The GBS dataset: measurements of satellite site diversity at 20.7 GHz in the UK

The GBS (Global Broadcast Service) dataset is a series of radio attenuation measurements made at three sites in the UK: Chilbolton and Sparsholt, both in southern UK, and Dundee in Scotland. The aim of the experiment was to make long term measurements of the signal strength received from a 20.7 GHz beacon on the US Department of Defense satellite UFO‐9 at multiple sites, in order to determine whether the use of site diversity as a fade mitigation technique would be effective. The dataset spans a period of 3 years, from August 2003 to August 2006 with signal attenuation sampled once per second.


Introduction
Radio signals at frequencies above 10 GHz suffer significant attenuation from rain, clouds and atmospheric gases. In spite of this there is significant pressure on the radio spectrum to open these (so far) unused higher frequencies to commercial exploitation. To do this, methods of compensating for atmospheric and rain fading must be developed (Castanet et al., 2003;Panagopoulos et al., 2004). One fade mitigation technique (FMT) proposed is the use of site diversity, where multiple satellite receive stations are set up sufficiently far apart so that the rain experienced at each site is de-correlated. In a properly configured site diversity system the sites experience fading at different times, and switching between the sites can improve the total communications system availability considerably.
The primary use of the GBS dataset is to provide the raw data for the development and testing of radio propagation models. However, it can also act as a proxy for rain rate measurements as there is an empirical relationship between the rainfall-induced attenuation and the part-averaged rainfall intensity (Goldshtein et al., 2009).

Experimental set-up
The Radio Communications Research Unit (now the Chilbolton Group), part of STFC-Rutherford Appleton Laboratory, installed the Chilbolton GBS receiver in August 2003, the Sparsholt receiver in October 2003 (beginning the site diversity measurements) and the Dundee receiver was installed in February 2004. All three receivers were recording the signal strength received from the 20.7 GHz Global Broadcast Service (GBS) beacon carried on the US Department of Defense (DoD) satellite UFO-9 at approximately 23°W until contact with the satellite was lost in August 2006.
The UFO-9 satellite was in a slightly inclined (approximately AE3.5 degrees) geosynchronous orbit requiring a tracking system on the ground receiver to ensure that the receiver pointed correctly. Ephemeris information was not available as UFO-9 was a military satellite. Unfortunately, the cost of implementing a tracking system on the receivers was prohibitive, so wider beamwidth antennas were used capture the complete movement of the satellite without the need for tracking. This resulted in a reduction in receiver gain of 6 dB, but the dynamic range of the receivers was wide enough to record a full range of measurement. The relative movement of the satellite and lack of tracking introduced a sinusoidal diurnal variation into the recorded data which was removed during preprocessing. The GBS receivers had a dynamic range of approximately 13 dB, sufficient to measure attenuation at 20.7 GHz down to time percentages of 0.02% (approximately 2 h in a year). Table 1 gives the technical specifications for the GBS receivers. Both receivers were designed and custom-built in-house by members of the RCRU (Figures 1 and 2), and were situated in specially designed cabins, looking along the path to the satellite through radomes of woven PTFE acquired from Gore Associates. More information on the GBS project can be found in (Callaghan et al., 2005) and details of the results of the experiment can be found in (Callaghan et al., 2008).

2003.
• Chilbolton (51°08′N, 01°26′W) Chilbolton and Sparsholt are approximately 7.8 km apart and were chosen to investigate short range site diversity. Dundee is approximately 600 km north of Sparsholt and Chilbolton, allowing an investigation of long range diversity. Co-located with the Chilbolton and Sparsholt receivers were a range of supporting meteorological instruments, such as a drop counting rain gauge and instruments to measure wind speed and direction, as well as pressure and temperature. These meteorological measurements were augmented by the multi-parameter 3 GHz radar, CAMRa, also located at Chilbolton. The signal strength in dB was recorded at the three receive sites at a sampling frequency of 1 Hz. Due to the diurnal variation of the satellite, this data was preprocessed to produce the archived attenuation time series, using the same methods as was used for the ITALSAT data (Ventouras et al., 2006). This process is outlined below:    The upper plot shows the signal levels as recorded by the beacon receivers. The blue dotted lines indicate the variation that the signal would have if the day were completely free of attenuation. These lines are fitted during the data pre-processing to remove the daily sinusoidal variation of the satellite relative to the beacon receiver, as well as adjusting for any sudden change in beacon signal power or receiver sensitivity. (The upper of the dotted lines indicates the attenuation relative to vacuum, whereas the lower dotted line indicates the attenuation relative to clear sky, i.e. attenuation including the gaseous attenuation due to atmospheric water vapour and oxygen, but not due to clouds or rain.) The lower plot in Figure 3 shows the resulting processed attenuation time series for the day, which is the time series stored in the archived attenuation time series.
In general the data availability for this experiment was very good, with monthly availabilities generally greater than 95%. The Sparsholt and Chilbolton datasets are continuous between the start of data collection and the loss of the UFO-9 satellite. This is not the case with Dundee, where funding issued resulted in a 2 month gap when data was lost due to networking problems. Figure 4 shows example monthly cumulative distributions of attenuations for the 3 sites.