radiation from a resident increases when a resident enters a sensing area, the PIR sensor outputs an increasing voltage. Conversely, the voltage decreases as the resident leave the sensing area. If the resident does not move within the sensing area, the variation in the infrared radiation does not exist and the PIR sensor outputs zero voltage. Therefore, it is very difficult to deter-mine when a resident is staying resident within a specific sensing area using only the voltage or current threshold of a PIR sensor.
Fig. 6. Signal output of PIR sensor.
In order to guarantee the location accuracy of the system, the resident-detection method must meet several requirements. First, if no resident is present within a sensing area, the PIR sensor should not output ?ON? signal. That is, the PIR sensor must not malfunction by other disturbances such as a moving pet, temperature change and sunlight. Second, it should be possible to precisely determine the point in time when a resident enters and leaves a sensing area. That is, in spite of variations in sensor characteristics, resident?s speed and height, it should be possible to determine the time point exactly. Finally, because the output voltage of a PIR sensor does not exceed the threshold voltage when the resident does not move within a sensing area, it is necessary to know if a resident stays within the sensing area.
In order to satisfy these requirements, this paper introduces the following implementation method for the resident detection method for PIR sensors. First, in order to eliminate PIR sensor malfunctioning due to pets or temperature changes, a Fresnel lens, which allows human infrared waveforms to pass through it while rejecting other waveforms, is installed in front of the PIR sensors. Second, when the output of a PIR sensor exceeds the positive threshold voltage, and this state is maintained for several predefined sampling intervals, that the resident has entered a sensing area. Here, the threshold must be sufficient for the method to distinguish variation in the resident?s infrared from an environmental infrared signal caused by pets or temperature change. Moreover, when the sensor?s output falls below a negative threshold voltage and this status is maintained for several sampling intervals, it is assumed that the resident has left the sensing area. Finally, when the output voltage remains between the two threshold voltages, for example when the resident is not moving inside the sensing area, the output of the corresponding PIR sensor is changed from ?ON? to ?OFF?. At this time, if other sensors installed near this sensor do not
9
output ?ON? signal, the method regards the resident as remaining within the corresponding sensing area.
3.2 Performance evaluation using an experimental test bed
In order to verify the feasibility of the PILAS, an experimental test bed was implemented. Since the intelligent location-based service in the smart home does not require very high location accuracy, we designed the system to have a location accuracy of 0.5 m. Figure 7 shows the experimental test bed in a room measuring 4 × 4 × 2.5 m (width × length × height). In the experiment, twelve PIR sensors were fixed on the ceiling, using the arrangement shown in Fig. 4(c). An Atmel AT89C51CC001 microcontroller [17] was used for signal processing and judging ?ON/OFF?, and a Nippon Ceramic RE431B PIR sensor [18] and NL-11 Fresnel lens were used. Especially, a horn was installed on each PIR sensor to limit the sensing area to the circle with 2 m diameter. Fig. 8 shows the experimental results with the horn. In the figure, the RE431B sensor outputs the signal shown in (a) when a resident passes through the sensing circle, while it outputs the irregular signal shown in (b) when the resident moves within the circle. Finally, no signal is detected when the resident moves outside the circle, as shown in (c). From these experimental results, we verified that the PIR sensor detects residents within the sensing area only. In addition, in order to judge whether the signal is ?ON? or ?OFF?, it is necessary to choose a threshold for the RE431B sensor that considers external environmental disturbance. Initially, several experiments were performed to determine the threshold with respect to the internal temperature change caused by a air conditioner or heater and other disturbances, such as wind or sunshine. Based on these experimental results, when the threshold of the RE431B sensor was ±0.4 V, external environmental temperature change did not affect its performance at detecting the resident. In addition, we verified that pets did not affect the sensing performance with the same threshold.
Fig. 7. Experimental test bed for the PILAS.
10
Fig. 8. Ensuring the exact sensing range with a horn.
Next, in order to determine the resident?s location using the information received from PIR sensors, a PC-based locationrecognition algorithm was implemented, as shown in Fig. 9. Here, a PC collects data from the PIR sensors every 10 msec using an NI 6025E data acquisition (DAQ) board [19]. In the figure, the line in the left window was drawn using a mouse to show the path of the resident graphically, while that in the window on the right is the estimated movement trajectory of the resident drawn by connecting the resident?s locations acquired using the DAQ board.
Finally, in order to verify the efficacy of the system, three experiments were performed with residents between 160 and 180 cm tall, moving at speeds between 1.5 and 2.5 km/h. Figure 9 shows the trajectory of a resident moving along a Tshaped path. The trajectory made by connecting the resi-dent?s locations recognized by the PILAS, shown on the right, was similar to the target path shown on the left. We know that the maximum location error is about 30 cm without compensating for the outer sensors. Fig. 10 shows the trajectory when the resident follows an H-shaped path. In this experiment, the location accuracy was similar to that in Fig. 9. We verified that the system could locate a resident with accuracy of 0.5 m, even if three or more sensors were activated. Figure 11 shows the trajectory of a resident moving along a square path. In this case, the location error is the largest, and the trajectory is not a straight line. We note that serious location errors occurred at each point marked by A due to the inaccurate judgment of the outer sensors. Nevertheless, the location error is still smaller than 0.5 m when moving in the square path. Here, the compensation method for outer sensors, which was explained in Fig. 5, reduces the location error at each point A. When the resident moves in a straight line, as shown in Fig. 12(a), the location error is relatively large without using the compensation method, as shown in Fig. 12(b). However, after applying the compensation method, we verified that the detection results for the areas in the small circles are enhanced by roughly about 30%.
11
4.SUMMARY AND CONCLUSIONS
This paper presents a PIR sensor-based indoor location aware system that estimates the resident?s location for location-based intelligent services in the smart home. This paper introduces the framework of smart home for the location-aware system, and a location-recognition algorithm that integrates the information collected from PIR sensors. In addition, this paper presents a resident-detection method. Finally, an experiment is implemented to evaluate the efficacy of the PILAS.
Based on several experiments conducted under various conditions, we verified that the PILAS can estimates resident?s location sufficiently well. Moreover, because the location accuracy of the system is less than 0.5 m without any terminal for location recognition, the system can be very practical. Furthermore, it should be possible to enhance the location accuracy of the system by increasing the number of sensing areas, by equalizing the sensing areas based on the sensor arrangement, or by compensating for the centers of outer sensors.
Since the location accuracy of this system differs according to the sensor arrangement, it is necessary to determine the optimal sensor arrangement that offers the greatest location accuracy. In order to enhance the location accuracy, it is also necessary to enhance the method of processing the PIR sensors using more advanced techniques such as probabilistic theories and soft computing. Finally, the proposed PILA system should be extended to deal with a room occupied by more than one residents.
12
相关推荐：
loading