Noise and Interference Reduction in Air-Launched Antennas used for GPR Evaluation of Roads and Bridges Jeffrey Feigin, Roger Roberts, Robert Parrillo, John Rudy, Alan Schutz, Jami Thomas Geophysical Survey Systems Incorporated 12 Industrial Way, Salem, NH 03079 603-893-1109 feiginj@geophysical.com Introduction Ground penetrating RADAR (GPR) is a technology used to assess the composition and location of heterogeneous materials. Common applications include locating the precise position of rebar within a concrete wall/floor, identifying and locating buried objects underground, assessing the quality and uniformity of an asphalt or concrete highway surface, and detecting deterioration on bridge decks. It is a particularly attractive technology due to the fact that not only is it non-destructive, but it is also non-ionizing. In fact, GPR utilizes radio energy that is similar in frequency to that of conventional pocket cellular telephone, but at far lower power levels. Common usage of GPR for roadway and highway applications requires resolution of features of less than one inch (2.54 cm) and is most convenient when the measurement system is capable of operating at legal highway speeds. There exist two common types of GPR for road/bridge surface measurement: Ground-coupled and airlaunched. Ground-coupled systems rely upon an antenna that is placed very close to the roadway/surface while airlaunched systems utilize directional antennas aimed at the surface from a height of 12-20 inches (30-50 cm). Ground-coupled antennas have a reputation of being less prone to Radio Frequency Interference (RFI) from sources such as FM and TV broadcasting, but typically operate at very slow speeds that are below normal highway limits. Air-launched antennas, even when travelling at 65 mph (105 km/h), are located at a safe distance from the surface. This paper presents a new interference-rejection technique that allows air-launched antennas to operate in the presence of the most prevalent, strongest, and widespread interference sources that one would encounter in highway and bridge measurement scenarios. Unlike simple interference filtering, this method preserves and even improves the resolution of the measurement data. Actual highway measurement data is collected according to a route that intentionally includes the strongest interference sources in the Boston area in order to demonstrate the effectiveness of this new technology and compare it to the conventional Antennas. Finally, the actual levels of interference ingress is precisely measured and recorded in order to characterize the prevalence and degree of interference caused by the various sources. Background GPR relies upon Ultra Wideband (UWB) RADAR technology, in which very broadband (where the bandwidth or aggregate bandwidth is more than 20% of the carrier frequency) Radio Frequency (RF) electromagnetic waves are used to assess and characterize the medium according to its measured impulse response by means of backscatter. The achievable feature resolution is strictly defined by the absolute bandwidth of the RADAR system. GPR technologies typically utilize antennas that are parameterized by either the center-frequency
or maximum pass-band frequency, while the bandwidth is left as an implicit parameter of that number. In general, achievable RF bandwidth is proportionately related to the frequency. Therefore, higher frequency antennas offer greater imaging resolution, but at the expense of penetration depth. An optimal choice for bridge and roadway analysis is a 2GHz maximum passband frequency, where the measurement bandwidth is approximately 1.5 GHz. The difficulty with such a measurement system, which is only loosely coupled with the measurement surface, is the susceptibility to RFI ingress. Extraneous RF sources, particularly broadcast sources such as FM and TV, enter the RADAR measurement system at the same location the desired backscatter energy is captured. It is of further difficulty that the levels of ingress for broadcast FM sources may be equal or greater than the backscattered RADAR energy while successful high resolution measurements require that such undesired interference be approximately a factor of 1 million times weaker than the intended energy. A possible solution is to attempt to remove the interference by means of an analog hardware filter that removes extraneous FM and TV energy. A filter that exhibits sufficient rejection is realizable in hardware. Unfortunately, all implementable hardware filters (filters that operate on voltages, currents, or mechanical vibrations as opposed to those that operate upon digital representations of these signals) with this level of rejection exhibit variations in time delay at different frequencies and seriously distort and blur the UWB RADAR data. Alternatively, digital filters that operate either in software or in the digital hardware of the RADAR system are capable of performing filtering without introducing any kind of blurring, but cannot remove the residual effects of overloading that occur in the analog circuitry. Therefore, a new interference-rejection technology was developed that operates in analog hardware as well as software. This technique utilizes specially designed hardware filters that remove the influence of the interference but preserve the recoverable information of the RADAR backscatter. A software algorithm, which acts in conjugate to the hardware filter, removes residual interference and focuses the measured impulse response from maximum resolution. Not only has the waveform been stripped of most of the interference sources, but the resultant waveform is extremely sharp and the level of detail is improved to even beyond that of the conventional air-launched antenna. Figure 1 shows the measured response of a metal plate placed 18 inches (46 cm) in front of three types of measurement systems 1) conventional air launched antenna, 2) a conventional antenna with hardware filtering, and 3) and interference rejecting air launched antenna. It is observed that cleanest and sharpest response is produced by the interference rejecting antenna while the hardware-filtered antenna exhibits an unacceptable degree of waveform dispersion.
Figure 1. Comparison of a Metal-Plate Reflection for a Conventional Air-Launched Antenna, A Conventional Antenna with Hardware Filtering, and An Interference-Rejecting Air-Launched Antenna Data Collection and Measurement System A truck was configured for data collection by mounting a conventional air launched antenna (GSSI Model 4105) on the center of the front bumper while the interference rejecting version (GSSI Model 4105NR) was mounted on the back. A measurement system was devised that captures the amount and frequency of RFI ingress into the antenna by connecting an additional antenna to the driver-side front-bumper of the vehicle. The feed point (the place where RF energy is converted to electrical energy) of this third antenna is connected to a spectrum analyzer for interference measurement. Both the RADAR and spectral measurement system data are recorded using laptop computers that also capture GPS positioning information. The RADAR system also provides a measurement of the data quality, which is used to assess the impact of RFI. Figure 2 shows a photograph of the data collection vehicle.
Figure 2. A Photograph of the Data Collection Vehicle A route through Boston and along the major highways was taken. Route 95/128 South of the Massachusetts Turnpike was specifically included since the TV and FM broadcasters that are found in this area are among the strongest found in the United States. The junction of Routes 93 and the Massachusetts Turnpike in Boston coincide with at least 9 extremely powerful commercial broadcast interference sources that have been known to cause difficulty for the conventional antenna measurement system. Figure 3 Shows the relative levels of the various classified interference sources along the route. Figure 4 shows a split-screen example of an outage caused by an FM Broadcast source (top scan) where the interference rejecting antenna is mostly unaffected by noise (bottom scan); both scans were taken at the same time.
Figure 3. Plot of Relative RFI Ingress for Various Radio Services vs. Route. Larger Circles Indicate Stronger Interference. Figure 4. A Screen Capture from GSSI RADAN Showing the RADAR Scans for the Conventional Horn Antenna (top) and the Interference Rejecting Antenna (Bottom) for an Identical Stretch of Highway. Note that the Layer Details are Completely Obscured by Noise (Due to an FM Transmitter) for the Conventional Antenna while the Interference Rejecting Antenna is Mostly Unaffected. Data Analysis
Figure 5. Example of Plotted RADAR Data vs. Distance showing the Effects of RFI The measured RADAR data for each location is categorized according the level of disruption for both antennas. Compared to the metal plate reflection magnitude, a measurement noise level of -44 db or lower is considered excellent while -44 to -38 db is considered acceptable. Levels -38 to -32 db are considered marginal while noise greater than -32 db is unusable. Figure 5. Shows RADAR data where the level of induced noise is Unusable and Marginal (Excellent/Acceptable are not labeled). From the data analysis for this specific route, it was found that the conventional antenna suffered outages (unusable or marginal data) 20% of the time while the results were either acceptable or excellent 80% of the time. Figure 6 shows a pie chart of this result. Figure 6. The Conventional Horn Antenna Data Quality is Excellent or Acceptable 80% of the Time, but Marginal or Unusable 20% of the Time for this Specific Route Around Boston Next, the data outages were classified according to the interference source. It was found that commercial broadcasting was responsible for the overwhelming majority of disruption where 51% was caused by FM radio
stations and 42% by television transmitters. Cellular/ PCS base stations (but not the phones, themselves) accounted for 5% of the disruptions which were observed to be highly localized rather than widespread. Figure 7 shows a pie chart of the various causes of outages. Figure 7. Causes of Disruptive Interference of GPR for the Conventional Antenna for this Specific Route Finally, the data for the interference rejection antenna was considered. It was found that outages were reduced from 20% to less than 8% by using this technique. Further, 79% of the data is now excellent versus the previous 62%. Figure 8 depicts a pie chart for this interference rejection antenna data result.
Figure 8. The Interference Rejecting Antenna Data Quality is Excellent or Acceptable 93% of the Time, but Marginal or Unusable < 8% of the Time for this Specific Route Around Boston Conclusions Air-launched antennas offer a significant advantage over ground-coupled GPR RADAR measurement systems because they can be safely operated at highway speeds. Conventional antennas, however, are subject to undesirable outages due to predominantly broadcast interference sources such as FM and television. While a simple filter could remove these types of interference, it would greatly deteriorate the quality of the data. A new interference rejection technology was developed that combines both hardware and software techniques in order to remove the influence of most broadcast interference sources while retaining and improving the data resolution to a level even beyond that of the conventional antenna. This new interference rejection technique was shown to reduce the amount of outage from 20% to less than 8% under identical scenarios in the presence of extremely powerful broadcasters and other interference sources in the Boston area.