Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Real-time Outdoor Positioning System Exploiting Ultrasonic Wireless Sensor Nodes

Young-yeol Choo1, Eung-Joo Lee1, Sooyol Ok1,


1

College of Information & Communication, Tongmyong Univ., 535 Yongdang-Dong, Nam-Gu, Busan, Korea {yychoo, ejlee, sooyol}@tu.ac.kr

Abstract. In this paper, we present a passive positioning system using Cricket Mote sensors. In order to improve accuracy of the passive system, the transmission speed of ultrasound was compensated according to air temperature at the moment. Upper and lower bounds of a distance estimation were set up through measuring minimum and maximum distances that ultrasonic signal can reach to. Distance estimations beyond the upper and the lower bounds were filtered off as errors in our scheme. We measured distances with Cricket nodes at several points which are 50 cm apart respectively and compared them with real distances to compensate the deviation at each point. A filtering method was proposed to discard error data during movement of target device. At outdoor tests, average location error and average location tracking period were 3.5 cm and 0.5 second, respectively, which outperformed original Cricket location system. Keywords: Wireless Sensor Network, Positioning System, Ultrasonic Sensor, Pervasive Computing.

Introduction

The WSN (Wireless Sensor Network) is a basic technology Ubiquitous Computing and is a new-concept IT technology that can implement a variety of services by collecting and analyzing information delivered from various kinds of sensors using wireless networks. In order to implement a smart environment in Ubiquitous Computing, various information is necessary for context-awareness, including temperature, humidity, luminance, pressure, acceleration, gradient, image data, and so on [1], [2]. The location of objects, in particular, is important information, on the basis of which many services are being created [3], [4]. To get information on the location, different technologies are applied indoors and outdoors, depending upon environmental conditions. In outdoor environments, at present, various location sensing services are used through the GPS (Global Positioning System) and GIS (Geographic Information System), while for indoor situations, studies on location tracking technologies using ultra sound, RF and image processing are in progress. Each system, however, has

different applicable environments and limitations due to technical characteristics. A GPS has just few meters of precision outdoors, but because it receives signals from satellites it cannot be used indoors. In the case of indoor location tracking, diversified techniques have been proposed, such as ultrasound, camera, and RF signals [2-5]. These technologies have a lot of limits, including multi-path problem, sensing distance and area. Location sensing systems of mobile devices indoors can be divided into active mobile device method and passive one according to where signals for measuring distance are generated from [2]. The active system employs a structure in which signals for distance measure are transmitted from a positioning target. Among the active methods are Active Badge [6] of AT&T using ultra sounds, Microsoft Corporation's RADAR [7] using Active Bat [8], RSS (Receive Signal Strength) of RF signals, and PAL (Precision Asset Location System) of Multispectral Solutions Inc. using UWB (Ultra Wide-band) signals [9]. On the other hand, in the passive system, target objects equip a receiver which senses signals for distance measurement. An example is MIT's Cricket Location System [10] using ultrasounds. In Active Bat, a representative active mobile device method, a sender called "Bat" transmits a synchronized radio signal and an ultrasonic pulse, and receivers measure the distance between each receiver and the sender by making use of a time difference of arrival between these two signals. Each receiver delivers this distance information to the host, connected via wired lines, which then calculates the location by using collected information. In case of active systems adopting this principle, as many receivers receive a signal generated at a certain time even while the object is moving, the location at that time can be calculated, thereby being able to have a comparatively high precision. However, it is necessary to establish a separate wired network for synchronizing the receivers, and there are other problems, such as high cost and complexity of installation in conjunction with the network, and privacy caused by the host-based centralized processing. On the contrary, the passive locating tracking technique like MIT's Cricket utilizes a method that calculates the distance between each beacon and receivers by using RF and ultrasonic signals transmitted from three or more beacons fixed inside the building, and thus three-dimensional locations can be calculated directly at mobile devices. Systems of this type do not need a separate synchronization among transmitters sending beacon signals, thereby providing advantages such as low cost, simplicity of installation, and solving the privacy problem, which are primary drawbacks of the active system. But because the distance to each transmitter becomes closer or farther as receivers move, a distance error arises due to Doppler effect in the case of ultrasonic waves. In addition, because beacons are not synchronized, each beacon sends RF and ultrasonic signals at different time. Hence, listener waits until three different signals are delivered at least, which makes update speed slow. In the meantime, a Hybrid Architecture [5] combining the active technique and the passive one has been proposed, but it accompanies a delay in processing time due to performing EKF (Extended Kalman Filter) and LSQ (Least Square Minimization) algorithms, stills demonstrating a lower precision in comparison with active systems. In this paper, we have designed and implemented a system in order to upgrade precision and location update speed, which are drawbacks of the passive method, by using the Cricket hardware. We also have described methods to get over the

inaccuracy of ultrasonic signals and their applicability through comparative experiments.

Location Calculation Algorithm

The distance measuring system comprises four beacons, a listener, and a PC as base station. Four fixed beacons send RF and ultrasonic sounds at the same time to the listener mounted on a vehicle, which then transmits the data to the base station to calculate coordinate of an object. The ultrasonic sensor communicates using a RF frequency of 433MHz band and ultrasonic signals are transmitted and received at 40KHz approximately. The software was developed with NesC language under TinyOS [11]. To find a 3-dimensional coordinate using the TDoA (Time Difference of Arrival) algorithm [12], it is necessary to establish three or more beacons and one or more listeners. In this paper, a hexahedron in the Cartesian coordinate was composed using four beacons. Each beacon transmits a value randomly at an average cycle of 0.25 second.

Fig. 1. Installation of beacons and listeners to calculate three dimensional coordinate. Bi is the position of i-th beacon. a and b represent the minimum and the maximum distance values allowable for a beacon, respectively.

With this method, the four distance values are saved each in a different buffer and the coordinate vector of the listener is obtained by triangulation with three distance values among them, as follows:
( x xB1 ) 2 + ( y yB1 ) 2 + ( z z B1 ) 2 = d12
(1)
2

( x xB 2 ) 2 + ( y y B 2 ) 2 + ( z z B 2 ) 2 = d 2

(2) (3)

( x xB3 ) 2 + ( y y B3 ) 2 + ( z z B 3 ) 2 = d 3

In the above equations, xBi, yBi and zBi denote the x, y and z coordinates of beacon Bi, respectively. After installation, each beacon is fixed and thus its absolute location can be found. di is the distance value measured by the i-th beacon and received by the
B

listener. The 3D coordinate of a location-tracked object, therefore, can be found by correlating the above three di. With the four beacons installed at the same height, the z coordinate of each beacon becomes equal and Equations (1), (2), and (3) can be simplified as follows:
x = [( y B1 y B 2 )((d 32 d 2 2 ) ( xB32 xB 2 2 ) ( y B32 y B 2 2 )) + ( y B 2 y B 3 )((d12 d 3 2 ) ( x B12 x B 2 2 ) ( y B12 y B 2 2 ))] /

2[( xB 2 xB1 )( y B 2 y B3 ) + ( xB 2 xB3 )( y B1 y B 2 )]


y = [( xB1 xB 2 )(( xB3 xB 2 ) + ( y B 3 y B 2 ) (d 3 d 2 ))
2 2 2 2 2 2

(4)

+ ( x B 2 x B 3 )(( x B12 x B 2 2 ) + ( y B12 y B 2 2 ) (d12 d 3 2 ))] /

2[( xB 2 xB1 )( y B 2 y B 3 ) + ( xB 2 xB 3 )( y B1 y B 2 )]

(5) (6)

z = z B1 + d12 ( x x B1 ) 2 ( y y B1 ) 2

The coordinate (x, y, z) of an object to be tracked can be found by solving equations (4), (5) and (6). The values of a and b in Fig. 2 become a minimum value and a maximum one of di that can be obtained between beacons and the listener. Accordingly, values out of the range a d i b are considered as error data. Three distance values received after going through this filtering are saved in buffers and used for coordinate calculation. This process is performed at base stations and its flow diagram is shown in Fig. 2.

Fig. 2. Software flow diagram

2.1

Data Compensation with Temperature Measurement

In TDoA scheme, the distance measurement is accomplished by measuring propagation times of ultrasound signal on air. The propagation time is dependent on the speed of sound and greatly influenced by a change in air temperature. As the speed of sound becomes faster in proportion to temperature increase, it works as an error factor to the real distance. Hence, the distance between a beacon and a listener was corrected according to Equation (7) below. speed_of_sound = 331.4+0.60*T , (7)

where T is the temperature measured from sensor on top of a Cricket node at the moment. 2.2 Correcting Distance Error caused from Ultrasonic Signal Attenuation

As the distance becomes farther from a vibration plate generating an ultrasonic signal in an ultrasonic sensor nodes, the signal strength is reduced and distance error increases. This is inherent feature of ultrasonic signal and reduces precision of positioning systems based on ultrasonic technology. To compensate these errors, we conducted a test to evaluate the deviation between real distance and measured distance by the location system. At first, we measured outdoor ground every 50cm with a laser distance meter and designated points from the origin to 11m. That is, first point is 50cm apart from the starting point, the second point is 1m apart from it, and the third one is 1.5m apart. Then, a beacon was positioned at the origin to send RF and ultrasonic signals. A Cricket node listened RF and ultrasonic signals to measure the distance at each designated points. These measured distances by the Cricket node were compared with real distances, respectively. This experiment revealed that the distance error at each point increased as the point went farther from the origin. The result of this experiment is shown in Fig. 3. As shown in Fig. 3, distance error was proportional to the distance from the starting point. The experiment was repeated for other Cricket nodes and revealed same results.

Fig. 3. Distance error between real data and measured data. Errors increase with distance.

Distance error of a point locating between two adjacent designated points was achieved by interpolating their error values. Using these data, Equation (8) could be derived to compensate errors caused from inherent feature of ultrasonic signal.
d = 1.1(d m 0.5)

(8)

where dm and d, are real distance and measured distance with a Cricket node, respectively and unit of distance value is meter. 2.3 Discarding Errors on Movement

Although errors due to the characteristics of the Cricket's ultrasonic sensors were reduced, an error within the minimum/maximum values of measure distances continued to exist. In order to solve this problem, we proposed the following algorithm. The values of distance measured with each beacon are saved in temporary buffers, and after the distance values in temporary buffers are used for calculating the location, are saved in calculation buffers. When a measured distance value is saved in a temporary buffer, the variation from the distance saved in the calculation buffer is compared with its allowable error. In this paper, allowable error is defined as two times the standard deviation of the data collected at section 2.2. It is used as a criterion for distance deviation. If the deviation is the same as or less than the allowable error, average value of the distance value saved in the temporary buffer and the one saved in the calculation buffer is calculated to be saved in the calculation buffer. This value is used to calculate the location of an object at the moment. Fig. 4 explains this process through an example.

Fig. 4. Buffer processing due to a change in distance

On the other hand, if the deviation is larger than the allowable error, the system waits for distance value from another beacon. Upon arrival of the distance value, the system tests it. If the deviation of the second distance is less than the allowable error, the first measured distance is considered as an error and removed (refer to Fig. 5).

Fig. 5. An error distinguishing process

If the second variation is larger than the allowable error, listener is considered to have moved and then the distance of the temporary buffer is reflected into the calculation buffer. In this case, the location is calculated using the distance value of the calculation buffer. The distance of unmeasured beacons are corrected by using the calculated location (refer to Fig. 6).

Fig. 6. Location calculation after movement

3.

Experiments

The ultrasonic sensor has properties similar to general speakers, causing a multi-path phenomenon through diffraction or reflection without avoiding obstacles. The ultrasonic sensors are also subject to the effect of fluorescent tube (stabilizer of the tube), man or wall indoors. There were rather more errors indoors than testing outdoors. Taking account of this phenomenon, in order to secure the sensing area of rectangular parallelepiped space as Fig. 1, a method suitably adjusting the angle and direction was taken by referring to the transmission pattern in [14]. The transmission pattern of ultrasound is 80 to 90 degrees in sending angle and the angle decreases as the distance goes farther. This could also be verified through experiments. In case of installing them by considering the characteristics of devices and the range, within which a stable distance can be sensed and thereby adjusting the angle and height of beacons, it could be confirmed that the maximum size of a rectangular parallelepiped that locations can be obtained is (x, y, z) = (430, 430, 630). We experimented and compared with the MIT Cricket Location System under the following conditions so as to verify the effectiveness of proposed algorithm. As an experimental environment, we mounted a wireless listener in a car and measured a position error in a space whose length, width and height are 150cm each, and took the average of three measurements while moving by a 10-cm increment of measurement unit. The results with applying the algorithm and those without applying it are displayed in Fig. 7, respectively. Fig. 7 (a) shows positioning error data measured by the MIT Cricket Location System, which adopted Extended Kalman filter for error reduction. The numbers in the graph reveals position error values in centimeter. Position errors ranged from 3cm to 6,5cm and average of them was 4.4cm. It took 1,000-3,000ms or longer to update location of an object.

(a) (b) Fig. 7. Positioning error distribution according to (x, y, z) coordinate. (a) represents the error distribution of Cricket Mote system. (b) represents the error distribution when proposed methods were applied. Proposed system reveals less error values and concentrated pattern.

Fig. 7 (b) shows the result obtained by applying compensation scheme proposed above. The scheme comprises five methods as follows: Adjusting ultrasound propagation speed according to temperature Defining the minimum and the maximum values of a distance Correcting distance errors caused from ultrasonic signal attenuation Discarding errors of moving object using buffers and queue Avoiding ultrasonic collision of each beacon The measured errors were in the range of 0-6cm with an average of 3.5cm, and the update time of location tracking was 250-750ms, which is very close to the active location tracking method. The measured value around the center, in particular, had an accurate value of 0-1cm, and it was confirmed that the error became larger as approaching closer toward the corner right below each beacon. The result of position measurement errors between the two systems is summarized in Table 1.

Table 1. Comparison of location error MIT(EKF) Proposed system Ave. 4.43 3.5 Location error Min 3 0 (cm) Max 6.5 6.0 Location tracking cycle (ms) 1000~ 3000 250~750

Conclusions

In order to reduce the tracking cycle of a passive location tracking system and improve its precision, this study designed a system based on a few essentially simple and clear contents and have induced the grafting of a new idea through experimental assessments and conclusions. Proposed system used Cricket mote, and for reducing a distance measuring error, integrated the data of temperature sensors, corrected ultrasonic transmission speeds, and proposed an algorithm for error signal removal method and beacon signal error correction. With proposed algorithm, positioning errors of the system were reduced by 20% in average and the location tracking cycle was improved four times than that of conventional Cricket Location System. Especially, the location data were very accurate at the center of sensing area. Bluetooth communication facilitates configuration of positioning system with ease data exchange between sensor node and base station. On the other hand, while there have been improvements in location tracking update time and precision, energy consumption was increased, which was caused from shortening the transmitting cycle of beacons for faster update. Acknowledgments. This research was supported by the MIC(Ministry of Information and Communication), Korea, under the ITRC(Information Technology Research Center) support program supervised by the IITA (Institute of Information Technology Advancement)" (IITA-2006-C1090-0602-0013).

References
1. Akyildiz, I. F., Su W, Sankarasubramaniam, Y., Cayirci, E.: A Survey on Sensor Networks, IEEE Communication Magazine, (2002) 102-114 2. Chen, G., Kotz, D.: A Survey of Context-aware Mobile Computing Research, Technical Report TR2000-381, Dept. of Computer Science, Dartmouth College (2000). 3. Zeimpekis, V., Giaglis, G. M., Lekakos, G.: A Taxonomy of Indoor and Outdoor Positioning Techniques for Mobile Location Services, ACM SIGecom Exchanges, Vol. 3. (2003) 19-27 4. Dijk, E. O., van Berkel, C. H., Aarts, R. M., van Loenen, E. J.: A 3-D Indoor Positioning Method Using a Single Compact Base Station, Proc. of the Second IEEE Annual Conf. on Pervasive Computing and Communications, (2004) 101-110 5. Smith, A., Balakrishnan, H., Goraczko, M., Priyantha, N.: Tracking Moving Devices with the Cricket Location System, Proc. of Int. Conf. On Mobile System, Applications and Services, Vol. 13 (2004) 190-202 6. Want, R., Hopper, A., Falcao, V., Gibbons, J.: The Active Badge Location System, ACM Transaction on Information System, Vol. 10. (1992) 91-102 7. Harter, A., Hopper, A., Steggles, P., Ward, A., Webster, P.: The Anatomy of a Contextaware Applications, Proceedings of 5th Annual ACM/IEEE Int. Conf. On Mobile Computing and Networking, (1999) 59-68 8. Bahl, P., Padmanabhan, V. N.: RADAR : An In-building RF-based User Location and Tracking System, INFOCOM, (2000) 775-784

9. Fontana, R. J., Richley, E., Barney, J.: Commercialization of an Ultra Wideband Precision Asset Location System, Proc. of IEEE Conf. on Ultra Wideband System and Technologies, (2003) 10. Priyantha, N. B., Chakraborty, A., Balakrishnan, H.: The Cricket Location-support System, ACM MOBICOM (2000) 32-43 11. Tinyos home page. http://www.tinyos.net 12. Ilyas, M., Mahgoub, I.: Handbook of Sensor Networks: Compact Wireless and Wired Sensing System, CRC PRESS LLC, New York (2005) 13. Duff, P., Muller, H.: Autocalibration Algorithm for Ultrasonic Location System, ISWC, (2003) 62-68 14. Balakrishnan, H., Nakamura, T., Matsuo, M.: Location Handling System for Mobile Computing, NTT Technical Review, Vol. 1. (2003)

You might also like