Pressure sensors provide indoor competency for navigation


by Gerhard Lammel and Julia Patzelt, Bosch SensorTech

Inside of buildings, even the newest GPS navigation units quickly hit their limits–their navigational abilities are so inexact that a few floors can stand between the goal they indicate and the actual one. These are extremely poor qualifications for location-based services. An experimental study recently proved that the BMP085 pressure sensor can resolve not only the imprecision of GPS indoor navigation in a cost-efficient manner and without requiring additional infrastructure, it can also substantially improve locating in urban canyons.

In order for a navigation unit with GPS to provide exact positioning information within a few meters, the device has to receive data simultaneously from as many GPS satellites as possible. This requirement is fulfilled most easily on flat land under the open sky. However, as soon as obstacles block the unit’s view of the satellites, and it receives data from fewer than four satellites, the 3D positioning quickly becomes a game of chance.

Older GPS units are practically blind inside of buildings. A single wall within a structure generally damps a GPS signal in the 1.5GHz range by 20–30dB (factor of 100 to 1000), while reinforced concrete has the greatest dampening impact. The weak, useful signal is then drowned in the static of broadband HF receiving modules in these situations; a signal-to-noise ratio sufficient for determining position can no longer be reached using correction algorithms, and the navigation unit fails.

Indoor navigation: More than comfort

Unerring navigation within buildings is as advantageous for pedestrians as outdoor navigation is for drivers. In large foreign airports, for example, passengers could effortlessly find the correct check-in counter and from thence their departure gate. In large shopping centers and malls, they could be guided directly to businesses or restaurant without detours. New forms of location-based services, linked to GPS-equipped cell phones, could be offered for these environments. Imagine a note about a special sale at a certain store, or a restaurant’s reasonably priced lunch menu; once a person expresses interest, a GPS-equipped cell phone could immediately provide navigation to the new goal. For professional usage, examples include various types of support–a maintenance technician unfamiliar with the plant could traverse an extensive industrial complex, or merchandise could be unerringly delivered, even by someone not well acquainted with the locale. The possibilities for use by first responders are of incalculable value–no longer would they have to fumble through complex corridors, instead they could be directly guided to the location where they were needed.

Small errors have embarrassing effects without floor accuracy

All of the uses for indoor navigation listed are based on the concept that navigation within buildings decisively requires vertical “floor accuracy” (along the z-axis). If one assumes that building floors are 4m in height, then a positional accuracy of only 10 vertical meters for a GPS unit can easily steer one to the wrong floor. The people so affected would, however, only notice the error when they had arrived at the putative goal, then have to retrace their steps, perhaps along both floors in question, in order to finally arrive at their destination. This type of indoor navigation is both worthless and counterproductive. A positioning error of 10m has far less impact when it occurs horizontally (along the x- or y-axis); in the former case, floors and ceilings do not bar one’s view of the goal, which can generally be quickly reached by traveling a few more steps.

Figure 1. Indoor altitude measurement values comprised using a modern GPS navigation device in an office building. The fluctuations from the actual value (30m) are substantial over time.
Click here to enlarge image

A limited type of GPS navigation within buildings has only become possible with the most recent navigation devices, equipped with new, highly sensitive receivers. However, multipath reception is omnipresent in buildings, caused by repeated signal reflections off walls, ceilings and floors; distortions of the signal propagation time significantly reduce the accuracy of positioning within buildings, especially in comparison with navigation under the open sky. Figure 1 quantifies the quintessential positioning error along the z-axis. The statistical measuring point was located at a height of 30m in the third floor of a three-story office building. The navigational unit used for the test could determine the elevation within the building, but without great accuracy. Over the course of the observation time of ~13min, during which time the actual position of the device at 30m did not change, the position measured along the z-axis fluctuated between 10–50m. Despite the most modern reception technology, the indoor positioning error of ±20m greatly exceeded that which is acceptable for floor-accurate navigation (=4m).

Pressure sensor for floor-accurate GPS indoor navigation

With its BMP085 pressure sensor, Bosch SensorTech already has the solution to the “floor-accurate GPS navigation” problem, with a very small (5mm × 1.2mm), yet highly exact digital barometric pressure sensor, constructed using MEMS (microelectromechanical systems) technology. At every point in the earth’s atmosphere, air pressure and elevation (as it relates to sea level) have a fixed relationship; by measuring the air pressure, the exact altitude of a measuring point can be calculated. The BMP085 has extremely high-pressure resolution of max. ±0.03 hPa (RMS), which when converted to altitude corresponds to a resolution of ±0.25m (at sea level). At this level of accuracy, the sensor can essentially recognize the difference in altitude that a person undergoes as they move from one step to the next on a flight of stairs.

Figure 2. The SiRFstarIII chip set offers a user input function for storing altitude information gained using a separate pressure sensor.
Click here to enlarge image

A prerequisite for transmitting exact knowledge of momentary elevation to a chip set is excluding the less exact altitude information, generated in parallel via GPS. Bosch SensorTech’s SiRFstarIII chip set (Figure 2) incorporates the usual NMEA protocol, plus a proprietary binary protocol that enables settings that reach deeper on the chip set, making it possible to influence the type of positional calculation. In normal operation, the chip set decides automatically about the best type of calculation (four possibilities are available) with regard to the situation, independent from the number of satellites the receiver can see at the moment. The “2D fix” type of calculation is the most favorable for inputting barometrically measured altitude; the chip set normally uses this when it only has reception from three satellites. By this means it sets the elevation to a fixed value in order to reduce the number of unknowns in its formula for calculating position.

Results of an experimental study using a BPM085/GPS coupling

An “Alt Hold Mode” feature with the SiRF technology enables storage of the navigating engine in the SiRFstarIII chip with the measured values of the BMP085. Bosch SensorTech conducted an experimental study using practical tests to examine which results this BMP-GPS coupling would involve. The goal was to determine if the coupling of the GPS chip set with the pressure sensor provided not only a better level of accuracy for vertical navigation, but also whether this positively affected horizontal navigation.

The barometric pressure measurement increases the accuracy of GPS navigation: vertically, in every condition horizontally, and only in urban canyons.
Click here to enlarge image

After setting the Alt Hold Mode parameter to “always use input altitude,” the SiRF chip set adopted the barometrically measured elevation as the new input variable. Also, a software interface was installed between the pressure sensor and the GPS chip set to turn off several negative influences on the positioning accuracy:

Recognition of climatic and artificial pressure fluctuations. Weather can cause strong fluctuations in air pressure (up to ±40hPa) that should not be interpreted as changes in elevation. An already extant algorithm analyzed the course of pressure fluctuations, and excluded rather slow changes typical of weather influences (<2.5hPa/h) from measured values. It similarly resolved abrupt, artificial pressure fluctuations, caused by air conditioning, ventilation systems, or a strong wind through open windows.

Statistical reduction of measurement errors. A BMP085 can perform up to 128 pressure measurements per second. Averaging over several measured values eliminates statistical outliers.

Figure 3. Principle of dynamic zero compensation. During indoor navigation, the high-resolution BMP085 signal replaces the weak, inexact GPS signal (left). During periods of strong reception, the highly accurate GPS signal calibrates the pressure sensor (middle). In urban canyons, the pressure sensor again replaces the GPS altitude signal, which has become unreliable due to multipath reception.
Click here to enlarge image

Calibration compensates for absolute measurement errors. Independent of its height resolution, the BMP085 has an absolute measurement error of ±2.5hPa (±20m). This error was demonstrated to be easily compensated for–at good GPS reception, i.e. at the maximum GPS positional accuracy, height values determined via GPS by the navigation device were used automatically as calibration values for the pressure sensor. If the GPS reception deteriorated, then the high-resolution barometric altitude values–based on the most recently calibrated values–came into play (Figure 3). A Kalman filter determined whether or not to use situational calibration in this example.

The results of the experimental study of a SiRFstarIII with a BMP085 as external user altitude input are compiled (see table on p.22). The pressure sensor guaranteed indoor navigation at a floor-accurate level, thereby allowing the user to reach his or her destination; it also significantly increased the vertical positioning accuracy when used outdoors. In addition to a drastic reduction in the altitude error (visible only in the bar graph), the error in horizontal direction due to sensor support is also significantly smaller, with a standard deviation reduced by ~60%. This unexpected positive effect on horizontal positioning could, however, only be observed in comparable situations, such as in “urban canyons”.

BMP085 pressure sensor

Introduced in 2008, the BMP085 in an LCC-8 housing has a measuring range of 300–1100hPa and high over-pressure resistance at 10,000 hPa. Specifically developed for use in consumer electronic mobile devices, the sensor requires only 3µA of power (in ultra-low-power mode), with a low idle current consumption in stand-by mode of 0.1µA. The minimum supply voltage was also reduced to its current 1.8V. A 19-bit measurement operation on applications takes place inclusive of calibration data for temperature compensation in serial via an I2C 2-wire interface, which simplifies the integration of the BMP085 into already extant applications and eliminates the need for additional external components for wiring. At its slowest, a new measured value is ready for collection every 7.5ms.

Dr. Gerhard Lammel is manager of engineering at Bosch SensorTech GmbH, a wholly owned subsidiary of Bosch SensorTech.
Julia Patzelt is responsible for marketing communication and public relations at Bosch SensorTech GmbH.

No infrastructure required

At present, alternatives to satellite navigation are being developed and tested globally for indoor and urban navigation. Locating takes place by cross bearing, determining position by using the distance-dependent intensity of HF signals, which can be emitted by terrestrial transmitters from exactly known locations. The specific advantage is that the current transmission networks for cellular structures can be used for this form of navigation, without requiring a completely new system of transmitters.

This type of system, which functions using currently available WLAN islands, was developed by the Fraunhofer Institute for Integrated Circuits (IIS). The navigation unit constantly analyzes the reception field strength of nearby WLAN spots, and by means of their SSIDs, obtains the exact location of the WLAN transmitter from a central databank, which it then uses to navigate. However, this is not sufficient to also navigate within buildings with a sufficient degree of accuracy. For indoor navigation to function correctly, the navigation unit has to recognize the additional field strength divisions in every building, which in turn requires preliminary mapping measurements. Solutions related to this one convert the same navigational principle, but use different radio systems, such as GSM or DECT.

A communications network infrastructure is imperative for each of these solutions, which in turn requires an (at least periodic) administrative expenditure (e.g. mapping) which is not insignificant. The combination of the BMP085 pressure sensor and GPS is completely different does not require any foreign infrastructure; it functions everywhere, all over the planet, and is completely autarchic, in that the pressure sensor can fundamentally increase the vertical resolution of each of the alternative navigation systems.