High-Sensitivity GNSS Limitations in RF Perturbed Environments
چکیده
Low signal power make stable GNSS signal acquisition and tracking difficult under RF perturbed conditions such as in signal shaded environments and in the indoors. These difficulties result in higher measurement noise, multipath (signal reflection) and, in some cases, signal denials and associated poor geometry. Starting in the mid 90s, methods have been developed and tested to somewhat overcome these difficulties. A review of these, which consist mainly of longer signal integration, assisted GNSS (AGNSS) techniques, and advanced signal processing methods, is presented. Associated limitations such as, thermal noise, receiver oscillator limitations, signal reflections and RF interference are discussed. Positioning capabilities and limitations are illustrated with examples in various outdoor and indoor environments as a function of signal attenuation and shading. These include natural and urban canyons. New developments to further enhance performance such as aiding with other sensors are introduced and some examples are given. High-Sensitivity GNSS Limitations in RF Perturbed Environments 2 2 STO-EN-SET-197 1.0 INTRODUCTION The focus of this lecture is on pedestrian users, a challenging user group given its rapid outdoor/indoor transitions, indoor use and unpredictable trajectories. In addition, this user group generally moves in concealed terrain or under the forestry canopy in the outdoors and along buildings and recesses in outdoor urban environments when not completely indoors. In these environments, GNSS is at least partly denied and, if available, suffers from signal degradation due to higher noise, multipath (signal reflection) and poor geometry. Aiding or replacement is therefore required. Even under outdoor line-of-sight (LOS) conditions, GNSS signals may be affected by unintended or intended electronic interference, the latter becoming an increasing concern for military applications, even raising doubts as to the very use of GNSS for many military uses. User requirements are not well defined for many applications, especially civilian ones in view of the rapid evolution of the technology and mismatch between ideal requirements and what technology can deliver. As performance increases, new applications become possible. In many cases, no technology is yet able to deliver the performance required. This is particularly true for soldiers and first responders such as rescue teams operating inside buildings where accuracies of 1 to 2 m are desirable. Such requirements help drive innovation and enhance technology. Prior to discussing aiding of and alternatives to GNSS, selected signal characteristics and interference sources are reviewed. 2.0 SIGNALS AND INTERFERENCE SOURCES GPS satellites are orbiting 20,000 km above the earth. Satellite transmitting antenna beams have to be wide enough to cover the planet as shown in Figure 1. The signal energy is dissipated and only a fraction thereof reaches any one point on the earth surface. As a consequence, the resulting free-space loss of 184 dB results in very weak signals available at a receiver on earth. The limited beam of the transmitting antenna and other gains nevertheless reduce this loss and the final signal strength on the earth for the GPS L1 C/A code is 160 dB (recently increased by 1.5 dB for new satellites). The nominal receiver SNR for the L1 C/A code under LOS conditions is -19 dB for a noise density of -204 dBW/Hz in a 2 MHz bandwidth. The typical detection threshold is 14 dB, hence the nominal processing gain required is 33 dB. When LOS is not achieved, additional attenuation occurs. Foliage attenuation depends on tree leaf type and moisture, the higher the moisture, the greater the attenuation. Values of up to 20 dB are common, which means that a processing gain of up to 53 dB for detection is required. The same order of magnitude holds for canyons, natural or urban. In the indoors, the attenuation depends much on the surrounding material and values of up to tens of dB occur, which means that signals cannot be acquired or tracked over certain thresholds, in which case GPS is partly or totally denied. The normal integration time of LOS signals is 1-5 ms. An increase of the integration time of IF in-phase and quadri-phase baseband measurements improves the SNR. This is the basis of high sensitivity GNSS (HSGNSS). Assisted GNSS (AGNSS) is also used during the signal acquisition phase in the form of information on the navigation messages, Doppler shift and timing from a nearby LOS stationary receiver. This information requires a communication link and is now available on most civilian cellular networks. The effective integration time is limited by many factors, the major ones being oscillator noise and user motion induced Doppler shifts; the latter effects can be dealt with using High-Sensitivity GNSS Limitations in RF Perturbed Environments STO-EN-SET-197 2 3 an IMU. Low cost oscillator noise remains a limitation. In practice, integration of up to a few hundred milliseconds is used. The gain achieved with HSGNSS as compared to standard GNSS is currently of the order of 25 to 30 dB. However this comes at a cost of higher noise, namely up to about 20 m for pseudorange measurements and high multipath in natural and urban canyons and the indoors. Non-LOS multipath in not bounded; in practice multipath errors of up to 100 m occur. Some signals are simply not available due to high attenuation as a result of masking from buildings and topography, which results in geometry degradation. The dilution of precision (DOP) numerically increases and, since position accuracy is the DOP multiplied by the error (sum of all errors), position accuracy drops. Figure 1: Satellite signal beam and energy dissipation Signal interference comes in many forms as illustrated in Figure 2. Ionospheric scintillation was initially a problem, especially for carrier phase tracking on L2; however, thanks to improvements in phase lock loop design, this issue is now minor, even under harsh conditions. Doppler shifts induced by antenna motion is a problem for high sensitivity technology as it limits the signal integration time; this can be mitigated with the use of IMUs (inertial measuring units) to measure short term Doppler effects, such as those induced by a hand held receiver. Noise is low for LOS (line of sight) applications; when the signal strength decreases, code and noise increases accordingly, up to 20 to 25 m; mitigation using IMUs is an option when available. Signal masking results in obscuration of signals on some satellites, thereby weakening satellite geometry (increase of DOP numerical value); reflected signals result in unbounded multipath that degrades position accuracy although, in many cases, it increases availability, although with a diminished accuracy. Unintentional jamming occurring through the increasing use of low cost wireless electronic equipment is a concern and the regulated spectrum is increasingly crowded as shown in Figure 3; weak signal acquisition and tracking in the indoors is vulnerable to some frequency leakages from nearby devices. Intentional jamming is an increasing concern for civilian and military use of GNSS and counter-measures are the subject of intense research. Low cost hand-held jammers, sometime called privacy devices are now legally or illegally available on the market (see Figure 3) and are a major concern for scores of civilian applications. Spoofing aims to coerce receivers into generating false positions without detection, a major concern for military operations with unaided GPS. Portable spoofers are likely to become available and affect civilian applications. An example of the cause and consequence of a spoofer is shown in Figure 4; the spoofed signal is stronger that the real signal, resulting in false positions and misleading of the vehicle on an incorrect trajectory. Spoofing methods are becoming increasing sophisticated and counter-measures are evolving equally rapidly. High-Sensitivity GNSS Limitations in RF Perturbed Environments 2 4 STO-EN-SET-197 Figure 2: Interference sources Figure 3: Non-intentional and internal electronic interference Figure 4: GNSS spoofing cause and consequence on a vehicle trajectory High-Sensitivity GNSS Limitations in RF Perturbed Environments STO-EN-SET-197 2 5 3.0 HSGNSS PERFORMANCE EXAMPLES Examples are given herein using hand-held HSGPS receivers aided with barometers, which help not only the height component, but the horizontal components as well through the additional redundancy. The first example, shown in Figure 5, consists of vehicular (blue) and pedestrian (red) trajectories determined with three different high sensitivity units simultaneously in selected streets of downtown Calgary where nearby buildings reach 60 stories, creating signal attenuation, multipath and poor geometry. The route was driven three times consecutively with the vehicle and walked once. The two trajectories on the upper part of the figure were obtained with a GPS unit equipped with a helix antenna; the remaining four trajectories were obtained with GPS-GLONASS units equipped with microstrip antennas. The vehicle was driven at velocities up to 50 km/h while the pedestrian was walking on sidewalks, much nearer to buildings. The vehicle trajectories are better than the pedestrian ones due to the above and, possibly, to more effective position filtering due to the higher velocity of the vehicle as compared to that of the pedestrian. Figure 5: HSGNSS trajectories in urban canyons (Vehicle on left, pedestrian on right) High-Sensitivity GNSS Limitations in RF Perturbed Environments 2 6 STO-EN-SET-197 The vehicle trajectories deviate from the streets by up to 50 m but are generally within a few tens of metres, which is remarkable given that receivers would have locked on reflected signals in many cases. Commercial vehicle navigation systems use the integration of GNSS with maps to enhance performance and generally perform well in most city cores; additional aiding with wheel, steering and inertial sensors available in modern vehicles is also possible (e.g. Li 2012). For off-road vehicles, wheel slippage and uneven terrain limit the effectiveness of some of these sensors. The pedestrian trajectories are noisier due to the reasons given earlier and are off by up to half a block or about 50 m in the south-north direction. In urban canyons, the satellite geometry at the time of the test matters and the above results are not necessarily reproducible at any time of the day. The next example, shown in Figure 6, consists of a pedestrian test with several hand held HSGNSS receivers, including a GPS-GLONASS unit in a natural canyon with mask angles of up to 80 ̊. The narrow canyon runs east-west, is 1.5 km long and was travelled forth and back on exactly the same route. The C/No values shown in Figure 7 reveal attenation of up to 25 dB from the maximum of 50 dB-Hz observed; LOS signals are limited and attenuation is caused by multipath. Some satellites were not available as the attenuation was too high. This is a classic case of poor DOP, high noise and high multipath. Differences between units and between forth and back trajectories reach 100 m. Comparisons with a reference trajectory known to an accuracy of 1 m results in the same error level. The other segments running roughly south-north are outside the canyon and are more accurate because due to better LOS conditions. The GPSGLONASS unit did not perform better than the GPS ones because the high mask angle and high multipath and noise defeated the advantage of additional satellites in this case. Reliability is also an issue which will be discussed later. Given the exceptionnaly harsh conditions encountered in this environment, near continuous availability of unaided GPS in hand held units with an accuracy of 100 is impressive and useful for many location/navigation needs but clearly fail to meet sub 10 m requirements. Figure 6: HSGNSS in natural canyon High-Sensitivity GNSS Limitations in RF Perturbed Environments STO-EN-SET-197 2 7 Figure 7: C/No in natural canyon 4.0 NAVIGATION IN PARTLY AND TOTALLY DENIED GNSS ENVIRONMENTS Technologies to aid or replace GNSS in these environments are evolving rapidly and a full review is beyond the scope of this lecture. Excellent reviews are provided by Mautz (2012) and Deak et al (2012). Two types of technology that can be used over wide areas are reviewed herein, namely ground-based RF technologies and self-contained sensors.