarization of the VOR antennas. Due to three blades such maxima occur three times the rotation speed what obviously is in good agreement with the measured periodicity shown in Figure 16. Figure 17 shows a measurement setup with a rotation speed of 33 per minute.
Amplitude variations are not clearly visible for the much larger rotation speed as the channel is obviously under sampled with the current timing scheme. This is a good example that the characterization of a time variant propagation channel requires an adequate timing scheme, which can be flexibly adjusted with the proposed channel sounder architecture.
In order to fully characterize the multipath propagation channel, especially in the context of the VOR, not only amplitude variations but also frequency shifts need to be measured. For a high frequency resolution with respect to Doppler shifts longer emitting durations for signal components S1 and S2 are required. Figure 18 shows the timing sequence suitable for measuring such Doppler shifts. In the inset-graph the block wise Fourier transform is highlighted where Doppler shifts can be identified.
Signal durations for S1 and S2 are 470 ms, the individual characteristic preceding pause durations are 20 ms and 40 ms, thus a whole period for both signal components takes 1 s.
As a comparison in all Doppler spectra the maximal Doppler shifts in the IF-spectrum are marked which are calculated by: (1)where fLO is the local oscillator frequency of the receiver (cp. Figure 1), fcarrier the emitting frequency, rmax is the length of a blade of 44 cm and frot is the rotation frequency, and c the speed of light. It is anticipated that (1) only applies to one single emitter and one single moving object, thus it can not generally be applied to a bistatic scattering scenario where a scattering object always has two relative motion speeds with respect to the receiver’s and the emitter’s position. Additionally, (1) assumes that the relative motion of the scattering object is towards, respectively away from the emitter or receiver only along the x or z axis (cp: Figure 15). So these theoretical Doppler limits shall just give an approximation of what can be expected.
Figures 19–21 show measurement results with the timing scheme proposed in Figure 18 for a wind turbine with a blade orientation of 90° and three different rotating speeds. For a better readability the carriers at the intermediate frequency are normalized, respectively. For a blade orientation of 90° the motion of the wings is transversal to the scaled VOR, thus besides the relativistic Doppler shifts (far from being measureable) there is none. Consequently, this orientation only has a motion relative to the receiver that leads to a Doppler shift. The Figures 19–21 demonstrate the ability of measuring spatially differential as the signal component that is pointing directly towards the receiver does not contain any Doppler shifts in its spectrum.
A Doppler shift spectrum can be seen in Figures 20 and 21 within the boundaries of calculated Doppler limits. What is remarkable to mention is that the Doppler spectrum is not symmetric with respect to the static intermediate frequency. This could intuitively be expected as the rotating blades always have the same velocities towards and away from the emitter.
In fact, this asymmetry is caused by the asymmetry with respect to ground. Frequencies lower than the static IF correspond to the blades below the hub of the wind turbine (for y<0, see Figure 15) as they move away from the receiver. Frequencies higher than the static IF occur if blades move towards the receiver (for y>0. Clearly visible are the standing-wave-like side spectra due to the conducting ground that obviously vary along the length of the blades; as expected the amplitude increases with the height above ground. Even more interesting observations with respect to Doppler shifts in a bistatic scattering scenario are discussed in the following.
5.3 A deeper discussion of bistatic Doppler-shifts
Whereas the measurement examples in the preceding section only have one relevant relative velocity of the scattering object, some results are given in the following where the motion of the scatterer has different relative velocities with respect to receiver and emitter location. For such cases the limitations and assumptions given in (1) become obvious and are proven by measurements. An obvious constellation for that is the blade orientation of 0° in the same geometry presented in Figure 15. In this case the relative motion of the upper blade (the blade currently above the hub) is both away from the receiver and away from the scaled VOR. For the monostatic case where emitter and receiver are collocated this constellation would lead to a maximum Doppler shift twice as large as calculated with (1). Figure 22 shows the measurement results for this blade orientation.
The velocity of the rotating blade has a component both relative to the emitter and the receiver that respectively moves away from them. This is why the measured Doppler spectrum is larger than the calculated limit that only refers to one single relative motion. An even more interesting example is measured with a blade orientation of 45° the results of which are shown in Figure 23.
For the blade orientation of 45° there are no Doppler shift spectra visible at all. This is because the velocity components of the blade that move away from the emitter, leading to induced currents with a negative Doppler-shift, move towards the receiver causing a positive Doppler-shift. Figure 24 illustrates the components of the blade velocities relative to emitter and receiver. The subscripts “emit” and “rec” refer to velocity components v with respect to the emitter and the receiver location, “long” denotes a motion longitudinal to the observers (directions of motion relevant for Doppler shifts), and “trans” means a motion perpendicular to the longitudinal component (not relevant for Doppler shifts).
The most obvious case for such an elimination of Doppler shifts is if blades are oriented in-line between the receiver and the emitter.
5.4 Conclusive remarks to the scaled VOR disturbances
Within the scope of VOR disturbances measurement examples both for amplitude and frequency modulation of a time variant multipath propagation channel are presented. Whereas for the ILS case presented in Section 4 the performance degradation can already be quantitatively assessed with respect to allowed ICAO tolerances, this step still needs to be done for the scaled VOR in future work. But the proposed channel sounder architecture provides all means to measure both amplitude and frequency shifts for different states of the rotating antenna of a VOR, thus spatially differential. Though the presented measurement examples only refer to a single rotating wind turbine the whole parameter space of even wind farms can and will be analyzed in future work.
Though it is quite obvious and proven by mentioned measurement examples that Doppler shifts of rotating wind turbines strongly depend on the blade orientation (wind direction) and the aircraft’s position relative to the VOR and the wind turbine, corresponding considerations are not part of the ICAO recommendations . Therein actually only the distance between VOR and wind turbine is the measure for potential interferences. Further measurement results within the proposed scaled environment will give reliable and traceable data and will allow a refinement of existing recommendations towards a more physical as well as validated concept.
This contribution proposes a universal hardware architecture and a coding concept for scaled navigation systems in order to investigate the influence of multipath propagation on their integrity. As a first example landing course disturbances of the ILS localizer due to large taxiing aircraft are presented and fully validated by analytical means. As a second example the compatibility of the VOR and rotating wind turbines is discussed with generic examples that qualitatively verify the feasibility of the scaled approach. With respect to flight safety as well as economy mentioned scaled environments provide a powerful optimization tool that allows investigations in a very flexible way with moderate costs for the setup.
The concept of scaling navigation systems and their environments enables in the first place measurements to be performed with an accuracy which can precisely be specified. Measurement examples and analytically treatable scenarios already demonstrate that.
In particular, with respect to states of wind farms and the large distances between a navigation system and a receiver it is not possible that all boundary conditions are monitored and taken into account in a real 1:1 measurement environment. The proposed scaling concept incorporates that easily.
Especially since flight safety lies in the responsibility of public administration, a fully traceable approach for investigating a navigation system’s integrity is desirable. In this context fully traceable means that all individual components of the measurement system, such as antenna gains or HF-components can be measured and characterized separately with high accuracy essentially back to basic SI quantities.
A further benefit of the segregation into antenna states is that possible disturbances of a navigation system can be related to individual spatial directions, into which the navigation system radiates its signal components. Such a differentiation is hardly feasible with measurements where the coding of the navigation information only allows an integral assessment of the disturbances, irrespective whether they are real or scaled.
Though demonstrated examples originally operate in the VHF frequency range and are scaled to 16 GHz, the concept is expected to operate at even much higher frequencies. It should be mentioned that the Friis transmission equation with a scatterer is independent of such a scaling as it contains both the ratio of wavelength and distance from emitter to object and the ratio of radar cross section and distance from object to receiver . Consequently, dynamic limitations for setups at higher frequencies, e.g. radar or GPS, only arise from limited output power and receiver sensitivity. So it could be possible to operate even at 77 GHz in the range of automotive radar, where a variety of HF components is available. Such applications are a focus for future research.
The authors want to thank the institute for geodesy and photogrammetry, university of Braunschweig, namely Mr. W. Schellin, for his assistance in setting up the geometry for the scaled ILS measurements. Special thanks also to the German metrology institute, PTB in Braunschweig, for providing the open area test site for all measurements. Also thanks to NATS/UK for providing data for Heathrow ILS.
 R. Geise, B. Neubauer, and G. Zimmer, “Electromagnetic scattering problems – numerical issues and new experimental approaches of validation”, in AIP Conf. Proc. 1648, 560002, 2015. Available: http://dx.doi.org/10.1063/1.4912781. International Conference of Numerical Analysis and Applied Mathematics, Rhodes, Greece, Sept. 2014, pp. 1–4. Crossref
 International Civil Aviation Organization, Tech. Report (2011, Nov.). “Assessment of ILS protection areas impact on large aircraft operations”. Available: http://www.paris.icao.int/documents_open_meetings/show_file.php?id=161, looked up Dec. 2013.
 T. M. Macnamara, Introduction to Antenna Placement & Installation, ISBN 9780470019818. Chichester, UK: John Wiley & Sons, Ltd, 2010, pp. 273–291. Google Scholar
 I. E. E. E. Standard, Dictionary of Electrical and Electronic Terms, ANSI/IEEE Std. 100–1984, 3rd ed. New York: IEEE Press, 1984. Google Scholar
 R. Geise, A. Enders, H. Vahle, and H. Spieker, “Scaled measurements of instrument-landing-system disturbances due to large taxiing aircraft,” IEEE Trans. EMC Special Issue on Aerosp., EMC, vol. 50, no. 3, pp. 485–490, Aug. 2008. Google Scholar
 Y. Zhang, A. Huston, R. D. Palmer, R. Albertson, F. Kong, and S. Wang, “Using scaled models for wind turbine EM scattering characterization: techniques and experiments,” IEEE Trans. Instrum. Meas., vol. 60, no. 4, pp. 1298–1306, 2011. Google Scholar
 F. Kong, Y. Zhang, and R. D. Palmer, “Wind turbine radar interference studies by polarimetric measurements of a scaled model,” IEEE Trans. Aerosp. Electron. Sys., vol. 49, no. 3, pp. 1589–1600, 2013. Google Scholar
 R. Geise, J. Schüür, L. Thiele, K. Notté, T. Beckers, and A. Enders, “A slotted waveguide setup as scaled instrument-landing-system for measuring scattering of an A380 and large objects,” presented at the the Forth European Conference on Antennas and Propagation (EuCap2010), Barcelona, Spain, pp. 1–5, 12–16 April 2010.
 C. A. Balanis, Antenna Theory, Analysis and Design, ISBN 0-471-59268-4. New York, US: John Wiley & Sons, Inc, 1997, pp. 349–352. Google Scholar
 R. Geise, G. Zimmer, B. Neubauer, and A. Enders, “Investigating multipath propagation for navigation systems in a miniaturized airport environment – ILS and extension to VOR”, presented at the International Flight Inspection Symposium, IFIS 2014, Oklahoma City, Oklahoma, US, June 2014, pp. 355–364.
 C. Jansen, N. Krumbholz, R. Geise, T. Probst, O. Peters, A. Enders, and M. Koch, “Alignment and illumination issues in scaled THz RCS measurements,” presented at the 34th International Conference on Infrared, Millimeter, and Terahertz Waves, IRMMW-THz 2009, Sept. 2009, pp. 1–2.
 E. F. Knott, J. F. Shaffer, and M. T. Tully, Radar Cross Section, ISBN 1-891121-25-1, 2nd ed. Raleigh, NC: SciTech Publishing, 2004, pp. 119–120. Google Scholar
 W. Mansfeld, Funkortungs- Und Funknavigationsanlagen ISBN 3-7785-2202-7. Heidelberg: Hüthig Verlag, 1994, pp. 180–185. Google Scholar
 S. Anderson and R. Flint, “The CAA Doppler omnirange,” Proc. IRE, vol. 47, no. 5, pp. 808–821, 1959. Google Scholar
 B. Neubauer, G. Zimmer, and R. Geise, “VHF omnidirectional radio range and rotating wind turbines – a review on a timely topic”, presented at the 8th European Conference on Antennas and Propagation, EuCAP 2014, The Hague, Netherlands, April 2014, pp. 1–5.
 International Civil Aviation Organization, Tech. Report, “European guidance material on managing building restricted areas” (EUR DOC 015), 2nd ed. Available: http://www.baf.bund.de/SharedDocs/Downloads/DE/ICAO_Docs/EUR_Doc015.pdf, looked up Dec. 2013.
 E. F. Knott, J. F. Shaffer, and M. T. Tully, Radar Cross Section, ISBN 1-891121-25-1, 2nd ed. Raleigh, NC: SciTech Publishing, 2004, pp. 45. Google Scholar
About the article
Published Online: 2015-09-04
Published in Print: 2015-11-01
Funding: This work was supported by the German Ministry of Education and Research under Grant 03V0253.
Citation Information: Frequenz, Volume 69, Issue 11-12, Pages 527–542, ISSN (Online) 2191-6349, ISSN (Print) 0016-1136, DOI: https://doi.org/10.1515/freq-2015-0080.
©2015 Robert Geise, published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. BY-NC-ND 3.0