Assessment of spatial and temporal properties of Ka/Q band earth-space radio channel across Europe using Alphasat Aldo Paraboni payload

2.2.1 Chilbolton and Chilton sites (UK)

RAL Space stations of Chilbolton and Chilton are equipped with identical Ka band and Q band beacon receivers and a meteorological station measuring rain rate, relative humidity, ambient temperature, and wind speed (see Figure 2). The design of RAL receivers is based on the equipment developed and tested in prior campaigns.17, 26 Detailed description of the receivers can be found in Woodroffe et al.27

Each of the Ka and Q band receivers uses a single downconversion to a 70-MHz IF; the signal power is measured with a Ferranti SBR100 70-MHz phase locked tracking receiver with a noise bandwidth of 30 Hz and a tracking range of ±100 kHz. The local oscillators (LOs) are dielectric resonators (DROs) phase locked to an internal crystal reference. For the Q band receivers, the LO frequency is multiplied by 3 to drive the mixer (see Figure 3).

For both sites, the two Ka and Q band receivers (and the antennas) are placed indoor where the temperature is controlled. The antennas view the satellite through radomes of woven PTFE windows (see Figure 4). The PTFE dialectic loss for both frequency bands is negligible.

Each beacon receiver is equipped with a commercial open-loop antenna tracking system (S3 SatCom Satsio Antenna System) using ephemeris TLE data provided by NORAD.28

Rain is measured by two different rain gauges29: drop counting (10-s integration time, with a quantisation of 0.0033 mm of rain accumulation per drop) and tipping bucket (one tip of the bucket is equivalent to 0.2 mm of rain) measuring the number of tips per 10 seconds.30 For this project, the drop counting rain gauge is used predominantly due to its higher sensitivity. Chilbolton site is equipped with additional instruments that could support the estimation of total atmospheric attenuation.31

All the time series from the beacon and meteorological instruments are stored on a daily basis at CEDA- RAL in NetCDF format.

2.2.2 Vigo site (Spain)

The Vigo equipment consists of a dual band, single polarisation Ka and Q band beacon receiver (see Figure 5) and a co-located set of meteorological instruments: a weather station and a 2D optical disdrometer. About 60 km from the station, there is a C-band weather radar whose data are available to the experiment.32

The design of the Ka band RF front-end is based on the outdoor unit described in Machado et al.33 It comprises a 35-cm diameter lens-horn antenna and a low noise block (LNB) unit. The LNB performs the low-noise amplification (G_LNB-Ka = 55 dB and NF_LNB-Ka = 1.3 dB) and the first frequency conversion (IF_1-Ka = 1.451 GHz).

The Q band RF front-end has been specifically developed for the Alphasat experiment and is composed of a 60-cm diameter Q band Cassegrain antenna and a custom LNB.32 The designed LNB consists of three main units: (a) a low noise amplifier (LNA), (b) an image rejection (IR) mixer, and (c) a phase-locked coaxial resonator oscillator (CRO) locked to a common high-stability oven controlled crystal oscillator (OCXO) reference at 10 MHz. The LNB performs the low-noise amplification (G_LNB-Q = 20 dB and NF_LNB-Q = 2.5 dB) and the first frequency conversion (IF_1-Q = 142 MHz).

The receiver's data acquisition (DAQ) and signal processing have already been described in Machado et al.32 Two ETTUS USRPs, configured in MIMO mode for synchronisation purposes, perform a second downconversion, digitise the I and Q components of both channels, decimate the signals to a rate of 200 kS/s, and send these components to a PC through a Gigabit Ethernet link (Figure 5). In the PC, a beacon receiver developed using LabVIEW SW carries out the signal acquisition, the spectral processing, and the data storage. In the default configuration, the software performs a 16384 point fast Fourier transform (FFT) to the 200-kS/s I/Q components, thus the frequency resolution (resolution band-width, RBW) would be 12 Hz, with an amplitude measurement rate of 12 S/s. Therefore, considering a Hanning window increase factor (B = 1.5) and three power-summed bins, the measurement bandwidth (MBW) of the beacon receiver would be 54 Hz. Considering these values and a 10-dB CNR margin to limit the amplitude errors, the dynamic ranges in the measurements are of approximately 26.2 dB (Ka band) and 30 dB (Q band).

Both Ka and Q band antennas are equipped with an open-loop antenna tracking system that consists of a pan-tilt unit and two digital controlled servomotors. The azimuth servomotor is coupled to the pan-tilt unit by means of a gearshift and the elevation motor is coupled by means of a ball screw linear actuator. The designed pointing system has an azimuth resolution of 0.009° and an elevation resolution of about 0.001°. The servomotors are controlled trough an RS-232 connection, using an antenna control procedure developed using LabVIEW SW. This procedure performs the regular pointing correction according to the orbit ephemeris message (OEM) files provided by ASAPE.

The dual band beacon receiver was installed in an indoor cabinet at the roof of the Telecommunications Engineering School with a 0.5-cm polycarbonate microwave transparent window, where the temperature is controlled (see Figure 6). The attenuation through the window is estimated to be less than 0.4 dB.

2.2.3 Aveiro site (Portugal)

The Aveiro site includes the following: a co-located Ka-Sat beacon receiver at 19.68 GHz (see Rocha34), a meteorological station and the Alphasat Q band beacon receiver (see Figure 7).

The design of the Q band beacon receiver minimises the number of RF components and employs an SDR sampling directly the IF signal (see Figure 8). The RF hardware comprises a 62-cm diameter antenna and a low noise converter (LNC). After a low noise amplification (NF < 4 dB), with about 20 dB gain, a first downconversion is made to 2.2 GHz by using an LO derived from a PLO at 9.3 GHz and 4× frequency multiplier. After a first IF amplifier, a second and last downconversion to 142 MHz is performed using an LO at 2.06 GHz. Both LOs use a common low phase noise reference at 100 MHz. An IF amplifier delivers a signal with −52 dBm to an ELAD FDM-S2 software defined radio that performs the sampling at about 122.88 MS/s. After a digital downconversion and decimation, an I/Q data stream at 192 kS/s is delivered to a PC and stored as consecutive 100 MB wav files by the SDR Console software radio. A Matlab software on APC cyclically loads from the SDR the WAV file when ready, performs the spectral-based beacon amplitude estimation at 8 S/s, estimates also the noise spectral density, and finally, stores the estimated data in a processed file detected data and finally deletes the analysed WAV before restarting the cycle. The resulting MBW of this beacon receiver would be 64 Hz.

The antenna pointing system (see Figure 7) is based on a rotation/inclination axis system that is motorised by standard linear actuators fitted with optical encoders, to improve the pointing resolution to about 0.022° in each axis. A USB controller starts and stops the linear actuators according to commands from the control PC. The elevation and azimuth of the satellite are obtained from the OEM file distributed by ASAPE. The elevation and azimuth coordinates are converted to inclination and rotation coordinates and sent to the controller periodically (1 min) at each predetermined period of time. The system was calibrated against the satellite to remove any bias introduced by the mounting alignment inaccuracies and periodically verified, and its operations have been always very reliable.

The Q band receiver is installed in an indoor room with controlled temperature. The attenuation introduced by the window (see Figure 7) is estimated to be 0.4 dB at Q band.

The DAQ system of the Ka-Sat beacon receiver logs simultaneously the beacon and the meteorological channels: temperature, relative humidity, wind speed, and rain rate (both drop counter and tipping bucket rain gauges). The number of drops and tipped buckets are counted each second offering an opportunity to investigate the impact of integration time on the rain rate statistical distribution.30

2.2.4 Athens and Lavrion sites (Greece)

The beacon receivers installed at NTUA Campus (Athens) and Lavrion Technological and Cultural Park (LTCP) consist of an outdoor unit (antennas, tracking systems) and an indoor/enclosed unit (including processing and system control elements) (see Figures 9, 10, and 11).

The outdoor units include (a) an offset 1.2-m dish antenna for the Ka band and (b) a 60-cm shrouded parabolic antenna for the Q band followed by LNBs and passed on to the indoor/enclosed unit. After single downconversion, the signal is brought down to an IF of 1.451 and 1.902 GHz for Ka and Q band, respectively. The downconverted signal is further filtered, amplified, and fed to Ettus USRP B210 units for signal estimation. The signals are digitised by the USRPs and are processed using a GNU Radio–based software application using an FFT-based algorithm. After running FFT on the samples, the bin with the maximum power along with three more bins in either of its sides are summed to calculate the received beacon power (using also a Blackman-Harris window to limit any spectral leakage). The FFT bin size is approx. 10 Hz. In addition, noise power is measured by averaging 1000 FFT bins at an offset of 1100 bins at either side of the beacon lobe's maximum and by multiplying the result by the equivalent beacon bandwidth (ie, 3 + 1 + 3 = 7 bins). Then resulting data are time-stamped and saved. The dynamic range for the Ka band receivers is about 40 dB for NTUA Campus and 37 dB for NTUA LTCP, whilst for the Q band ones is about 37 dB for both sites, at a data sampling rate of 10 Hz. Figure 10 depicts the receivers' simplified block diagram. The resulting MBW of this beacon receiver would be 71.72 Hz.

To get a very accurate time reference for the different receiver elements and sites, the LNB's LOs, the USRP units, and the on-board computer are all synchronised using a highly accurate GPS disciplined oscillator (GPSDO). The GPSDO feeds a 10-MHz reference signal to the LNBs and to the USRP units and a pulse per second (PPS) clock reference to align in time the received samples originating from the USRPs in time; the on-board computer runs network time protocol (NTP) in kernel GPS-PPS mode.

Each terminal is equipped with an antenna tracking system (see Figure 11) working only in elevation (because Athens and Lavrion are on the same longitude as Alphasat satellite position). It employs a heavy-duty linear actuator, installed together with gas springs in a push-pull configuration to dampen movement and to limit effect of wind loading. The absolute position of the antenna is measured in real-time using digital high precision inclinometers. The antenna control system is based on a scheduler that downloads OEM file distributed by ASAPE, runs the necessary calculations for each site, and issues pointing commands to the antenna tracking system. The tracking system's resolution is better than 0.05°.

Sites are equipped with meteorological sensors for temperature, humidity, wind speed and direction, atmospheric pressure, and precipitation; the precipitation is measured using tipping bucket rain gauges collocated to the receivers. The rain gauges deliver an accuracy of 0.2 mm of rain at a sampling rate of 1 Hz; each tip increases a counter recorded together with a time stamp. These data are converted to rain rate time series with 1-minute integration window.30

2.2.5 Ljubljana site (Slovenia)

The beacon receiver system at JSI consists of three major subsystems, (a) the RF front-end, (b) the digital detector, and (c) the pointing system. The complete configuration is schematically depicted in Figure 12.

The RF front-end consists of a 1.2-m offset dish antenna with a custom designed 4-port feed and four LNC channels at two bands (19.7 and 39.4 GHz) and two polarisations (co-polar and cross-polar). The nominal antenna gains are 45.8 and 50.8 dBi, with half power beam widths (HPBWs) of 0.84° and 0.44°, at Ka band and Q band, respectively. A dual band/dual polarisation feed was custom designed for this specific application and the overall antenna cross-polar discrimination (XPD) is 12 and 28 dB, at Ka and Q band, respectively. To avoid losses of high frequency signal transfer, the first and the second downconversion for all four channels are performed at the outdoor unit, right after the feed. At the output, the converters provide a standard IF2 signal in the VHF range (140 MHz) led through coaxial cables to the indoor unit performing signal sampling, power estimation, and data storage.

The digital detector is implemented on a PC in a GNU radio application. It employs four universal software radio peripheral (USRP) N200 platforms with BasicRX daughter-boards, ie, one platform per channel. The same PC also runs an application to control the antenna tracking system. The signal processing at first filters the data samples with a band-pass filter. Then the samples are arranged into a vector (with 65 536 elements) and multiplied with a flat-top window function. The vectors are converted to the frequency domain, where the signal power is estimated and recorded together with the sampling time in daily files. With a sampling rate of 400 kS/s and FFT size of 16 bit, every second, 6.1 samples of estimated power are obtained. Figure 13 shows the flow of the signal in the digital processing. The resulting MBW of this beacon receiver would be 90 Hz.

The antenna tracking system mechanism employs the two-line element (TLE) sets provided by NORAD. In addition, adaptive corrections based on received signal strength for fine-tuning are performed twice per day. The pointing resolution of the system is 0.005°.

A tipping bucket rain gauge (YOUNG 52202) is placed close to the beacon receiver and measures the number of tips per minutes and per seconds.30 It has a collecting area of 200 cm², a measurement resolution of 0.1 mm, and includes a debounce circuit. The acquisition system includes also temperature and humidity sensors.

2.2.6 Prague site (Czech Republic)

Two separate Alphasat receivers for Ka and Q bands channels (see Figure 14) are installed at this site. They are equipped with 1.8-m diameter offset parabolic antennas (Andrew Company, type 183) with electrical heating (gain: 48 and 55 dB at Ka and Q, respectively), efficiency 64%. Antenna 3-dB beam width is 0.62° and 0.32° at Ka and Q band, respectively; the antenna cross-polar discrimination is 35 and 30 dB at Ka and at Q band, respectively. The LNB is placed in the feeder at the antenna and is maintained at a constant temperature of 40°C. The LNBs downconvert the Ka and Q band signals to 1.451 and 1.702 GHz, respectively. The receivers are mounted in an outdoor rack together with power supply and communication interfaces and connected with a monitoring server via an RS485 bus using protocol Modbus.

The IF1 is converted to an IF2 of 465 MHz (for both receivers) sent to an RSSI receiver (HopeRF chip RFM65) equipped with switchable filters (100, 10, and 1 kHz), automatic frequency control (AFC) and automatic gain control (AGC) to optimise SNR performance. The gain variation is monitored and compensated using a calibration reference. Synthesiser oscillators are controlled by an OCXO at 10 MHz. The total gain of the receiver chain is 76.7 and 76.6 dB at Ka and Q band, respectively. An extra detector is used to control the closed-loop antenna tracking system, which is performed only in elevation because of the small difference between the longitude of Prague and Alphasat orbital position. The resolution of the antenna tracking system is 0.05°. A splitter and band pass filter is also used to extract the signal at 39 GHz for an IF radiometer (“total power”) channel operating on 1.7-GHz frequency to measure atmospheric noise emission with a total power radiometer.

The station is equipped with a meteorological station (WS 981 ANEMO company) measuring temperature, relative humidity, air pressure, wind speed, wind direction, and rain rate. The rain rate is measured with a tipping bucket rain gauge (resolution of 0.1 mm per tip) measuring the time of the tips.30 A rain distrometer (JR 2D-Video 3rd generation) is installed at 100 m from the receivers. The site is also equipped with a 12-GHz “Dicke switch” type radiometer.35

2.2.7 Louvain-la-Neuve site (Belgium)

The Ka/Q band Alphasat terminal is equipped with a 1.2-m offset feed antenna (Vertex/General Dynamics) with a rotation over inclination mount (span of ±10° and a dual-band, dual-polarisation feeder (LC Technologies) directly connected to the LNC block converting the Ka and Q band signals to 1.1 and 1.9 GHz, respectively (see block diagram in Figure 15).

The indoor unit contains the second converters and software defined radio receivers (ELAD FDM-S2). The detection bandwidth (MBW) of this beacon receiver is 164 Hz. The frequency reference section contains a GPS disciplined reference oscillator. As illustrated in Figure 16, the antenna positioning system contains three linear actuators controlled by the positioner/tracker controller. The controller converts the OEM files data (provided by ASAPE) into linear movements.

UCL has installed a new OTT Pluvio2 weighting rain gauge in a small meteo park located in the experimental farm, at a distance of 2 km from the receiving station, measuring the number of tips per minute.30 The data of three pluviometers (measuring the number of tips per minute) belonging to the Walloon Region pluviometric network30 and located at the vertices of a triangle centred on the Alphasat station are also received and processed.

2.2.8 Edinburgh site (UK)

This site is equipped with Q band terminal. The block diagram of the front-end is shown in Figure 17. It exploits a series of mixers to downconvert the Q band signal to a 5-MHz IF frequency, which is then sampled by a DAQ card (12-bit National Instruments 5124) followed by digital processing (software defined radio). The LO of the downconverter are driven by an ultra-stable 10-MHz reference oscillator. The receiver can capture the spectrum that includes the beacon signal and implements frequency tracking of the beacon signal. A 10th order type 2 Chebyshev digital bandpass filter of bandwidth 50 kHz and measured noise bandwidth of 10.6 Hz is applied; see Figure 18B. A more detailed description of this receiver can be found in Nessel et al.36

The Q band antenna and RF front-end are shown in Figure 18A, where the temperatures of the LNA and RF box are monitored. The indoor computer performs data processing and logging. A weather station is also located at approximately 70 m southwards from the terminal.

The terminal is equipped with an open-loop antenna tracking system (QPT 200 positioner with stepper motors) using OEM data provided by ASAPE.

3.2.1 RAL method

It is based on the Fourier series expansion of the received beacon signal for both the event identification and the estimation of the clear air signal template (ie, the so-called 0-dB reference). The method was originally developed for OLYMPUS26 and ITALSAT17 campaigns, and it is now fully automated and implemented in Python.

This method is currently used for Chilbolton, Chilton, Athens, Lavrion, and Edinburgh data.

3.2.2 Aveiro method

This is a complete GUI SW processing tool implemented in Matlab.38 In this tool, data are calibrated, then possible data gaps are flagged as invalid data, and finally, time series are plotted to perform visually analysis and classification of all the channels (beacon and meteorological data).

Then clear air signal templates are derived first by plotting the beacon copolar signal levels (Ka and Q band if available) and in a separate plot an ancillary channel (usually rain rate or the noise spectral density at the Q band). Then the software automatically marks attenuation event periods (eg, with concurrent rain) that the user can manually change (eg, add/remove period of events or their start/end times). The attenuation event periods are interpolated, and FFT or polynomial techniques are used to get the clear air template (ie, the 0-dB reference level).

The template is then compared with beacon signal level, and in this case, the template can be fine “tuned” by manual interpolation with second-order polynomial or linear interpolation. The excess attenuation is derived (as the difference between the template and the signal level) and in case accepted or further processed.

This method is used for Aveiro and Vigo data.

3.2.3 Propagation data processing system

In the frame of ESA contract,39 a web-based propagation data processing system (PDPS) has been implemented to receive via network the L1 data from different measurement sites and to generate L3 data in a semiautomatic process. L1 data are coded using the ASAPE SPEX format39 and are automatically processed to remove outliers, time inversions, and nonphysical values. L3 data are then automatically derived according to specific user's configuration and available ancillary data (eg, event detection can be automated thresholds on rain rate, radiometric attenuation, integrated liquid water content, or sky status index). Event detection can be corrected manually offline. Then the 0-dB reference level is derived using the days before and after the one under analysis to extract either FFT harmonics, coefficients of polynomial interpolation, or using low-pass filter. This method is used for Louvain-la-Neuve data.

3.2.4 IAP method

The IAP method is based on FFT/IFFT and polynomial method. The most appropriate data for the 0-dB reference level is selected and combined by visual inspection, considering either clear air periods or parts of signal level during other days (up to 2 days) before and after the one under period of analysis (ie, −1 d = yesterday, −2 d, +1 d or +2 d). To avoid discontinuities between the start and the end of the templates of different periods, the FFT/IFFT is extended to cover overlap between consecutive periods (ie, from −3 h of the last day of the previous period to +3 h of the first day of the next period). This method is used for Prague data.25

3.2.5 JSI method

The procedure starts with an automatic detection and removal of signal outliers. Periods of events are selected by visual inspection. During clear-sky conditions, polynomial functions are fitted to the copolar signals to derive the 0-dB reference level (see previous studies23, 40). This method is used for JSI data.

At the beginning of the activities, an initial analysis of the data processing methods adopted by each experimenter has been done, to verify their applicability to campaign progress monitoring.

Data quality control checks are done first by the experimenter on time series and statistics and then reviewed on a monthly basis with RAL, as responsible for statistical data analysis and ESA. In case of anomalies of failures, corrective actions were discussed within the consortium, selected and executed with the objective to minimise the loss of data.

4.1.1 Data availability

Table 4 lists the monthly data availability, intended as percentage of time of valid processed beacon data (Level 3) for each station of ASALASCA network and each month in the period July 2016 to December 2017. Measurements collected when attenuation exceeded the dynamic range of the receiver are still considered valid for the specific purposes of assessing statistical distributions and for correlation analyses. These data are marked as loss-lock for the calculation of second-order (ie, conditioned) distributions.

Table 4 does not include the availabilities for Edinburgh and Q band signals in Athens and Lavrion, as the processing of these data has only started recently. However, at both sites, the operational availability, intended as amount of raw data (level 1) is almost 100%. This table includes also effects of unavailability of space segment, like the planned switch off of the Alphasat Aldo Paraboni payload between 2 and 7 March 2017 during spacecraft eclipse period.

4.1.2 Power spectral analysis

Power spectral analysis is performed on a daily basis to check the quality of the received signals, according to practice recommended by the ASAPE group. Figure 21 shows two examples of power spectral analysis of dry scintillation and a rain event in England (sampling rate 10 Hz) compared with the theoretical slopes for rain and scintillation (ie, −20 dB/decade and −80/3 dB/decade, respectively).48 Figure 22 shows the power spectral analysis for an intense rain event in Aveiro (sampling rate 8 Hz).

4.1.3 Geographical variability of distributions of excess attenuation

Figure 23 provides an example of the variation of the distribution of excess attenuation at Q band when passing from Southern England (Chilbolton and Chilton) to Greece (Athens and Lavrion at the other edge of the coverage area). The rain rate exceeded for the 0.01% of the average year, as predicted from the ITU-R Rec. 837-7, is around 30 and 64 mm/h for the Southern England and Athens, respectively.

4.1.4 Spatial characteristics of the radio channel

Figure 24 shows examples of excess attenuation at Q band measured simultaneously on the same day at different sites at short (plots a and b, Chilton-Chilbolton, 47.9 km) and long distances (plots c and d, Vigo-Aveiro, 173 km) as an example of effects relevant for application of site diversity and smart gateway FMTs.

Figure 25 shows an example of the application of an ideal system switching between Chilbolton and Vigo at an attenuation threshold of 5 dB, for the same day of Figure 24. The joint signal is defined as the less attenuated signal at Vigo and Chilbolton.

Future statistical analyses will include the characterisation of the correlation coefficient and the statistical dependence index,11 as also required by ITU-R SG3.

4.1.5 Ka/Q band frequency scaling of excess attenuation

Figure 26 shows an example of the variability of the instantaneous frequency scaling of excess attenuation from Ka to Q band measured at Vigo on the same day of previous examples.

Figure 27 gives the distribution Q band excess attenuation conditioned to excess attenuation at Ka band derived from observations in Chilbolton from July 2016 to December 2017. The plot includes also the curve of the equiprobable values of Ka and Q band excess attenuation, which is used in comparison with the 50% decile of the conditioned distribution to detect eventual statistical data biases (see previous works17, 26).

4.1.6 Meteo and remote sensing (ancillary) data

Figure 28 shows an example of measurements of wind speed and direction at Chilbolton in relation to baseline distance and orientation of the Alphasat link with the stations in Chilbolton and Chilton.

4.2.1 Simulated time series of the attenuations

Figure 29 shows a 1-day time series of the different contributions to the Q band attenuation generated by the ANS for the Chilton station. The gaseous attenuation (dashed) is always present and varies slowly around 1.2 dB in agreement to previous ITALSAT observations.17 The cloud attenuation (dashed-dotted line) gives a small contribution mostly during about the second half of the day. Rain attenuation (solid line) contributes mostly with a small peak of about 1 dB at 8 am UTC.

The temporal resolution of ANS data is 5 minutes, which is deemed sufficient to catch the variability of gaseous (slow varying) attenuation making possible to use ANS data in clear air conditions and beacon measurements to identify the 0-dB reference level and to derive total atmospheric attenuation. In addition, the spatial and temporal resolution of the current ANS process could be sufficient to reproduce the dynamic of rain attenuation for percentages of time lower than 0.1% (see previous literature49, 50). It still remains to be investigated the applicability of ANS data for the derivation of cloud and rain attenuation, considering that the max value of attenuation of ANS data is only limited by spatial and temporal resolution whilst VSAT class propagation terminals have a limited dynamic range.

4.2.2 Attenuation statistics derived from the simulated time series

Figure 30 shows statistics for the gaseous attenuation produced by the ANS over a 12-month period compared with the yearly ITU-R prediction. An agreement well within 0.1 dB can be expected based on the observed results. In the absence of independent measurements of clear-sky attenuation, this value is assumed to be indicative of the accuracy of gaseous attenuation estimated by the ANS.