Adaptive Ultrasonic Distance Measurement Technique for Handwriting ...
2240
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
Correspondence Adaptive Ultrasonic Distance Measurement Technique for Handwriting Digitization Using Reconfigurable Analog Blocks Chang Hoon Han, Jinyong Chung, and Sang Hoon Hong
Abstract—This paper proposes a low-power dynamically adaptive distance measurement technique for ultrasonic (US) pen digitizers. The technique utilizes a new clock frequency-throttling technique for a reconfigurable analog front-end in implementing a frequency-hopping scheme to minimize power dissipation. As a result, a noise-tolerant distance measurement which is not possible in conventional schemes is realized. Furthermore, the use of reconfigurable analog blocks facilitates wider bandwidth usage of the Piezo film transducers that are used for ultrasonic transmissions and receptions. Index Terms—Frequency hopping, noise reduction, reconfigurable analog block, ultrasonic (US) pen digitizer, voltage throttling.
I. I NTRODUCTION For handwriting digitization, an ultrasonic (US) pen digitizer system has an advantage of larger writing space over the touch screen interface of modern smartphones. The relatively larger writing space provided by the pen digitizer will allow the system to be incorporated into numerous handheld devices [1]. However, as more pen digitizer systems become incorporated into handheld devices, the increased interference noise due to other similarly equipped devices will make it more difficult to accurately track the handwriting of the individual users. Fig. 1 shows the basic operation of the US pen digitizer. For an initial synchronization, a host device transmits an infrared (IR) sync signal to the IR sensor attached to the pen. Subsequently, the pen responds with a US signal output. The response times from the sync signal to the first edge of US signal on both the transmitter and receiver sides can be used to compute the distance, which is proportional to t2-t1. Thus, the distances DL and DR shown in Fig. 1(b) can be obtained from each of the two receiving sensors that are at fixed locations. Since the distance DS between the two sensors is known, the location of the pen can be represented in rectangular coordinates (X, Y) using the Pythagorean theorem for the two right angle triangles shown in Fig. 1(b). However, conventional pen digitizers using this method were found to show poor accuracy in noisy environments that have a frequency that coincides with the resonance frequency of the US transducers such as blow dryer sounds for even a weak noise source. To understand why the weak noise source can affect the measurement,
Fig. 1. (a) Operation of the transmitter (pen) and the receiver (host). (b) Relationship between the distances measured at the host (DL, DR) and the pen position (X, Y).
Fig. 2. Manuscript received January 29, 2010; revised March 20, 2010; accepted March 21, 2010. Date of publication May 10, 2010; date of current version July 14, 2010. This research was supported by the Kyung Hee University Research Fund 2006 (KHU-20060606). The Associate Editor coordinating the review process for this paper was Dr. V. R. Singh. C. H. Han and S. H. Hong are with the Department of Electronics and Radio Engineering, Kyung Hee University, Yongin 446-701, Korea (e-mail: [email protected]; [email protected]). J. Chung is with the School of Information and Communication Engineering, Inha University, Incheon 402-751, Korea (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIM.2010.2047972
Basic mechanism to estimate pen coordinates.
the basic mechanism for distance measurement is illustrated in Fig. 2. The sensed signal is amplified and bandpass-filtered to detect the rising edges of the US wave. The amplification factor depends on the size of the specified writing space. If the writing space is large, the amplification factor should be higher. Unfortunately, higher amplification factor also amplifies noise. The distance is computed by recording the clock cycles spent before the first US wave edges are detected following the IR sync signal. The US wave edges are detected by recording the clock cycles spent between two adjacent edges. Various digital processing
0018-9456/$26.00 © 2010 IEEE
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
Fig. 3.
2241
(a) Characteristics of the Piezo film sensor (receiver). (b) Test setup to evaluate the power efficiency of the proposed adaptive frequency hopping system.
techniques can be used to filter out the noise. For example, determining the likelihood of positive detection and using the relative handwriting speed difference for adjacent sample intervals were proposed to accomplish improved performance [2]. Using an extra sensor at the host to triangulate among three possible combinations of sensor pairs could also improve performance. This additional physical placement or mechanical relocation of the sensors reduces the probability of false detection similar to the approach proposed in [3], which rotates the US transducer to take multiple measurements at predefined positions to reconstruct the final image. Analyzing the surface characteristics of the written area is another possibility using the two sensors attached to the host to recalibrate distance measurement result as studied in [4]. However, the noise reduction capability also depends heavily on the counter size and the clock frequency of the counter because the higher resolution, provided by higher frequency and larger counter size, allows for more accurate signal period detection. Obviously, for low-power operations, the counter size and clock frequency should be as small as possible. Thus, the use of more sophisticated digital processing cannot effectively reduce the error in a low power configuration because the accuracy is already too degraded by the reduced counter size and the clock frequency. Even if power dissipation due to the counter was ignored, the digital processing techniques mentioned above cannot handle the burst or continuous US noise that coincides with the transducer frequency. It is inevitable that these problems will become even more critical if multiple pen digitizers are operating in close proximity. One way to solve this problem is to use a code division multiple access (CDMA) technique [5]. The CDMA technique was not considered, since it is inappropriate for small data communication due to its relatively large overhead.
II. P ROPOSED M ETHOD OF P EN L OCATION M EASUREMENT To improve the measurements while minimizing power consumption, this paper proposes a new adaptive frequency-hopping method utilizing reconfigurable analog blocks in both the transmitter and the receiver. The proposed system starts with a minimum frequency set, F(0), where only one frequency, f0 , is effective, and hence, no hopping occurs during this operation period. If the error rate becomes higher than upper threshold in F(0) operation period due to US noise, a higher frequency, f1 , is added to the current frequency set and F(1) is effective in the next operation period. In general, a frequency set F(i), can be represented as {f0 , f1 , f2 , . . . , fi }, and during the F(i) operation period, (i + 1) frequency components are involved in pseudorandom frequency hopping. With higher number of frequency components, the error rate will decrease. Since it is found that lower frequency operation of the transmitter and receiver can significantly reduce power dissipation, when the error rate is below the lower threshold, the highest hopping frequency from F(i), is removed to minimize the power consumption, and the new frequency set becomes F(i-1). III. R ESULTS In order to determine the feasibility of using reconfigurable analog blocks to perform more accurate measurements, a Piezo film sensor was driven directly by Anadigm’s field programmable analog array (FPAA) AN221E04. The FPAA chip is constructed using switched capacitor circuits. Piezo film sensors were used because they have broadband characteristics [5]. To accommodate frequency-hopping, the test range was set between 40 kHz to 400 kHz. Fig. 3(a) shows the measured responses of the receiving sensor for different distances
2242
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
TABLE I R ECONFIGURATION VALUES OF FPAA
Fig. 5. Demonstration of noise elimination through the proposed frequency hopping technique.
Fig. 4. Power dissipations of transmitter and receiver analog front-end FPAAs.
and driving frequencies. The results show that received signal strength is weaker for longer distances and higher frequencies. It is important to note that for robust edge detection at the receiver side, the output of the analog front-end at higher hopping frequency ranges must be amplified with higher gain for normalized signal strength. However, this amplification increases noise as well as power dissipation. Therefore, using lower hopping frequency ranges are preferable for power savings and noise reduction. To demonstrate the power efficiency of the proposed adaptive frequency hopping method, a test structure comprising two Piezo film sensors, two FPAA boards, and a field programmable gate array (FPGA) board was setup as shown in Fig. 3(b). The FPGA board directly programs both the transmitter and the receiver analog frontend FPAA through a direct digital interface. The transmitter FPAA chip, which consists of an oscillator and a gain component, outputs the US driving signal with 4 V peak-to-peak. The receiving FPAA chip, which consists of a low- and high-pass filter, representing the bandpass filter, is connected to a programmable gain that acts to compensate for the sensitivity drop of the Piezo film sensor in the higher frequency ranges. The reconfiguration parameters are listed in Table I. It is important to note that the key factor in reducing power dissipation was the switching frequency of the switched capacitors. To ensure similar accuracies, the switching frequencies were set to be at least 16 times as large as the maximum input frequency. The FPAA used for this experiment only offered switching frequencies of 2 MHz, 4 MHz, and 16 MHz. Fig. 4 shows the power dissipation differences for various reconfiguration values. The results clearly indicate that low-frequency operation is preferable for low-power operation. If frequency-throttling reconfiguration was not possible, up to 24% more power would be consumed. The maximum dynamic reconfiguration time was 116.5 µs. Since, approximately, a 75-Hz sampling frequency is required for handwriting digitization, the dynamic reconfiguration
does not affect the measurements. In addition, the capability to control the switching frequency more freely will provide more efficient power savings. The ultrasonic noise removal using the proposed frequency-hopping technique was verified in Fig. 5. This test compared the conventional scheme (top traces) with the proposed hopping scheme (bottom traces) in a noisy environment for a fixed distance. The traces show the outputs of the receiver’s analog front-end. The most critical noise is the noise from a similar device in close proximity. Assuming allowable range for the two similar systems is 30 cm, and when the edge detection is activated after the analog front-end output signal crosses over half of the full swing level, it was found that at 100 kHz US frequency, 67 cm distance between two hosts can cause problems. This distance is frequently encountered in a typical office environment. The problem becomes more severe if the writing range is made larger. Therefore, to simulate the office environment, noise of equivalent strength was directly fed in at the analog front-end. While the conventional pen digitizer system was not able to provide the correct distance measurements when the noise (other user’s host) frequency was equivalent to the 100-kHz US frequency of the transmitter, the proposed system, utilizing multiple frequencies, allowed the distance measurements for the same noise conditions. IV. C ONCLUSION The paper proposed a new power-efficient measurement technique, utilizing the reconfigurable analog blocks at both the transmitter and the receiver side, which dynamically reconfigured to obtain noise tolerant measurements. In addition to removing interferences from neighboring devices using the proposed adaptive frequency hopping, the analog front-ends are also concurrently reconfigured with a lowpower consideration by appropriately throttling the switched capacitor frequency of the FPAA for each hopping frequency.
R EFERENCES [1] J. Lee and Y. Chung, “Design of a wireless handwriting input system for mobile devices,” in Proc. IEEE Int. Symp. Consum. Electron., Jun. 2005, pp. 222–225. [2] J. Lee, S. Hong, and Y. Chung, “Robust tracking of ultrasonic pen for handwriting digitizer applications,” IEEE Trans. Consum. Electron., vol. 53, no. 3, pp. 1098–1102, Aug. 2007. [3] A. Fenster, S. Tong, H. N. Cardinal, C. Blake, and D. B. Downey, “Threedimensional ultrasound imaging system for prostate cancer diagnosis and treatment,” IEEE Trans. Instrum. Meas., vol. 1, no. 4, pp. 32–35, Dec. 1998. [4] V. C. Protopappas, D. A. Baga, D. I. Fotiadis, A. C. Likas, A. A. Papachristos, and K. N. Malizos, “An ultrasound wearable system for the monitoring and acceleration of fracture healing in long bones,” IEEE Trans. Biomed. Eng., vol. 52, no. 9, pp. 1597–1608, Sep. 2005. [5] M. Hazas and A. Hopper, “Broadband ultrasonic location systems for improved indoor positioning,” IEEE Trans. Mobile Comput., vol. 5, no. 5, pp. 536–547, May 2006.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
Correspondence Adaptive Ultrasonic Distance Measurement Technique for Handwriting Digitization Using Reconfigurable Analog Blocks Chang Hoon Han, Jinyong Chung, and Sang Hoon Hong
Abstract—This paper proposes a low-power dynamically adaptive distance measurement technique for ultrasonic (US) pen digitizers. The technique utilizes a new clock frequency-throttling technique for a reconfigurable analog front-end in implementing a frequency-hopping scheme to minimize power dissipation. As a result, a noise-tolerant distance measurement which is not possible in conventional schemes is realized. Furthermore, the use of reconfigurable analog blocks facilitates wider bandwidth usage of the Piezo film transducers that are used for ultrasonic transmissions and receptions. Index Terms—Frequency hopping, noise reduction, reconfigurable analog block, ultrasonic (US) pen digitizer, voltage throttling.
I. I NTRODUCTION For handwriting digitization, an ultrasonic (US) pen digitizer system has an advantage of larger writing space over the touch screen interface of modern smartphones. The relatively larger writing space provided by the pen digitizer will allow the system to be incorporated into numerous handheld devices [1]. However, as more pen digitizer systems become incorporated into handheld devices, the increased interference noise due to other similarly equipped devices will make it more difficult to accurately track the handwriting of the individual users. Fig. 1 shows the basic operation of the US pen digitizer. For an initial synchronization, a host device transmits an infrared (IR) sync signal to the IR sensor attached to the pen. Subsequently, the pen responds with a US signal output. The response times from the sync signal to the first edge of US signal on both the transmitter and receiver sides can be used to compute the distance, which is proportional to t2-t1. Thus, the distances DL and DR shown in Fig. 1(b) can be obtained from each of the two receiving sensors that are at fixed locations. Since the distance DS between the two sensors is known, the location of the pen can be represented in rectangular coordinates (X, Y) using the Pythagorean theorem for the two right angle triangles shown in Fig. 1(b). However, conventional pen digitizers using this method were found to show poor accuracy in noisy environments that have a frequency that coincides with the resonance frequency of the US transducers such as blow dryer sounds for even a weak noise source. To understand why the weak noise source can affect the measurement,
Fig. 1. (a) Operation of the transmitter (pen) and the receiver (host). (b) Relationship between the distances measured at the host (DL, DR) and the pen position (X, Y).
Fig. 2. Manuscript received January 29, 2010; revised March 20, 2010; accepted March 21, 2010. Date of publication May 10, 2010; date of current version July 14, 2010. This research was supported by the Kyung Hee University Research Fund 2006 (KHU-20060606). The Associate Editor coordinating the review process for this paper was Dr. V. R. Singh. C. H. Han and S. H. Hong are with the Department of Electronics and Radio Engineering, Kyung Hee University, Yongin 446-701, Korea (e-mail: [email protected]; [email protected]). J. Chung is with the School of Information and Communication Engineering, Inha University, Incheon 402-751, Korea (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIM.2010.2047972
Basic mechanism to estimate pen coordinates.
the basic mechanism for distance measurement is illustrated in Fig. 2. The sensed signal is amplified and bandpass-filtered to detect the rising edges of the US wave. The amplification factor depends on the size of the specified writing space. If the writing space is large, the amplification factor should be higher. Unfortunately, higher amplification factor also amplifies noise. The distance is computed by recording the clock cycles spent before the first US wave edges are detected following the IR sync signal. The US wave edges are detected by recording the clock cycles spent between two adjacent edges. Various digital processing
0018-9456/$26.00 © 2010 IEEE
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
Fig. 3.
2241
(a) Characteristics of the Piezo film sensor (receiver). (b) Test setup to evaluate the power efficiency of the proposed adaptive frequency hopping system.
techniques can be used to filter out the noise. For example, determining the likelihood of positive detection and using the relative handwriting speed difference for adjacent sample intervals were proposed to accomplish improved performance [2]. Using an extra sensor at the host to triangulate among three possible combinations of sensor pairs could also improve performance. This additional physical placement or mechanical relocation of the sensors reduces the probability of false detection similar to the approach proposed in [3], which rotates the US transducer to take multiple measurements at predefined positions to reconstruct the final image. Analyzing the surface characteristics of the written area is another possibility using the two sensors attached to the host to recalibrate distance measurement result as studied in [4]. However, the noise reduction capability also depends heavily on the counter size and the clock frequency of the counter because the higher resolution, provided by higher frequency and larger counter size, allows for more accurate signal period detection. Obviously, for low-power operations, the counter size and clock frequency should be as small as possible. Thus, the use of more sophisticated digital processing cannot effectively reduce the error in a low power configuration because the accuracy is already too degraded by the reduced counter size and the clock frequency. Even if power dissipation due to the counter was ignored, the digital processing techniques mentioned above cannot handle the burst or continuous US noise that coincides with the transducer frequency. It is inevitable that these problems will become even more critical if multiple pen digitizers are operating in close proximity. One way to solve this problem is to use a code division multiple access (CDMA) technique [5]. The CDMA technique was not considered, since it is inappropriate for small data communication due to its relatively large overhead.
II. P ROPOSED M ETHOD OF P EN L OCATION M EASUREMENT To improve the measurements while minimizing power consumption, this paper proposes a new adaptive frequency-hopping method utilizing reconfigurable analog blocks in both the transmitter and the receiver. The proposed system starts with a minimum frequency set, F(0), where only one frequency, f0 , is effective, and hence, no hopping occurs during this operation period. If the error rate becomes higher than upper threshold in F(0) operation period due to US noise, a higher frequency, f1 , is added to the current frequency set and F(1) is effective in the next operation period. In general, a frequency set F(i), can be represented as {f0 , f1 , f2 , . . . , fi }, and during the F(i) operation period, (i + 1) frequency components are involved in pseudorandom frequency hopping. With higher number of frequency components, the error rate will decrease. Since it is found that lower frequency operation of the transmitter and receiver can significantly reduce power dissipation, when the error rate is below the lower threshold, the highest hopping frequency from F(i), is removed to minimize the power consumption, and the new frequency set becomes F(i-1). III. R ESULTS In order to determine the feasibility of using reconfigurable analog blocks to perform more accurate measurements, a Piezo film sensor was driven directly by Anadigm’s field programmable analog array (FPAA) AN221E04. The FPAA chip is constructed using switched capacitor circuits. Piezo film sensors were used because they have broadband characteristics [5]. To accommodate frequency-hopping, the test range was set between 40 kHz to 400 kHz. Fig. 3(a) shows the measured responses of the receiving sensor for different distances
2242
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 59, NO. 8, AUGUST 2010
TABLE I R ECONFIGURATION VALUES OF FPAA
Fig. 5. Demonstration of noise elimination through the proposed frequency hopping technique.
Fig. 4. Power dissipations of transmitter and receiver analog front-end FPAAs.
and driving frequencies. The results show that received signal strength is weaker for longer distances and higher frequencies. It is important to note that for robust edge detection at the receiver side, the output of the analog front-end at higher hopping frequency ranges must be amplified with higher gain for normalized signal strength. However, this amplification increases noise as well as power dissipation. Therefore, using lower hopping frequency ranges are preferable for power savings and noise reduction. To demonstrate the power efficiency of the proposed adaptive frequency hopping method, a test structure comprising two Piezo film sensors, two FPAA boards, and a field programmable gate array (FPGA) board was setup as shown in Fig. 3(b). The FPGA board directly programs both the transmitter and the receiver analog frontend FPAA through a direct digital interface. The transmitter FPAA chip, which consists of an oscillator and a gain component, outputs the US driving signal with 4 V peak-to-peak. The receiving FPAA chip, which consists of a low- and high-pass filter, representing the bandpass filter, is connected to a programmable gain that acts to compensate for the sensitivity drop of the Piezo film sensor in the higher frequency ranges. The reconfiguration parameters are listed in Table I. It is important to note that the key factor in reducing power dissipation was the switching frequency of the switched capacitors. To ensure similar accuracies, the switching frequencies were set to be at least 16 times as large as the maximum input frequency. The FPAA used for this experiment only offered switching frequencies of 2 MHz, 4 MHz, and 16 MHz. Fig. 4 shows the power dissipation differences for various reconfiguration values. The results clearly indicate that low-frequency operation is preferable for low-power operation. If frequency-throttling reconfiguration was not possible, up to 24% more power would be consumed. The maximum dynamic reconfiguration time was 116.5 µs. Since, approximately, a 75-Hz sampling frequency is required for handwriting digitization, the dynamic reconfiguration
does not affect the measurements. In addition, the capability to control the switching frequency more freely will provide more efficient power savings. The ultrasonic noise removal using the proposed frequency-hopping technique was verified in Fig. 5. This test compared the conventional scheme (top traces) with the proposed hopping scheme (bottom traces) in a noisy environment for a fixed distance. The traces show the outputs of the receiver’s analog front-end. The most critical noise is the noise from a similar device in close proximity. Assuming allowable range for the two similar systems is 30 cm, and when the edge detection is activated after the analog front-end output signal crosses over half of the full swing level, it was found that at 100 kHz US frequency, 67 cm distance between two hosts can cause problems. This distance is frequently encountered in a typical office environment. The problem becomes more severe if the writing range is made larger. Therefore, to simulate the office environment, noise of equivalent strength was directly fed in at the analog front-end. While the conventional pen digitizer system was not able to provide the correct distance measurements when the noise (other user’s host) frequency was equivalent to the 100-kHz US frequency of the transmitter, the proposed system, utilizing multiple frequencies, allowed the distance measurements for the same noise conditions. IV. C ONCLUSION The paper proposed a new power-efficient measurement technique, utilizing the reconfigurable analog blocks at both the transmitter and the receiver side, which dynamically reconfigured to obtain noise tolerant measurements. In addition to removing interferences from neighboring devices using the proposed adaptive frequency hopping, the analog front-ends are also concurrently reconfigured with a lowpower consideration by appropriately throttling the switched capacitor frequency of the FPAA for each hopping frequency.
R EFERENCES [1] J. Lee and Y. Chung, “Design of a wireless handwriting input system for mobile devices,” in Proc. IEEE Int. Symp. Consum. Electron., Jun. 2005, pp. 222–225. [2] J. Lee, S. Hong, and Y. Chung, “Robust tracking of ultrasonic pen for handwriting digitizer applications,” IEEE Trans. Consum. Electron., vol. 53, no. 3, pp. 1098–1102, Aug. 2007. [3] A. Fenster, S. Tong, H. N. Cardinal, C. Blake, and D. B. Downey, “Threedimensional ultrasound imaging system for prostate cancer diagnosis and treatment,” IEEE Trans. Instrum. Meas., vol. 1, no. 4, pp. 32–35, Dec. 1998. [4] V. C. Protopappas, D. A. Baga, D. I. Fotiadis, A. C. Likas, A. A. Papachristos, and K. N. Malizos, “An ultrasound wearable system for the monitoring and acceleration of fracture healing in long bones,” IEEE Trans. Biomed. Eng., vol. 52, no. 9, pp. 1597–1608, Sep. 2005. [5] M. Hazas and A. Hopper, “Broadband ultrasonic location systems for improved indoor positioning,” IEEE Trans. Mobile Comput., vol. 5, no. 5, pp. 536–547, May 2006.
