Robust ultrasonic indoor positioning using transmitter arrays

Transcription

1 2010 INTERNATIONAL CONFERENCE ON INDOOR POSITIONING AND INDOOR NAVIGATION (IPIN), SEPTEMBER 2010, ZÜRICH, SWITZERLAND Robust ultrasonic indoor positioning using transmitter arrays Sverre Holm and Carl-Inge C. Nilsen Department of Informatics, University of Oslo, P. O. Box 1080, NO-0316 Oslo, Norway Abstract As time-delay based ultrasound positioning often is noise sensitive at long ranges, the goal of the research reported here is to achieve sub-room ultrasound positioning with other methods. By combining a portable ultrasound receiver which measures signal strength with a transmitter array that sends steered, coded beams inside a room, the tag can determine which beam it is located in and carry out fine-positioning. The concept is demonstrated in an experiment using a 40 khz system with 4-7 transmitter elements. I. INTRODUCTION Ultrasound positioning systems can be classified according to the need for RF, i.e. whether they are 1) based on ultrasound alone 2) hybrid, combining ultrasound and RF Usually the first class of systems involves portable devices with ultrasound transmitters. They are therefore infrastructurebased systems as the positioning will take place at a central processor which collects data from all receiver nodes. Examples of such systems are described in [1], [2]. The second class of systems are often privacy-based as the portable device usually contains an ultrasound receiver which is crucial for the positioning. The portable device performs its own positioning and may use the RF unit to communicate the positions to others such as in [3]. Ultrasound positioning systems can also be classified according to the positioning principle: A) based on time-delay, angle-of-arrival or timedifference-of-arrival B) based on the ability to communicate, signal level and/or Doppler shift Class A systems usually aim for accuracies in the 1-30 cm range and are the dominant systems if one looks at the current research literature. Systems in class B will have much lower resolution, down to room-level positioning, and they are not so well covered in the research literature, nevertheless they play a role in practical applications. We have previously developed ultrasound systems using only the ability to communicate as a positioning criterion (class 1B). Such a system may also be classified as a binary signal strength indicating systems, as they indirectly measure whether the signal can be received or not. Such a system has the capability to indicate in which room a transmitter is located [2], [4] and it is now used commercially for tracking of assets and personnel in e.g. a hospital environment ( This is our reference system and /10$26.00 c IEEE statements in this paper about fine-positioning should be understood in the context of being improvements to roomlevel accuracy. Also, when discussing robustness, it should be understood in the context of robustness of positioning over ranges that are 5-10 meters and more. Although the formerly described system is in commercial use, it has some shortcomings, the primary one is a relatively low update rate due to the low bit rate of the ultrasound communications channel. This results in a chance to miss items if several objects are to be located in a short time. Despite this, a small scale version of the system was tested in [5] and found to have greater zonal accuracy and reliability than similar RFID systems. The almost 100 % room-level reliability was one of the reasons why it was described in a feature article in Scientific American in 2008 [6]. Sometimes a second disadvantage of systems using only ultrasound (class 1), is that portable ultrasound transmitters may expose human bearers to levels that are near the maximum recommended levels, due to proximity to the ears. An in-depth analysis of output levels compared to international and national safety standards can be found in [7]. Such an analysis will have to be done more often as ultrasound systems move from the laboratory to commercial use. Many hybrid systems of class 2A have been developed, such as Active Bat [1], Cricket [3] and Dolphin [8]. Cricket claims an accuracy in the 1-3 cm range while Dolphin claims an accuracy of about 15 cm. Thus they have much higher accuracies than the previous system, but usually a much lower range, typically the experience is that they easily break down in real life when the range is more than a few meters. This is due to the reduction in received signal level, as level falls with 1/r, coupled with the large variation in the background noise level. This makes the received signal to noise ratio too low for reliable detection beyond a certain range. A classical test is key jingling. This is a test which will cause breakdown of most, if not all, time-delay based systems [9]. The main difference between the short range, high accuracy (class A) and long range, low accuracy (class B) systems is the bandwidth. Typically in order to measure time delay, one requires a time resolution much less than 1 ms. Since time resolution and bandwidth are inverse proportional, this means that the bandwidth has to be many khz. On the other hand, if one wants to measure only signal strength or Doppler shift, one can use a very low bandwidth, such as 25 Hz as in [2]. Actually this design is modelled after similar robust underwater acoustic communications systems for operation

2 under adverse conditions [10], [11]. In [7] we have compared the link budget of these two kinds of systems using bandwidths of 4 khz and 25 Hz as examples. Assuming a background of white noise in this bandwidth, not unrealistic for typical industrial noise sources [12], the timedelay based system will pick up 10log(4000/25) = 22 db more noise power. As signal power falls with 20 log(1/r) this is the same as decreasing the range by a factor of 10 22/20 = In practice one also has to take absorption into account [13], so the difference will not be as large as this simple calculation has indicated. In [7], link budget analysis with the above assumptions predicts a reliable range of 14.3 m for the 25 Hz bandwidth system and 5 meters for the 4000 Hz bandwidth system. Also, we have analyzed the variation in noise level in realistic environments and found that it may vary up to ±30 db. Thus the above claimed reliable range may be reduced to almost nothing in the worst case, especially for the wide-band system. This is why we claim that wide bandwidth, time-delay based systems often have a very low robustness to external noise, especially as the range approaches 10 meters and more such as is feasible for class 1B systems. To overcome the limitations of both class 1B systems and class 2A systems we therefore focus here on a hybrid system with a reversed flow of ultrasound, so that the portable tag only contains an ultrasound receiver and no transmitter. In addition there is an RF unit for communicating the data [7], making it a system of class 2B, using a narrowband ultrasound receiver for signal strength measurement. The concept described in [2], [4] is today used industrially in such a setting, as an augmentation of a WiFi-location system ( WiFilocation typically has a median accuracy in the 3-meter range, and with 97 % of the errors less than 10 meters [14]. Thus there is a chance for large outliers which in the worst case may indicate a false position in an adjacent room or even on an adjacent floor. There is also good indications that such outliers are unavoidable with RSSI-based fingerprinting and WiFi [14]. However, when combined with room-level indicating ultrasound, the large errors that fall in adjacent rooms and floors are eliminated. The goal of the research reported here is to determine if accuracies better than room-level can be achieved with a low bandwidth, long range system. The goal is therefore to find how accurately a system can perform in a real-life environment without relying on time-delay estimation. The objective is reliable positioning in rooms in a hospital, so a range of 10 meters or more is desirable. The required components are a portable tag with an ultrasound receiver which outputs the Received Signal Strength Indicator (RSSI) value, and in the future also the velocity, as described in [7] and one or more stationary, array-based ultrasound transmitters. II. ARRAY BASED SYSTEM A typical application where the proposed setup could be beneficial is in a hospital setting where the goal is to determine in which sector an instrument or a patient is located in a multibed hospital room as shown in Fig. 1. Two arrays can also be Fig. 1. Fig. 2. Determination of sectors in a multi-bed hospital room Use of 2 arrays to position in smaller cells. used on perpendicular walls in order to perform positioning in even smaller cells, say of size 1 m by 1 m, as in Fig. 2. The new feature of this paper is the array-based transmitter, which can be either 1- or 2-dimensional. The 1-D array is usually oriented in the horizontal plane, e.g. high up on a wall. Such 1-D arrays have been used for a long time in ultrasound imaging applications [15] and in sonar, and 2-D arrays are now used in 3-D ultrasound imaging systems [16]. The array is configured to transmit data and steer its beam electronically in the array plane. In this way sectors may be formed, typically 3-7, to cover a medium sized room. The transmitter first sends its data into sector 1, then shifts its beam to sector 2, and so on, and may repeat this pattern continuously. The transmitted data may consist of an optional room ID + sector IDs. The array is used in combination with RSSI measurements in the movable receiver. By comparing the RSSI-values, the tag will know in which sector it is located.

3 z Beam φ l sin φ x l Fig. 3. Generation of steered beam with delays Crucial parameters of this configuration are the width of the emitted beam, and the angular increment between the beams. A. Beamwidth An aperture will send out a beam of width Fig. 4. Photo of transmitter array θ λ/d. (1) The equation gives an estimate of the beamwidth in radians as measured between the 3 db points, where λ is the wavelength and D is the effective aperture. B. Beamsteering The beam steering takes place by controlling the relative delays between individual elements in the array as shown in Fig. 3. If all alements are excited in phase, the beam will be sent broadside to the array. The relatively small apertures can all be considered to operate in the farfield as the nearfield-farfield transition takes place at approximately D 2 /λ which for all the apertures considered here is less than or equal to /8.6 mm 0.16 m. Therefore the outgoing beams from the different parts of the aperture can all be considered to be parallel and in order to steer to an angle φ relative to broadside the relative delay between two elements has to be τ = lsinφ/c. (2) where l is the inter-element spacing and c is the speed of sound in air. A. Transmitter III. HARDWARE We built the array shown in Fig. 4. It is composed of 7 Murata MA40S4S transducers, each of outer diameter d o = 9.9 mm and with an inner, active aperture of d i 7 mm. Operating at 40 khz, the wavelength is λ = c/f = 343/ mm. In the horizontal plane, the long-axis of the array, where the aperture is D = 3 d o + d i, the beamwidth (Eq. 1) is θ h 8.6/36.7= 0.23 radians or In the vertical plane we used two different configurations. The first one used only a single row of 4 elements, i.e. a vertical beamwidth of θ v1 λ/d i = 8.6/7 = 1.23 radians or 70. Later we also tested with all seven elements so that the effective vertical aperture was more than doubled (15 mm), i.e. θ v2 8.6/15 = 0.57 radians or 33. We generated horizontally steered beams from the array of Fig. 4 at about 40 khz. Each element in the array was driven by an amplifier which was individually excited by digital output signals from a Xilinx Spartan-3 FPGA development board. The output level was between 5 and 30 V p p. The FPGA was programmed with a transmit cycle of three unique codes. Each step of the transmit cycle consisted of 1) Loading the FSK (frequency shift keying) modulated signal for the ID associated with the current step. 2) Transmitting the signal after an element-specific delay determined by the sector angle associated with the current ID as given by Eq. 2. The desired IDs, their FSKmodulated signals, and the sector angles with corresponding delays were all hard-coded in the FPGA software. B. Receiver The receiver was the same as used in [7]. It was built to be a demonstrator for the ultrasonic part of the system. The receiver is based on a low-power 16-bit microcontroller which also has some DSP capabilities. It implements the FSK receiver according to [2], [4]. The ultrasound receiver runs from a small button cell battery (CR2032) and its block diagram is shown in Fig. 5 and a photo is shown in Fig. 6. Instead of the RF-link for output which is shown in the block diagram, the detected ID was shown on a display, and the ID, RSSI and Doppler shift from the demonstrator was sent over an RS232 interface. In [7] it was shown how this receiver together with an ultrasound transmitter can be used for predicting relative

4 65 RSSI vs ID, 0,1,2 = red,green,blue 60 Fig. 5. Block diagram for ultrasound tag with processor for decoding the ID code sent per beam, as well as estimation of receiver signal level (RSSI) and Doppler shift (velocity). db Time [s] (sampled every 0.25 sec) Fig. 7. Measured RSSI as the tag is slowly moved across the three beams (7-element array) at a distance of 4 meters. Fig. 6. Photo of ultrasound receiver with the main functions contained within the rectangle in the lower right-hand corner. 65 RSSI vs ID, 0,1,2 = red,green,blue changes in distance between the transmitter and receiver via the RSSI-measurement. It was also shown how the velocity vector along the line between the transmitter and receiver can be estimated. When integrated it can also be used to find range. In [7] it was demonstrated how the range found from received signal level and the range found by integrating the velocity agree well with each other. This confirms that at least in the small room tested there, level falls off with range in a fairly predictable way, and that our hardware estimates RSSI and Doppler shift reliably. The Doppler shift may be used for estimation of the velocity vector in conjunction with the array transmitter described here. So far this feature has not been tested in the array setting, but this is desirable to do in later studies. IV. RESULTS The experiments were done in a relatively narrow room of size (height) meters. The array was high up in a corner, and every second a new beam with a new ID was sent. We compared the two arrays, 4 elements and 7 elements, both with a horizontal beamwidth of about This paper reports results from initial testing of the concept. These first measurements with transmitter arrays are only concerned with the use of the RSSI-measurement capability. Fig. 7 shows the RSSI measured from the 7-element array, as the tag is slowly moved across the beams at a distance of 4 meters and at a height of about 1.5 meters.. From 5 to 14 seconds it is in the beam with ID=0 (steered to φ 0 = 15, dashed), from 14 to 26 seconds it is in the central broadside beam with ID=1 (solid) and φ 1 = 0 and for the rest of the time it is in the ID=2 beam (steered to φ 2 = 15, dash-dot). Thus, from a simple comparison of RSSI the sector number can be found. Why there is a small bump in the amplitude db Time [s] (sampled every 0.25 sec) Fig. 8. Measured RSSI with a stationary tag in the center of the central beam (4-element array) at a distance of 6 meters. of the φ 2 = 15 (dash-dot) curve at 20 sec, we cannot say for sure. It could be the result of some combination of direct and reflected sound that adds constructively at this particular position, or it could be an artefact of the measurement system. The array with the wider vertical beam gives similar results but the contrast is smaller, i.e. the distance in db between the actual beam in a certain direction and the other beams. This is believed to be due to the increased number of reflections via the ceiling and the floor. Finally Fig. 8 shows a result at 6 meters distance with the basic 4-element array. The tag is stationary in the broadside beam and one can see that the two other beams are weaker all of the time. This figure demonstrates the stationarity and stability of the amplitude values.

5 V. CONCLUSIONS Our first few results demonstrate the feasibility of horizontal beam steering for positioning. The stronger the direct beam compared to reflections from surfaces above, below and to the sides, the better the concept will work. This means that the larger the room the better, but even in our narrow lab the concept worked well. The influence from the reflections from the ceiling and floor seemed to be reduced with a narrower vertical beamwidth. The system will be tested in more varied environments in future research. It is in particular important to find out if the ratio of direct and reflected energy is high enough in various realistic environments and at all heights in the room for the system to indicate transmitter sector in a reliable way. ACKNOWLEDGMENT We want to thank Birkeland Innovation AS at the University of Oslo and especially Bjarne Tvete for many helpful discussions and also for partial financial support. REFERENCES [1] A. Ward, A. Jones, and A. Hopper, A new location technique for the active office, IEEE Personal Communications, vol. 4, no. 5, pp , Oct [2] S. Holm, O. B. Hovind, S. Rostad, and R. Holm, Indoors data communications using airborne ultrasound, in Proc. IEEE Int. Conf. Acoust., Speech, Sign. Proc, Philadelphia, PA, Mar. 2005, pp [3] N. Priyantha, A. Chakraborty, and H. Balakrishnan, The cricket location-support system, in Proc. 6th Ann. Int. Conf. Mobile Computing and Networking. ACM, 2000, pp [4] S. Holm, Airborne ultrasound data communications: The core of an indoor positioning system, in Proc. IEEE Ultrason. Symp., Rotterdam, Netherlands, Sep. 2005, pp [5] D. Clarke and A. Park, Active-RFID aystem accuracy and its implications for clinical applications, in Proc. 19th IEEE Symposium on Computer-Based Medical Systems, Oct. 2006, pp [6] L. Greenemeier, A positioning system that goes where GPS can t, Scientific American, Jan [7] S. Holm, Hybrid ultrasound-rfid indoor positioning: Combining the best of both worlds, in IEEE Int. Conf. RFID, Orlando, FL, Apr. 2009, pp [8] Y. Fukuju, M. Minami, H. Morikawa, and T. Aoyama, DOLPHIN: An autonomous indoor positioning system in ubiquitous computing environment, in IEEE Workshop on Software Technologies for Future Embedded Systems, vol. 53, 2003, pp [9] G. Welch and E. Foxlin, Motion tracking: No silver bullet, but a respectable arsenal, IEEE Computer Graphics and Applications, vol. 22, no. 6, pp , Nov/Dec [10] D. Wax, MFSK The Basis for Robust Acoustical Communications, in Proc. OCEANS, 1981, pp [11] K. Scussel, J. Rice, and S. Merriam, A new MFSK acoustic modem for operation in adverse underwater channels, in Proc. OCEANS, vol. 1, 1997, pp [12] H. Bass and L. Bolen, Ultrasonic background noise in industrial environments, Journ. Acoust. Soc. Am., vol. 78, pp , [13] H. Bass, L. Sutherland, A. Zuckerwar, D. Blackstock, and D. Hester, Atmospheric absorption of sound: Further developments, Journ. Acoust. Soc. Am., vol. 97, pp , [14] E. Elnahraway, X. X. Li, and M. R. P., The limits of localization using signal strength: A comparative study, in Proc. of the First IEEE Int. Conf. on Sensor and Ad hoc Communications and Networks, Santa Clara, CA, Oct. 2004, pp [15] S. Holm and K. Kristoffersen, Analysis of worst-case phase quantization sidelobes in focused beamforming, IEEE Trans. Ultrason., Ferroelec. Freq. Contr., vol. 39, no. 5, pp , [16] S. Holm and B. Elgetun, Properties of the beampattern of weight-and layout-optimized sparse arrays, IEEE Trans. Ultrason., Ferroelec. Freq. Contr., vol. 44, no. 5, pp , 1997.