See Page 3 FUNDAMENTAL CHARACTERISTICS AND APPLICATION POTENTIAL et y op Pe ie ty ac Pl s le tic ci So y A ty ie oc s So Pl es ac y et A So y et ci DECEMBER SEPTEMBER ac e Pl es Pl ac S es rti oc A ci Pe A es rti e l e e r Pe e c s i P t s r t o c e y t l A t i A S e y e ic cl le ty pl Pl op op s rti oc So rti le es s Pe Pl e So ac le A So cl A le ie cl s ci a o e A c rti e r t c e et A Pe pl s ci y ie tic s es s rti Pe P cl y r e e P t o t P l y le Pl cl ty es Pl ic ac eo op pl e P s So e ac A l o a P e e p e e l s c p S e s P e r s le op es ci es A le tic oc op la A Pe et Pe rti le ce le ie rti le Pl Pe y S P op P c o s t s P c ac y le oc la So pl op Pl So le l le s a c e e P i A a s l ci c e e s c P e c l e s So ty rti ie ac la et es A s t P P cl rti CORELESS PRINTED CIRCUIT BOARD (PCB) TRANSFORMERS— A ISSN 1049-3654 le es Volume 11, Number 3, Third Quarter 2000 c and Systems S O C I E T Y op E E E Circuits Pe I N E W S L E T T E R A JUNE MARCH THE INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, INC. IEEE Circuits and Systems Society Newsletter Newsletter Homepage–http://www.nd.edu/~stjoseph/newscas Editor-in-Chief Michael K. Sain Electrical Engineering Department University of Notre Dame Notre Dame, IN, USA 46556–5637 Phone: (219) 631– 6538 Fax: (219) 631– 4393 E-mail: jordan@medugorje.ee.nd.edu In This Issue Call for Contributed Articles, The CAS Magazine Is Nearing Launch: page 25 Features Editors Guanrong Chen Rui J. P. de Figueiredo Department of Electronic Engr. Department of Electrical City University of Hong Kong and Computer Engineering Hong Kong, P. R. China University of California, Irvine (on leave from U. Houston) Irvine, CA, USA 92697–2625 Phone: (852) 2788– 7922 Phone: (949) 824– 7043 Fax: (852) 2788– 7791 Fax: (949) 824– 2321 E-mail: gchen@ee.cityu.edu.hk E-mail: rui@uci.edu IEEE Publishing Services Robert Smrek Production Manager IEEE Service Center 445 Hoes Lane P.O. Box 1331 Piscataway, NJ 08855–1331, USA Phone: (732) 562–3944 Frequency of Publication Quarterly Newsletter Deadlines Articles for the CAS Newsletter issues must be received by the Editor by the following dates: Issue First Quarter Second Quarter Third Quarter Fourth Quarter Due Date February 1 May 1 August 1 November 1 © 2000 IEEE. Information contained in this newsletter may be copied without permission provided that the copies are not made or distributed for direct commercial advantage, and the title of the publication and its date appear. IEEE Circuits and Systems Society Newsletter is published quarterly by the Circuits and Systems Society of the Institute of Electrical and Electronics Engineers, Inc., Three Park Avenue, New York, NY 10016–5997. Four dollars per member per year (included in Society fee) for each member of the Circuits and Systems Society. Printed in U.S.A. Periodicals Postage Paid at New York, NY, and at additional mailing offices. Postmaster: Send address changes to IEEE Circuits and Systems Society Newsletter, Attn: Change of Address, IEEE, 445 Hoes Lane, Piscataway, NJ 08855–1331. 2 Bing Sheu CAS President Clear Up Your Understanding of Fuzzy Systems: page 28 Do You Know Someone Outstanding in Their Field? Nominate them for a CAS Award: page 38 Find Out What You Missed at MWSCAS: page 40 See Who Won the Gold at the First CNN Young Researcher Contest: page 43 Articles Hui/Tang/Chung: PCB Transformers ............ Gruhl et al.: 3rd Gen. Wireless Networks ..... Call for Papers in the CAS Magazine ........... Ying: Fuzzy Systems Technology ................. Society CAS Awards Nominations ............................ MWSCAS 2000 Report ................................. CNN Young Researcher Contest 2000 ......... People CAS 2000 IEEE Fellow Profiles .................... Chua Receives Neural Networks Award ....... The Adventures of the ’Umble Ohm ............. Places Calls for Papers and Proposals ICECS ’01 .................................................. SSMSD ’01 ................................................ 2nd Dallas CAS Workshop ........................ Coming Events ISCAS 2001 ............................................... ICECS ’2K ................................................. APCCAS 2000 ........................................... ICCAD 2000 .............................................. CAS Workshop on Emerging Techologies SCS ’01 ..................................................... Analog & Mixed Signal Short Course 2000 3 16 25 28 38 40 43 26 44 44 46 46 46 48 45 45 45 45 46 46 Coreless Printed Circuit Board (PCB) Transformers – Fundamental Characteristics and Application Potential S. Y. (Ron) Hui A bstract – In this article, the fundamental concept, characteristics and application potentials of coreless printed-circuit-board (PCB) transformers are described. Coreless PCB transformers do not have the limitations associated with magnetic cores, such as the frequency limitation, magnetic saturation and core losses. In addition, they eliminate the manual winding process and its associated problems, including labor cost, reliability problems and difficulties in ensuring transformer quality in the manufacturing process. The parameters of the printed windings can be precisely controlled in modern PCB technology. Because of the drastic reduction in the vertical dimension, coreless PCB transformers can achieve high power density and are suitable for applications in which stringent height requirements for the circuits have to be met. A transformer’s power density of 24W/cm2 has been S. C. Tang H. Chung reported in a power conversion application. When used in an isolation amplifier application, coreless PCB transformers tested so far enable the amplifier to achieve a remarkable linear frequency range of 1MHz, which is almost eight times higher than the frequency range of 120 kHz in existing Integrated-Circuit products. PCB materials offer extremely high isolation voltage, typically from 15kV to 40kV, which is much higher than many other isolation means such as optocouplers. It is envisaged that coreless PCB transformers can replace traditional core-based transformers in some industrial applications. Their application potentials deserve more attention and exploration. Introduction The discovery of Faraday’s law of induction is indisputably a corner stone in electrical and electronic engineering. The consequent development of . . . continued on Page 4 3 Figure 1. Photograph of a coreless PCB transformer (right) and a core-based transformer (left). Coreless PCB Transformers … continued from Page 3 electric generators and transformers has made electricity a common form of energy in modern society. These days, transformers are commonly used for electrical isolation and energy and/ or signal transfer. Normally, traditional transformers consist of copper windings wound on magnetic cores. The use of magnetic cores in transformers is usually thought to be essential because the magnetic cores, which are made of ferromagnetic materials, provide good conducting paths for the magnetic flux. The core-based transformer concept has not faced much serious challenge in the past, probably because of the fact that most transformer designs were for low-frequency (50 or 60Hz) operations. Even when the operating frequency in many modern power electronics applications (such as switched mode power supplies) was significantly increased to several hundreds of kilo-Hertz in the 1990’s, the core-based transformer concept remained more or less intact. Although it is well known that the size of the magnetic components decreases with increasing operating frequency, the issue: “When will the size of the magnetic components approach zero and become zero?” was seldom addressed. The main reasons for the use of magnetic cores are primarily to pro4 vide a high degree of magnetic coupling and to reduce the leakage inductance. Transformers formed by using twisted coils without magnetic cores have been proposed [1] for high-frequency applications. In [1], it was demonstrated that the twisted-coil transformers could achieve a coupling factor of 0.8 at about 1 MHz. However, the parameters of twisted coil transformers are difficult to control precisely. In addition, it may not be easy to manufacture identical twisted coil transformers in large quantity with high quality control. Much research effort has been focused on the use of printed planar windings for inductor or transformers [2–10]. The use of printed planar windings not only eliminates the costly manual winding process in traditional transformers but, more importantly, makes it possible to manufacture inductors or transformers with precise parameters in an automated manner. In most of the literature [2– 9], magnetic substrates or materials are still used as parts of the magnetic core structures. An interesting attempt of printing two spiral windings on the same surface of a PCB without using a magnetic core is reported in [10]. In [10], an integral equation analysis method for predicting the parameters of the printed single-sided PCB transformer is presented. In the literature mentioned so far, the planar inductors and transformers are of low output power (typically less than 2W). Except in [1,10], the magnetic designs require the use of 10mm Figure 2. Dimensions of Tr6. magnetic cores in one form or another. In this article, we summarize the recent developments of coreless PCB transformer technology [11– 17]. The misunderstandings that PCB transformers without magnetic cores might have low coupling factor, low voltage gain and high radiated EMI problems are clarified. With the aid of a high-frequency circuit model, the basic characteristics and application examples of coreless PCB transformers are described. In particular, a resonant technique has been incorporated into the use of the proposed coreless PCB transformers so as to achieve a high voltage gain (to overcome the apparent low magnetic coupling) and take advantage of the leakage inductance (to turn the apparent disadvantage into an advantage). Optimal operating techniques for using coreless printed-circuit-board (PCB) transformers under (1) minimum input power conditions and (2) maximum energy efficiency conditions are described. Basic Structure and Equivalent Circuit of Coreless PCB Transformers The basic structure of a coreless PCB transformer is very simple. Essentially, transformer windings are printed on a double-sided PCB. An example of a coreless PCB transformer (right) is shown in Fig. 1, together with a core-based pulse transformer (left). In order to illustrate the characteristic of coreless PCB transformers, a prototype shown in Fig. 2 and labeled as transformer Tr6 is used as an example. The width and height of the copper track are 0.22mm and 0.025mm, respectively. The distance between adjacent tracks is about 0.28mm. The number of turns for the primary and secondary printed windings is 10. The equivalent circuit of a coreless PCB transformer is shown in Fig. 3, where R1 is the primary winding resistance, R'2 is the secondary winding resistance referred to the primary, RL is the resistive load, Llk1 is the primary leakage inductance, L'lk2 is the secondary leakage inductance referred to the primary, LM1 is the primary mutual inductance, C1 is the primary winding capacitance, C'2 is the capacitance in the secondary winding referred to the primary, C12 is the capacitance between primary and secondary windings, and n is the turns ratio. . . . continued on Page 6 Figure 3. Equivalent circuit of the PCB transformer with a parallel capacitive/resistive load. C'12 R1 C'1 L lk1 L'lk2 L M1 R'2 C'2 C'L 2 N RL 5 Coreless Printed Circuit Board (PCB) Transformers Coreless PCB Transformers … continued from Page 5 The no-load resonant frequency of the equivalent circuit is given by fo = 1 2π Leq C eq (1) where Leq = L'lk2 + Llk1 LM1 and Ceq = C'2 + C'12 . (Here C'2 includes the load capacitance.) The parametric values of Tr6 measured at 10MHz are shown in the equivalent circuit in Fig. 4. It is important to note from (1) that the no-load resonant frequency can be changed by connecting an external capacitor C2 across the secondary winding terminals. This feature enables the optimal operating frequency to be chosen for a particular application. For example, if the operating frequency of the coreless PCB transformer is limited to 10MHz, connecting a capacitor C2 of 680pF gives Tr6 a resonant frequency to be approximately 9 MHz. Figure 4. High-frequency Equivalent Circuit Model of Tr6. 3.5 pF 1.4 ohm 304 nH 304 nH 275 nH 6 1.4 ohm Characteristics of Coreless PCB Transformers Based on the equivalent model, the frequency response of Tr6 loaded with a capacitor C2 of 680pF and a dummy load of 2kΩ is shown in Fig. 5a and Fig. 5b. The voltage gain is the ratio of the output voltage to the input voltage (V2/V1). Observation of this typical frequency response leads to the following important points that can be considered to operate the coreless transformer in an optimal manner: (i) As expected, it can be observed that the voltage gain at low operating frequency (less than 200kHz) is very low. As the frequency increases, the voltage gain increases until it reaches its maximum at the resonant frequency. (ii) It is interesting to note that the voltage gain of the coreless PCB transformer can exceed 1.0 in the high-frequency region. This dispels the misunderstanding that coreless PCB transformers have low voltage gain. (iii) The voltage gain of the transformer drops to zero beyond the resonant frequency. Thus the useable frequency range should be below the resonant frequency. (iv) The operating frequency of the coreless transformer should be near but below the resonant frequency. This is the high-frequency end of the useable operating range where the magnetizing reactance is large. Otherwise, the equivalent behaves like a short circuit at low-frequencies. (v) Near the resonance region (just below the resonant frequency), the – Fundamental Characteristics and Application Potential 6 5.5 5 Measured 4.5 4 Calculated 3.5 Gain voltage gain is higher than the rest of the operating range. This is the “partial resonance” region with high gain and small phase shift. One can take advantage of this high-frequency and high-gain region for various applications. For Tr6, the partial resonance region is in the range of 6 MHz –8.5MHz. 3 2.5 2 1.5 1 0.5 0 0.00E+00 2.00E+06 4.00E+06 6.00E+06 8.00E+06 1.00E+07 1.20E+07 1.40E+07 1.60E+07 1.80E+07 2.00E+07 Frequency / Hz Figure 5a. Predicted and measured voltage gain versus operating frequency of Tr6. Transformers for Signal Transfer and Power Transfer Maximum-Impedance Frequency for Signal Transfer Transformers are often used to transfer signals with minimum power involved. One example is the gate drive circuit for power electronic devices such as power mosfets and insulated gate bipolar transistors (IGBTs). The gate drive circuit of the power electronic devices requires the gating signal to be transferred to the gate with a small amount of power involved. In . . . continued on Page 8 Phase Shift (Vs leads Vp) / Degree Figure 5b. Predicted and measured phase shift versus operating frequency of Tr6. 180 160 140 120 Measured 100 Calculated 80 60 40 20 0 -20 -40 -60 -80 -100 -120 -140 -160 -180 0.00E+00 2.00E+06 4.00E+06 6.00E+06 8.00E+06 1.00E+07 1.20E+07 1.40E+07 1.60E+07 1.80E+07 2.00E+07 Frequency / Hz 7 36 m m 400 Input Impedance / Ohm 350 Maximum-impedance Secondary Figure 7. Dimensions of the Transformer Tr9. 300 250 ment of the transformer for signal transfer applications. 200 150 Maximum-Efficiency Frequency for Power Transfer 100 50 0 0.00E+00 2.00E+06 4.00E+06 6.00E+06 8.00E+06 1.00E+07 1.20E+07 1.40E+07 1.60E+07 1.80E+07 2.00E+07 Frequency / Hz Figure 6. Predicted (solid-line) and measured (dotted) input impedance versus operating frequency of Tr6. Coreless PCB Transformers … continued from Page 7 order to minimize the input power of the transformer, one can consider the input impedance characteristic. Based on the transformer circuit model, the input impedance of Tr6 can be determined and is shown in Fig. 6. Observation of these plots leads to the following important points: 1. The magnitude of the input impedance peaks at a frequency (termed “maximum-impedance frequency”) which is within the useable frequency range and is slightly below the resonant frequency. For Tr6, this frequency is about 8 MHz and the impedance is about 150Ω. 2. The voltage gain at this “maximum-impedance frequency” is high (about 1.8). That is, the signal can be enlarged. The use of the partial-resonance technique overcomes the supposed low-gain problem and can make the voltage gain greater than unity. 3. Operating the coreless PCB transformer at or near this frequency would minimize the power require8 Primary Coreless PCB transformers can be used as power transformers for power transfer. Figure 7 shows a prototype (labeled as Tr9) that has been tested for use in a 94W DC-DC power converter [15]. Analysis of the energy efficiency of the equivalent circuit for various resistive loads indicates that the coreless PCB transformers can have a wide frequency range within which a high energy efficiency exceeding 90% can be achieved. A plot of the energy efficiency for Tr9 is shown in Fig. 8. Because the power consumption of the electronics driving the primary winding increases with operating frequency, the optimal operating frequency should be chosen at the low-frequency end of the highefficiency region and should be below the resonant frequency of the transformer circuit. Electromagnetic Field One common misunderstanding about coreless transformers is that there will be a serious radiated EMI problem. For a loop antenna, the radiation is primarily perpendicular to the x-y plane, i.e. θ =π/2. The intrinsic impedance η is 120π or 377Ω in free space. If the operating frequency Efficiency Versus Frequency 100% 90% 80% Efficiency 70% 60% High Efficiency Region 50% 40% RL=50 RL=100 RL=200 30% 20% 10% 0% 0E+00 1E+06 2E+06 3E+06 4E+06 5E+06 6E+06 7E+06 8E+06 9E+06 1E+07 Frequency Figure 8. Efficiency of the PCB transformer using FPC sheet with Cr = 390pF and various resistive loads, RL. is 8MHz, the wavelength λ of the radiated signal is λ= c 3 x 10 8 = = 37 .5m (2) fC 8 x 10 6 where fc is the operating (or carrier) frequency. The time-averaged radiated power (P) of a loop antenna is afc P = 160 π I c 6 2 o 4 (3) The radiated power depends on (i) the current Io (or power of the operation), (ii) the dimension (radius a) of the structure and (iii) the operating frequency fc. The radiated power drastically increases with increasing frequency and the dimension of the radiating structure. According to antenna theory, a good loop radiator should have a radius that is in the order of magntiude close to that of the wavelength of the radiated signal. For the transformer TR6, the radius of the outermost loop is 0.005m. This radius is only 0.13x10-3 of the wavelength λ (37.5m). The term (a/λ)4 is in the order of 10-16. For a current Io = 1A, the radiated power of a single loop antenna with –11 a radius of 5mm is P = 4.86 x 10 W. Therefore, the averaged radiated power of a single loop with a radius of 5mm is negligible. Although the coreless PCB transformer has 10 turns, the radiated power involved and its radiated EMI effects are still too small to be a concern. Therefore, the calculation indicates that the transformer TR6 is an extremely poor transmitting antenna as far as far-field radiation is concerned. By the reciprocity theorem, a poor transmitter is also a poor receiver Figure 9. 3-D Field plot of coreless PCB transformer Tr6. for a signal of a certain wavelength. The 3-D field plot of Tr6 excited at 8MHz is shown in Fig. 9. The magnetic flux essentially concentrates within and near the structure of the transformer. . . . continued on Page 10 9 Figure 10. Modulated gate drive circuit using coreless PCB transformer. 5 ohm 1.1mH 3 2 1 2 0.1uF 3 1N4148 1uF 680pF 1N4148 330pF 1uF 1N4148 3 2 5 6 3.9k 2SC2120 7 4 8 TLC 555 6 3 2 1 1N4148 6.8k APT5040 1uF 4 2SB561 1N4148 Buffer Module Coreless PCB Transformers … continued from Page 9 Some Application Examples Example 1 – Transformer Isolated Gate Drive Circuit with a Wide Frequency Range Isolation is often required between the gate drive circuits and the power electronic circuits so that the low-voltage control electronic circuits are electrically isolated from the power circuits. Aiming at (i) minimizing the input current requirement and (ii) providing a wide range of switching frequency, the isolated gate drive circuit and a test circuit with a resistive-inductive load (5Ω and 1.1mH) are Figure 11a. Measured input (Vin) and output (Vgs) signals of the gate drive shown in Fig. 10. The power circuit at fsw = 1 Hz. MOSFET driven by the proposed gate drive is APT5040, which has voltage and curFigure 11b. Measured input (upper: rent ratings of 500V and 16A Vin 10V/div.), carrier (middle: Vc 25V/div.) and output (Vgs) signals of the gate drive respectively. A capacitor of circuit at fsw = 300 kHz. 680pF is connected across the secondary winding so that the resonant frequency of Tr6 will be set at about 9.1 MHz. The large stabilizing capacitor of 1µF and the DC blocking capacitor of 0.1µF in the secondary circuit will not significantly affect the frequency characteristic of the transformer because they are 10 in series with a diode (1N4148) which has typical capacitance of only a few pico-Farads. A voltage-doubler is included in the secondary circuit in order to boost the gate drive voltage. The carrier frequency fc (i.e. the operating frequency) is set at 8 MHz which is the “maximum-impedance frequency” of Tr6. The output signal of the gate drive (i.e. the gate-source voltage Vgs of the power MOSFET) is controlled by the input signal Vin of the gate drive in the modulation/demodulation process. Thus, the frequency of Vin determines the switching frequency (f sw ) of the power MOSFET. Figure 11a and Fig. 11b show the waveforms of the input gating signal and the output gate drive signal for the power device at 1 Hz and 300kHz, respectively. The carrier signal of 8MHz is also shown in Fig. 11b. The input current of the entire gate drive circuit versus operating frequency is shown in Fig. 12. The minimum input current occurs at the maximumimpedance frequency as predicted. A photograph of the actual test circuit is shown in Fig. 13. The magnetic field of the circuit has been scanned by a Precision EMC scanner and is shown in Fig. 14. Most of the EMI emission comes from the upper lefthand region of the PCB (labeled as region ‘T’). This area contains the copper tracks that form part of the S. C. Tang power circuit. Thus, the major source of EMI emission is the conducting path of the power circuit (upper righthand side of the PCB) rather than the coreless PCB transformer. Even the gate drive circuit on the primary side (region ‘P’) and the gate drive circuit on the secondary side (region ‘S’) have higher EMI emission than the coreless PCB transformer (enclosed in the square box). The EMI emission from the coreless PCB transformer is relatively small compared with that from the power circuit and other electronic circuits in the entire circuit. Unlike the power tracks and the gate drive electronics where sharp voltage and current transients occur, the coreless transformer has some filtering effects and has no sharp rising and falling voltage and current edges. H. Chung Figure 13. Photograph of the top side of the PCB. (Coreless PCB transformer is enclosed by the square box.) P Figure 14. EMI from PCB with both the gate drive circuit and the power circuit turned on. Frequency range: 30MHz to 300MHz. T ▲ S. Y. (Ron) Hui ▼ ▲ . . . continued on Page 12 S Figure 12. Measured input current of the gate drive circuit versus carrier frequency fc under the condition of duty cycle of 1.0 (the worst-case situation). Supply Current / mA 240 220 200 180 160 140 120 100 80 60 40 20 0 5.00E+06 6.00E+06 7.00E+06 8.00E+06 9.00E+06 1.00E+07 1.10E+07 1.10E+07 Frequency / Hz 11 10mm 10mm Figure 15. Two secondary windings (left) and one primary winding (right) of the coreless PCB transformer. Gate Drive Gate Drive Figure 16. Coreless PCB Transformer with two secondary outputs used in complementary gate drive circuits. Figure 17. Measured primary gating signal (V1), Drain-source voltage of the two MOSFETs (Vds1 and Vds2) at 1MHz switching operation. (V1: 10V/div.; Vds1 and Vds2: 20V/div.) Coreless PCB Transformers … continued from Page 11 Although the coreless PCB transformer is placed fairly close to the power circuit, its normal operation is not affected by the EMI from the power circuit [16]. Example 2 – Transformer with Multiple Secondary Windings for Totem-Pole Gate Drives Multiple secondary windings can also be constructed for coreless PCB transformers. This can be done either by printing the two secondary windings on the same side or printing them in different layers in a multiple-layer PCB. Figure 15 shows the winding dimensions of a coreless PCB transformer with two secondary windings. This transformer has been used in two totem-pole gate drive circuits (shown in Fig. 16) which are commonly used in power inverters. Figure 17 shows the practical switching waveforms of power devices at 1 MHz. Example 3 – Isolation Amplifier with 1MHz Bandwidth Commercial isolation amplifiers have frequency bandwidths up to about 120kHz [18]. Because of the absence of the core limitations, coreless PCB transformers offer a much higher bandwidth up to at least 1MHz. Figure 18 shows a typical isolation amplifier circuit. The power transformer and the signal transformer are replaced by their coreless counterparts T1 and T2, respectively (Fig. 19). Figure 20 shows the voltage gain versus operating frequency in this application example. Example 4 – Transformers for Maximum Power Transfer Coreless PCB transformers have been tested for power conversion applications with different rated power outputs from 0.5W to 94W. The transformer Tr9 (Fig. 7) has been tested in a low-profile switched mode power . . . continued on Page 14 12 High Voltage Isolation Barrier POWER +15V DC Power Supply Pulse Generator High Voltage Isolation Barrier INPUT Power Transfer Transformer Drive Circuit OUTPUT Low-Pass Filter Rectifier and Filter T1 Isolated Output Signal Isolated Output Signal Common Demodulator CR1 +VISO1 -VISO1 Power Return Signal Low-Pass Filter Power CR2 Input Signal Transformer Drive Circuit Modulator Input Signal Common +VISO2 -V ISO2 Signal and Power Transfer Rectifier and Filter T2 Figure 18. Block diagram of an isolation amplifier. mm 5 9.7 mm 56 5.8 Figure 19. (a) Shape of T1, (b) Shape of T2. Gain versus Frequency 10 0 Gain (dB) -10 -20 -30 Measured Gain -40 Calculated Gain -50 -60 -70 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 Frequency (Hz) Figure 20. Gain versus Frequency of the Isolation Amplifier Prototype. 13 +120V References SW1 IRF630 SW2 IRF630 S. Y. (Ron) Hui S. C. Tang H. Chung 14 [1] S. Hayano, Y. Nakajima, H. Saotome, and Y. Saito, “A New Type High Frequency Transformer”, IEEE Transactions on Mag0.68 µ F netics, vol. 27, no. 6, November 1991, pp. 2µF 5205–5207. Cr RL VO [2] W. Roshen, “Effect of Finite Thickness of Magnetic Substrate on Planar Inductors”, 0.68µ F IEEE Transactions on Magnetics, vol. 26, no. Figure 21. Circuit schematic of the half-bridge converter. PCB Transformer 1, January 1990, pp. with FPC Sheet 270–275. [3] K. Yamaguchi, S. Ohnuma, T. Imagawa, J. Toriu, H. Matsuki, and K. Murakami, “Characteristics of a Coreless PCB Transformers … continued from Page 13 Thin Film Microtransformer with Circusupply with power output of 94W (Fig. lar Spiral Coils”, IEEE Transactions on Magnetics, vol. 29, no. 5, September 21). A maximum transformer effi1993, pp. 2232–2237. ciency exceeding 95% (Fig. 22) and a [4] M. Mino, T. Yachi, A. Tago, K. maximum converter energy efficiency Yanagisawa, and K. Sakakibara, “A of about 84% have been achieved. New Planar Microtransformer for Use in Micro-Switching Converters”, IEEE Conclusions Transactions on Magnetics, vol. 28, no. 4, July 1992, pp. 1969–1973. In this article, the characteristics [5] C. H. Ahn and M. G. Allen, and some application examples of “Micromachined Planar Inductors on coreless PCB transformers have been Silicon Wafers for MEMS Applicadescribed. Several misunderstandings tions”, IEEE Transactions on Industrial Electronics, vol. 45, no. 6, December of coreless PCB transformers have 1998, pp. 866–875. been clarified. Without the limitations [6] C. R. Sullivan and S. R. Sanders, “Deof the magnetic cores, coreless PCB sign of Microfabricated Transformers transformers offer better performance and Inductors for High-Frequency than their core-based counterparts in Power Conversion”, IEEE Transactions on Power Electronics, vol. 11, no. 2, the high-frequency operating range. 1996, pp. 228–238. Research into coreless PCB transform[7] C. R. Sullivan and S. R. Sanders, “Meaers is still in its early stage. It is envissured Performance of a High-Power-Denaged that coreless PCB transformers sity Microfabricated Transformer in a may find applications in many other DC-DC Converter”, IEEE Technology Update Series: Power Electronics and areas. In particular, the advantages of Applications II, IEEE Press, 1997, pp. coreless PCB transformers make them 104–111 attractive in micro-circuits and in low- [8] Balakrishnan, W. Devereux Palmer, W. profile applications in which stringent Joines, and T. G. Wilson, “The Inducheight requirements have to be met. tance of Planar Structures”, IEEE Power Transformer Efficiency Vs. Switching Frequency 100% 95% 90% 85% RL=30 Ohm RL=50 Ohm RL=100 Ohm RL=200 Ohm 80% 75% 70% 65% 60% 55% 50% 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06 3.5E+06 4.0E+06 Switching Frequency (Hz) Figure 22. Measured efficiency of the PCB transformer operated in the half-bridge power converter. Electronics Specialists Conference Proceedings, 1993, pp. 912–921. [9] W. G. Hurley, M. C. Duffy, S. O’Reilly, and S. C. O’Mathuna, “Impedance Formulas for Planar Magnetic Structures with Spiral Windings”, IEEE PESC Proceedings, 1997, pp. 627–633. [10] I. Marinova, Y. Midorrikawa, S. Hayano, and Y. Saito, “Thin Film Transformer and Its Analysis by Integral Equation Method”, IEEE Transactions on Magnetics, vol. 31, no. 4, July 1995, pp. 2432–2437. [11] S. Y. R. Hui, S. C. Tang, and H. Chung, “Coreless Printed-Circuit-Board (PCB) Transformers for Signal and Energy Transfer”, IEE Electronics Letters, vol. 34, no. 11, May 1998, pp. 1052–1054. [12] S. Y. R. Hui, S. C. Tang, and H. Chung “Coreless PCB-Based Transformers for Power MOSFET/IGBT Gate Drive Circuits”, IEEE Transactions on Power Electronics, vol. 14, no. 3, May 1999, pp. 422–430. [13] S. C. Tang, S. Y. R. Hui, and H. Chung, “Coreless Printed Circuit Board (PCB) Transformers with Multiple Secondary Windings for Complementary Gate Drive Circuits”, IEEE Transactions on Power Electronics, vol. 14, no. 3, May 1999, pp. 431–437. [14] S. Y. R. Hui, S. C. Tang, and H. Chung, “Optimal Operation of Coreless PCB Transformer-Isolated Gate Drive Circuits with Wide Switching Frequency Range,” IEEE Transactions on Power Electronics, vol. 14, no. 3, May 1999, pp. 506–514. [15] S. C. Tang, S. Y. R. Hui, and H. Chung, “Coreless Printed Circuit Board (PCB) . . . continued on Page 47 S. Y. (Ron) Hui received the B.Sc. degree (Hons) at the University of Birmingham, U.K., in 1984, and the D.I.C. and Ph.D. degrees at Imperial College of Science, Technology and Medicine, London, in 1987. He was lecturer in power electronics at the University of Nottingham, U.K., in 1987–90. In 1990 he went to Australia for a lectureship at the University of Technology, Sydney, where he became senior lecturer in 1991. He later joined the University of Sydney and was promoted to reader of electrical engineering and director of the power electronics and drives research group in 1996. Presently he is chair professor of electronic engineering and associate dean of the faculty of science and engineering at the City University of Hong Kong. S. C. Tang received the B. Eng. degree in electronic engineering (with first class honors) from the City University of Hong Kong, Kowloon, Hong Kong, in 1997. He is currently working towards the Ph. D. degree at the City University of Hong Kong. His research interests include coreless PCB transformers, high-frequency magnetics, MOSFET/IGBT gate drive circuits, isolation amplifiers and switched-capacitor converters. Henry Shu-hung Chung received the B.Eng. degree (with first class honors) in electrical engineering from The Hong Kong Polytechnic University in 1991, and the Ph.D. degree in 1994. Since 1995 he has been with the City University of Hong Kong. He is currently associate professor in the Department of Electronic Engineering. His research interests include time- and frequency-domain analysis of power electronic circuits, switchedcapacitor-based converters, random-switching techniques, digital audio amplifiers, and soft-switching converters. He has authored two research book chapters, and over 110 technical papers including 50 refereed journal papers in the current research area. Dr. Chung is presently associate editor of the IEEE Transactions on Circuits and Systems—Part I: Fundamental Theory and Applications. 15 S. Gruhl T. Rachidi TEs U M T S IP Network BS M. Link M. Söllner Real-Time Multimedia Applications over 3rd Generation Wireless Networks by S. Gruhl, T. Rachidi, A. Echihabi, M. Link, and M. Söllner A bstract—It is expected that in 3rd generation wireless networks in general, and in the Universal Mobile Telecommunications System (UMTS) in particular, complex control and transport mechanisms will influence the data communication. The wireless link quality varies for a given application’s data flow, while the application itself adapts to the system and thus influences the control mechanisms. In this paper, we study the effects of such a dy16 namic system on the user-perceived Quality of Service (QoS) for realtime multimedia applications over UMTS, by means of simulating the wireless link. The simulator comprises essential layer 2 and 3 protocol functionality of the UMTS Terrestrial Radio Access Network (UTRAN) for terminal equipment (TE) and base station (BS). It is used to demonstrate in real-time the effects of 3rd generation wireless network access on IPbased multimedia applications. UMTS The Universal Mobile Telecommunication System (UMTS) will extend the services provided by current second-generation systems (GSM, PHS, IS-95, etc.) from simple circuitswitched voice telephony to complex data services ranging from e-mail and web-browsing to voice over packets, media on demand, and video conferencing [1]. Users will be able to interact totally with their Wireless Information Devices to retrieve, store, and process data anywhere, anytime while being on the move. To this end, UMTS will support packet-switched data services for up to: • 144 kbps for high speed mobile users • 384 kbps for low speed mobile users • 2 Mbps for portable/fixed users Packet switched services use the system capacity more efficiently, and allow for user idle time and volume charging policy. The statistical gain of packet switching results from increased link utilization due to non-continuous bandwidth requirements from applications. Such a mechanism is well investigated in wireline networks where medium capacity does not vary. Major differences exist, however, between wireline and wireless networks, because the radio link constitutes a massive bottleneck: • Power limitation, interference and altering radio link conditions due to mobile terminal position cause the link capacity to change rapidly. • Handoff calls lead to additional and unpredictable load in a cell. • High bit error rates (BER) are encountered in wireless communications. Forward error correction (FEC) becomes important and retransmissions are more frequent than in fixed networks. Consequently, data rates are affected. Once deployed, UMTS is expected to interface seamlessly with the wide variety of interactive, media-on-demand and multimedia IP-based applications developed originally for the Internet, like the ITU standard H.323 “Packet Based Multimedia Communications Systems” [12] as depicted in Fig. 1. . . . continued on Page 18 Application Audio/Video Applications Audio/Video Codecs RTC H.323 Data Applications Terminal control and management H.225.0 terminal to gatekeeper signaling H.225.0 call signaling H.245 control T.12x RTP Unreliable transport (e.g. UDP) Reliable transport (e.g. TCP) Network layer (e.g. IP) T.123 Network Link layer (e.g. UMTS) Physical layer (e.g. UMTS) Figure 1. Principal scheme of an Internet multimedia protocol stack, as used in the ITU standard H.323 “Packet Based Multimedia Communications Systems” [12–17]. 17 Relative Traffic Data 250 200 Mobile Voice Fixed Voice 150 100 50 Time Data over Circuits Voice over Packets Figure 2. Estimated relative traffic mix. Wireless Networks … continued from Page 17 The volume of data generated by these applications is expected to grow over-proportional in terms of bandwidth consumption as depicted in Fig. 2. However, due to the above mentioned intrinsic characteristics of the wireless link, it is a challenge to deliver circuit-switched-like Quality of Service (QoS), such as bounded delay and jitter, which are essential for multimedia and interactive applications. The QoS needed for the broad variety of data applications that will be available over UMTS can be specified in terms of several QoS parameters and classes. Table 1 describes these classes. It is critical to investigate UMTS system performance from the perspectives of system efficiency and QoS contract fulfillment. To this end, it is essential to model the wireless subsystem in terms of its capacity and transmission technique. Wideband Code Division Multiple Access (WCDMA) as the transmission technique for UMTS has the following characteristics: • CDMA is a spread spectrum technique developed for military antijam applications. • Wide bandwidth supports high bit Table 1. The four UMTS traffic classes defined by ITU-R [2–3]. 18 Class Number Traffic Class 1 Conversational 2 Class Description Example Relevant QoS Requirements – Preserves time relation between entities making up the stream – Conversational pattern based on human perception – Real-time – Voice over IP – Video conferencing – Low jitter – Low delay Streaming – Preserves time relation between entities making up the stream – Real-time – Real-time video – Low jitter 3 Interactive – Bounded response time – Preserves the payload content – Web browsing – Database retrieval – Round trip delay time – Low BER 4 Background – Preserves the payload content – Email – File transfer – Low BER rates and helps to combat fading in multi-path radio channels. • Many users share the same radio carrier. • Each user is assigned a unique random code different from and approximately orthogonal to other codes. • Quality degrades as the number of users on a channel/carrier increases (interference limited system). WCDMA technology results in soft-capacity behavior. In classical schemes employing a combination of Time Division Multiple Access and Frequency Division Multiple Access (TDMA/FDMA schemes), the total capacity is static (see Fig. 3), and the MultipleAccess DS-CDMA (TD-CDMA) Duplex scheme FDD (TDD) Chip rate 3.84 MChip/s Carrier spacing Flexible in the range 4.6–5.0 MHz (200 kHz carrier raster) Frequency bands Frame length 10 ms; 15 time slots Inter-BS synchronization No accurate synchronization needed (synchronization needed) Multi-rate/ Variable-rate scheme capacity 1920–1980 / 2110–2170 paired (1900–1920 and 2010–2025 unpaired) Variable-spreading factor + Multi-code Spreading factor: 4–256 (1–16) Channel coding scheme Convolutional coding, Turbo coding, rate 1/2–1/3 Packet Dual mode on common and dedicated access channels Table 2. UMTS Key Parameters (source 3GPP). time frequency capacity user #3 user #2 user #1 Interference time 5 MHz frequency 10 ms Figure 3. Capacity in FDMA & TDMA (top) versus W-CDMA (bottom). QoS of an individual link is hardly correlated to other carriers in the cell. In CDMA the whole capacity is limited by the relative signal to noise ratio of the individual links (interference limited system). Adaptive techniques, e.g. power control and admission control are key in providing QoS for each individual service. This naturally imposes the need to model physical layer behavior with regard to individual services. Table 2 lists typical UMTS key parameters to be taken into consideration in the model. To address the specific requirements of services while achieving a high spectral efficiency there are dif. . . continued on Page 20 19 Real-Time Multimedia Applications over 3rd Generation Wireless Networks DTCH Information data Uplink DCCH 2560 Information data 2560 CRC detection 96 96 CRC detection CRC16 CRC16 Tail bit discard 2576 Tail8 112 Termination 12 Turbo Code R=1/3 7740 1st interleaving 7740 Viterbi decoding R=1/3 360 1st interleaving 360 Radio Frame segmentation #1 1935 #2 1935 #3 1935 #4 1935 90 90 90 Rate matching #1 2293 #2 2293 #3 2293 #4 2293 105 105 105 105 2293 107 2293 107 107 2293 90 2293 107 2nd interleaving 2400 2400 2400 2400 slot segmentation 1 240kbps DPDCH 2 1 2 … … Radio frame FN=4N 15 1 2 15 1 2 … … 15 1 2 15 1 2 Radio frame FN=4N+1 … … 15 1 2 15 1 2 Radio frame FN=4N+2 … … 15 15 Radio frame FN=4N+3 Figure 4. Example of how two different services data/voice receive different physical layer processing. Wireless Networks … continued from Page 19 ferent coding schemes applied to user data as shown in Fig. 4. For our investigations we concentrate on the UMTS terrestrial radio access network (UTRAN), see Fig. 5. Thus we explicitly model the UTRAN part with its protocols, see Fig. 6. As we work with the standard IP interface we are able to include possible transit networks in our investigation, by using existing networks as access networks to our demonstrator. These are not modeled but actually coupled with the real-time testbed. The detailed de20 scription of the testbed is out of the scope of this paper, and is presented in [11]. UMTS Protocol Stack The UMTS Radio Interface architecture is layered into a physical layer, a data link layer and a network layer (IP in this work). The data link layer is divided into a Radio Link Control (RLC) sublayer and a Medium Access Control (MAC) sublayer. The Logical Link Control (LLC) sublayer present in many early UMTS proposals is not considered. It is expected that this sublayer will be reduced to a Null-sublayer to minimize protocol overhead and/or will be merged with the PDCP Layer, which is doing header compression. Figure 5 shows protocol termination for UMTS dedicated channel (DCH). The following is a summary of the main services and functions of Layer 1 and Layer 2 [4–6] that have been RLC Sublayer Services: • Transparent data transfer • Acknowledged/unacknowledged data transfer • QoS settings RLC Sublayer Functions: • User data transfer • Segmentation and reassembly Application Services Radio Access Bearer Services TE TAF UE MT NodeB UU UE: User Equipment MT: Mobile Termination CN RNC Transit Network Terminating Network TE IU UTRAN TE: Terminal Equipment CN: Core Network TAF: Terminal Adaption Function Figure 5. Logical network architecture. considered in the system: Physical Layer Services: • Information transfer services to higher layers though Transport Channels Physical Layer Functions: • Error Detection and FEC on transport channels • Multiplexing and de-multiplexing of transport-channels/coded-transport composite channels • Synchronization • Measurements and indication to higher layers MAC Sublayer Services: • Data transfer • Radio resources and MAC parameters reallocation • Measurements reporting MAC Sublayer Functions: • MAC-scheduling and selection of an appropriate transport format. • Multiplexing/de-multiplexing of higher layer PDUs into/from transport frames delivered to/from the physical layer • Traffic volume monitoring and reporting to RRM • Padding • In-sequence/out-of-sequence delivery of higher layer PDU • ARQ—backward error correction and flow control UMTS provides besides other services a reliable RLC mode to be used RLC RLC MAC MAC PHY UE PHY BS SRNC Figure 6. Our model for the protocol in the UTRAN. by non-real-time or interactive applications, based on FEC using various channel coding schemes, radio blocks. As can be seen, most UMTS layer 1 and 2 connectionless services and functions have been considered. Functions for connection establishment on the other hand have been ignored. This assumption is valid for the scope of our . . . continued on Page 22 21 Wireless Networks … continued from Page 21 project, as we are only interested in the dynamic system behavior of an established link. Control Loops Stefan Gruhl Tajje-eddine Rachidi We have identified a set of challenging questions pertaining to wireless network performance in general and to UMTS in particular, which we address through the real-time testbed. These questions are: • How successful does a certain QoS enabling technology such as UMTS perform with a given application? • What will user-perceived QoS be like over a future wireless link? • What are the traffic arrival characteristics of a certain type of future service, e.g., a video real-time session? To illustrate what we consider as protocol performance in this context, Michael Link we reference some studies of dynamic TCP/IP behavior and its performance over erroneous/slow links. TCP has TCP-flow-control as an adaptation mechanism for network congestion situations. It acts on packet loss and adapts its sending rate. For lossy links, as experienced in wireless transmission, this mechanism can show unexpected behavior and perform poorly [7–10]. We view these investigations as one example of undesired control loop interaction in a multi-protocollayer wireless environment. We are interested in control loops in general, as e.g. also found in bandwidth adaptation methods in a scalable video codec. Our approach also allows us to address TCP performance investigations, but is not restricted to this. In the worst case we expect to find oscillation effects that arise due to the various dynamic controls that are applied simultaneously by higher and lower level protocols. These mechanisms are found in lower layers to overcome lossy and variable capacity links, and in higher layers specifically to adapt to changing network capacity due to congestion situations. Michael Söllner Stefan Gruhl received the masters in computer science in 1998 from the Friedrich Alexander University of Erlangen-Nürnberg. He is currently participating in a joint research program between the Department of Computer Architecture and Performance Evaluation Institute of the Friedrich Alexander University and the Global Wireless Systems Research Group at Bell Labs, Lucent Technologies. His research interest is quality of service for packetized data for cellular wireless systems, particularly GPRS and UMTS. Tajje-eddine Rachidi is associate professor of computer science at Alakhawayn University in Ifrane, Morocco. He received his first degree from the Ecole Nationale Supérieure d’Informatique et de Mathématiques Appliquées de Grenoble (ENSIMAG), France, in 1989 specializing in real-time systems, and earned the Ph.D. degree from Essex University, U.K., in 1995. His research interests include 3rd generation wireless networks and computer vision. He consults for various local companies in Internet/Intranet and database technologies. He is cofounder of CoreSoft: a software company developing intelligent components. Michael Link received the diploma in electrical engineering from University Erlangen-Nürnberg, Germany, in 1994. From 1994 to 1999 he worked as research assistant at the Telecommunications Laboratory, University Erlangen-Nürnberg. In September 1999 he joined the Network Technology Group at Lucent Technologies, Nürnberg. His research interests are in channel coding for the transmission of multimedia data in packet switched networks, channel modeling, and wireless multimedia protocols, especially for UMTS. Michael Söllner received the diploma in mathematics in 1979 and the Ph.D degree from the University of Bochum, Germany in 1982. Since 1984 he has held various positions in communication technology in industry, entering mobile communications with Philips in 1990, where he focused on traffic and protocol engineering for mobile radio systems, being also involved in ETSI GSM standardization. He is now responsible for a Bell Labs wireless research group with Lucent Technologies in Nürnberg, Germany. His current main research interests are radio system aspects of 3rd generation wireless services. 22 In particular, we have looked at adaptive video Client Client codecs on top of the Real-time TEs Transport Protocol (RTP), which run their independent control loops. These interact with other adapIP Network tive control loops from a wireless network, in a way comparable to the TCP phenomBS enon. There can Server be many control loops, e.g. power Server control to adjust the transmit power on the radio link, Figure 7: Standard Client/Server IP applications run through the UMTS protocol stack (red). which affects error rates or control performed by the abled system. As we offer a real-time Radio Resource Manager (RRM) on real-application interface, these effects Resource Allocation. These interac- will be accessible for measurements. tions, if not investigated and under- Furthermore, our system will allow demstood, can hinder the many benefits onstrating user perceived Quality of Serexpected from future wireless commu- vice over a simulated UMTS network. For many higher layer protocols nications networks. and applications we lack the full deControl Loops and QoS scription of internal control mechaFuture packet switched networks nisms, but often we find them as will have to incorporate QoS enabling implemented applications. Most applitechniques to address specific applica- cations today communicate via IP traftion requirements. Imagine a guaran- fic. We use this transparent standard teed bandwidth link with a low error IP-interface to transport application data seamlessly in and out of our real rate. Then you will find most of the TCP-flow control problems to be soft- time simulation of a third generation network. Thus standard IP-based apened to a large degree. How does this work for other control loops? What are plications such as FTP, web browsing, the QoS requirements that should be video applications or Microsoft satisfied? How sensitive is this ap- NetmeetingTM can be simulated conproach to wireless intrinsic link varia- currently over the simulated UMTS tion or actions like a handover to an- protocol stack and radio link (see Fig. 7). Our approach links existing appliother BS? cations during run-time with our proWith our approach we will be able to investigate the benefits of a QoS en- totype implementation of a UMTS U M T S . . . continued on Page 24 23 Wireless Networks … continued from Page 23 protocol stack and with our PHY-layer simulator. Thus we can run our investigations without the need to model the application/higher layer part. By using the real application we do not need any traffic models assumptions, but use the real dynamic behavior—including user interaction—for our investigations. Although there are QoS protocols for IP networks, such as the Resource Reservation Protocol (RSVP), these QoS negotiation protocols are rarely found in present applications. Therefore we have developed the concept of individually assigned QoS for individual flows. A flow is distinguished by its sending/receiving port number. This allows us to do investigations with a simulation of QoS enabled networks. The QoS requirements are for example for reliability, delay, jitter, and guaranteed bandwidth. The main advantage of an on-line simulation system is the possibility to demonstrate the user-perceived quality of service. We can investigate the effects of the various control mechanisms, altered radio conditions or a congestion situation in a cell. Conclusions Quality of Service will be highly important to enable new real-time multimedia services for 3rd generation mobile networks. Therefore it is important to investigate and understand the influence of all protocol layers from the physical link to the application. Therefore we have designed a real-time testbed for performance as- sessment of IP-based packet switched and emulated circuit switched multimedia applications over UMTS. Our proposed architecture allows concurrent access for an arbitrary number of application-flows, while we provide a QoS understanding on a per-flow basis even for non-QoS aware applications. This specifically allows investigating user-perceived QoS of today’s multimedia applications over a simulated wireless link. Starting from existing investigations on the performance of TCP over a wireless link we broaden the field of investigations to other higher layer control loops such as those found in adaptive RTP-based video applications, and combine them with simulated third generation wireless specific RRM functionality and wireless lower layer protocols. Acknowledgements We would like to thank Dr. Urs Bernhard and Dr. Jens Mueckenheim for their kind support on all areas of UMTS. References [1] General UMTS information, as found in http://www.3gpp.org/ [2] O. Lataoui, T. Rachidi, L. G. Samuel, S. Gruhl, and Ran Hong Yang, “A QoS Management Architecture in Packet Switched Mobile Systems”, submitted to INTEROP’00, IEEE, 2000. [3] Technical Specification Group Services and System Aspects, QoS Concept, Technical Report 3G TR 23.907 version 1.1.0., 1999. [4] C. Robool, P. Beming, J. Lundsjo, and M. Johansson, “A Proposal for an RLC/MAC Protocol for Wideband CDMA Capable of Handling Real-Time and Non Real-Time Services”, VTC’98, pp. 107–111. . . . continued on Page 47 24 NEW ARTICLES The IEEE Circuits and Systems Society Newsletter will become the IEEE Circuits and Systems Society Magazine in January, 2001. Style Considerations 1) Articles are readable by the entire CAS membership. 2) Articles are about eight published pages in length. We can, however, accommodate longer or shorter items if the situation seems appropriate. 3) Articles communicate primarily by graphs, diagrams, and pictures. Many authors have begun the inevitable transition to color, as may be seen in back issues of the Newsletter, available at www.nd.edu/ ~stjoseph/newscas/. In issue layout, the editors often build upon the colors chosen by the authors. 4) Equations are to be used sparingly, except for special situations. When equations are present, they may receive special graphical design treatment.* Submission Information 1) File format for diagrams, figures, graphs, and photos is .eps format. If another format is needed, permission should be obtained from the editor at sain.1@nd.edu. Such files should be provided separately from the text. If an embedded document is submitted, say for ease of review, then accompanying text and separate .eps files should also be provided. 2) Abstracts, or complete papers, are submitted to the Features Co-Editors, who will review them to determine if final papers will be a good fit. All papers are subject to review and requests for revision. Features Editors Guanrong Chen Department of Electronic Engr. City University of Hong Kong Hong Kong, P. R. China (on leave from U. Houston) Phone: (852) 2788– 7922 Fax: (852) 2788– 7791 E-mail: gchen@ee.cityu.edu.hk Rui J. P. de Figueiredo Department of Electrical and Computer Engineering University of California, Irvine Irvine, CA, USA 92697–2625 Phone: (949) 824–7043 Fax: (949) 824–2321 E-mail: rui@uci.edu * Note that December 1999 is an exception to the equation restriction. Future exceptions are likely to be rare. C A L L F O R A R T I C L E S 25 Pe op le Pe op le P e Pe op e Pe Pe op le op le Pe Pe Pe op le eo pl e Pe op le op le Pe Pe op le op le P Pe op Pe le Pe op op le le Pe Pe op op le Pe le op op le le Pe Pe op op Pe Pe le op op le le Pe Pe op op Pe le Pe le op op le le Pe Pe o Pe op Pe le op le IEEE CAS FELLOW PROFILES 2000 Weiping Li For contributions to image and video coding algorithms, standards, and implementations. 123456789012 123456789012 Weiping 123456789012 123456789012 Li Weiping Li received the B.S. degree from the University of Science and Technology of China (USTC) in 1982, and the M.S. and Ph.D. degrees from Stanford University in 1983 and 1988, respectively, all in electrical engineering. In 1987, he joined the faculty at Lehigh University, where he is professor in the Department of Electrical Engineering and Computer Science. Since 1998, he has taken a leave from Lehigh University to work on network streaming video in Silicon Valley, California. Weiping Li is currently editor-in-chief of the IEEE Transactions on Circuits and Systems for Video Technology. He served as associate editor of the same journal from 1995 to 1999. He was one of the guest editors for a special issue of the IEEE Proceedings on image and video compression in February, 1995. In 1998–1999, he served as past-chair, and from 1996–1998, chair of the Technical Committee on Visual Signal Processing and Communications of the IEEE Circuits and Systems Society. He is program chair of the MPEG-4 Workshop and Exhibition. He was co-chair in 1999 and chair in 1997 of the Technical Track on Multimedia and Communications at the IEEE International Symposium on Circuits and Systems. He served as chair of the Best Student Paper Award Committee for the 1999 SPIE Visual Communications and Image Processing Conference. From 1997 to 1998, he served as chair of the working group on reaffirmation of IEEE Standard 1180 (Specifications for the Implementation of 8X8 Inverse Discrete Cosine Transform). Since 1995, he has been a member of the Moving Picture Experts Group (MPEG) of the International Standard Organization (ISO). Weiping Li received the Spira Award for Excellence in Teaching in 1992 at Lehigh University and the Guo Mo-Ruo Prize for Outstanding Student in 1980 at University of Science and Technology of China. Norbert J. Fliege For contributions to analog and digital signal processing, and to engineering education. 123456789012 123456789012 Norbert J. 123456789012 123456789012 Fliege 26 Norbert J. Fliege received the Dipl.-Ing. degree and the Dr.-Ing. degree in 1971, both from the University of Karlsruhe, Germany. Since 1978, he has been associate professor at the same university. In 1980, he was visiting professor at ESIEE in Paris. From 1982 to 1996, he was full professor and head of the Telecommunication Institute at Hamburg University of Technology in Hamburg, Germany. Since 1996, he has been full professor of electrical engineering and computer technology at the University of Mannheim, Germany. Since 1968, Dr. Fliege has been engaged in research work in such fields as active filters, digital filters, communication circuits and software, digital audio, and multirate digi- tal signal processing. In addition, he served as department chairman and head of a research center. He has also founded a company providing telecommunication equipment. Dr. Fliege has published approximately one hundred papers, most of them in international magazines and conference proceedings, and four books, one of them with the title Multirate Digital Signal Processing, John Wiley and Sons, 1994. Dr. Fliege is a senior member of IEEE, a member of EURASIP, and a member of VDE (Germany). He has received several national and international awards. In 1997, he was given the honorary doctorate from the University of Rostock, Germany. Pe Pe op Pe le Pe op op le le Pe Pe op op le Pe le op op le Pe le Pe op P le Pe eo op pl l e Pe Pe e op op le Pe le op op le le Pe Pe o Pe op le eo pl e Pe op le pl e P Pe o o Pe op le Pe op le P e Pe op le op le Pe Pe op le op pl le Pe Pe op le p op Pe Pe o CIRCUITS AND SYSTEMS SOCIETY MEMBERS Henry Samueli For contributions to VLSI architectures and realizations for high-bit rate digital communication systems. Henry Samueli was born in Buffalo, New York, on September 20, 1954. He received the B.S., M.S., and Ph.D. degrees in electrical engineering from the University of California, Los Angeles (UCLA) in 1975, 1976, and 1980, respectively. From 1980 to 1985 he was with TRW, Inc., Redondo Beach, California, where he was section manager in the Digital Processing Laboratory of the Electronics and Technology Division. His group was involved in the hardware design and development of military satellite and digital radio communication systems. From 1980 to 1985 he was also part-time instructor in the Electrical Engineering Department at UCLA. In 1985 he joined UCLA full-time and is currently professor in the Electrical Engineering Department. His research interests are in the areas of digital signal processing, communications systems engineering, and CMOS integrated circuit design for applications in high-speed data transmission systems. In 1988 he co- founded PairGain Technologies, Inc., Tustin, California, a telecommunications equipment manufacturer; and in 1991 he co-founded Broadcom Corporation, Irvine, California, an integrated circuit supplier to the broadband communications industry. Since 1995 he has been on leave of absence from UCLA while serving fulltime as chief technical officer of Broadcom where he is responsible for all research and development activities for the company. Dr. Samueli was the recipient of the 1988/1989 TRW Excellence in Teaching Award of the UCLA School of Engineering and Applied Science, the Meritorious Paper Award of the 1991 Government Microcircuit Applications Conference, the 1995 Best Paper Award from the IEEE Journal of Solid-State Circuits, and the Jack Kilby Best Paper Award from the 2000 IEEE International Solid-State Circuits Conference. He received the 1999 Engineer of the Year Award from the Orange County Section of the IEEE. 1234567890123 1234567890123 Henry 1234567890123 1234567890123 Samueli Peter Feldmann For contributions to the analysis and simulation of electronic circuits. Peter Feldmann was born in Timisoara, Romania. He began his university studies at the Polytechnic Institute in Bucharest, Romania; and after two years he continued at the Technion, in Haifa, Israel. There he received the B.Sc. degree, summa cum laude, in computer engineering, in 1983, and the M.Sc. degree in electrical engineering in 1987. From 1985 through 1987 he worked for Zoran Microelectronics in Haifa, Israel, on the design of digi- tal signal processors. Feldmann continued his graduate studies at Carnegie Mellon in Pittsburgh, Pennsylvania, and obtained the Ph.D. degree in 1991. Currently, he is a distinguished member of the technical staff at Bell Labs in Murray Hill, New Jersey. In 1995 he was an adjunct professor at Columbia University in New York. His main research interests are simulation, analysis and design of electronic circuits and communication systems. 1234567890123 1234567890123 Peter 1234567890123 1234567890123 Feldmann 27 Fuzzy Systems Technology: A Brief Overview A bstract—In this article, the basic concepts of fuzzy sets and fuzzy logic are first introduced. The results of industrial and biomedical applications of fuzzy systems technology in the areas of image/signal processing, automation, and control are then briefly reviewed. Finally, a summary of latest theoretical advances on fuzzy control, fuzzy modeling and fuzzy approximation is provided. Recognized by the IEEE as one of the emerging information processing technologies, fuzzy systems technology has achieved, especially in the last several years, widespread applications around the globe in many industries and technical fields, ranging from control, automation, and AI to image/signal processing and pattern recognition. The basis of the technology is a fuzzy set that is an extension of the Membership 1 Young Figure 1. One possible description of vague concept “young” by classical sets. Age (year) 0 35 classical set. In traditional set theory, membership of an object belonging to a set can only be one of the two values: 0 or 1. An object either completely belongs to a set or does not at all. No 28 partial membership is allowed. Crisp sets handle black-and-white concepts, such as “chairs”, “ships”, and “trees”, where little ambiguity exists. Nevertheless, in our daily lives, there exist countless vague concepts that we humans can easily describe, understand and communicate with each other; but conventional mathematics, including the set theory, fails to handle in a rational way. The concept “young” is an example to the point. For any specific person, age is precise. However, relating a particular age to “young” involves fuzziness and is sometimes difficult. What age is young and what age is not? The nature of questions like this is deterministic, not stochastic. A hypothetical crisp set “young” is given in Fig. 1. A fundamental problem is the abrupt change of the membership value from 1 to 0 at a certain age (35 in this case), which means that a 34.9 years old person is completely young whereas a 35.1 years old person is not young at all. To address issues like this, fuzzy set theory generalizes 0 and 1 membership values of a crisp set to a membership function of a fuzzy set. Using the theory, one relates an age to “young” with a membership value ranging from 0 to 1; 0 means no association at all, 1 indicates complete association, and any number in between means partial association. One possible fuzzy set “young” is provided in Fig. 2. One sees that every age is “young” to a degree (equivalently, everyone is “young” to a certain extent). by Hao Ying Of course, no standard fuzzy set “young” exists. People have different views on the same (vague) concept. Fuzzy sets can be used easily to accommodate this reality. Continue the age example. Some people might think age 50 is “young” with membership value as high as 0.9 whereas others might consider that 20 is “young” with membership value merely 0.2. Different fuzzy sets can be used to represent these different versions of “young.” Not only do different people have different fuzzy sets for the same concept, even for the same person, the fuzzy set for “young” can be different when the context in which age is addressed varies. For instance, a 40 years old president of a country would likely to be regarded young whereas a 40 years old athlete would not be considered young. Two different fuzzy sets “young” are needed to deal with the two situations. In classical set theory, there are binary logic operators AND (i.e., intersection), OR (i.e., union), and NOT (i.e., complement). The corresponding fuzzy logic operators exist in fuzzy sets theory. Fuzzy logic AND and OR operations are widely used in fuzzy systems. Unlike the binary AND and OR operators whose operations are uniquely defined, their fuzzy counterparts are non-unique. Numerous fuzzy logic AND operators and OR operators have been proposed; some of them purely from a mathematics point of view. To a large extent, only Zadeh fuzzy AND and OR operators are most useful. Suppose that a specific age, say 30, is “young” (a fuzzy set) with a membership value of 0.8 and is “old” (another fuzzy set) with a membership value of 0.3. Then, the membership value for the age being “young and Membership 1 Figure 2. One possible description of vague concept “young” by fuzzy sets. Young Age (year) 0 10 30 50 70 90 old” (a newly formed fuzzy set) is 0.3 if the Zadeh fuzzy AND operator is used. The membership value for the age being “young or old” (another newly formed fuzzy set) is 0.8 if the Zadeh fuzzy OR operator is utilized. From these simple examples, one sees that (1) fuzzy sets can practically, quantitatively and flexibly represent vague concepts, and (2) this representation is mathematically precise. Fuzzy sets and fuzzy logic form the foundations for fuzzy mathematics, which may be viewed as an extension of the traditional mathematics. The new branches include fuzzy algebra, . . . continued on Page 30 29 Fuzzy logic operators Input variables x1 x2 : Fuzzy inference method Fuzzy rule base Fuzzy sets Fuzzification Fuzzy reasoning Defuzzifier Defuzzification : Output variables y1 y2 yN xM Figure 3. Structure of a typical rule-based MIMO fuzzy system, such as a fuzzy controller or a fuzzy model. Fuzzy Systems Overview … continued from Page 29 fuzzy calculus, fuzzy differential equations, fuzzy graphs, fuzzy topology, and fuzzy spaces, to name just a few. The extension is often technically dif- Fuzzy sets and fuzzy logic form the foundations for fuzzy mathematics, which may be viewed as an extension of the traditional mathematics. The new branches include fuzzy algebra, fuzzy calculus, fuzzy differential equations, fuzzy graphs, fuzzy topology, and fuzzy spaces, to name just a few. The extension is often technically difficult. ficult. For instance, solving a fuzzy quadratic algebraic equation ax2 + bx + c = 0 or fuzzy first-order differential equation ax' + bx + c = 0, where a, b, and c are fuzzy sets, is more challenging than their crisp counterparts. The answers depend on the fuzzy sets a, b, 30 and c, and may not exist or be unique. Although a vast volume of theoretical results has been reported in the literature, the majority of them are pure mathematics and have yet to find useful practical applications. Nevertheless, a small part of fuzzy mathematics has proved to be very useful for the creation and development of fuzzy systems technology. Loosely speaking, any system that uses fuzzy mathematics may be viewed as a fuzzy system. More strictly, fuzzy systems may be divided into two groups: pure fuzzy systems that are developed entirely on the basis of fuzzy mathematics, and hybrid fuzzy systems that are constructed by a mixture of fuzzy mathematics and non-fuzzy techniques. Even pure fuzzy systems only use a small portion of the fuzzy mathematics available; this portion is also mathematically quite simple and conceptually easy to understand. Most of the (rule-based) pure fuzzy systems, including all the fuzzy controllers and fuzzy models, accomplish their objectives by periodically executing the following three steps: fuzzification, fuzzy reasoning, and defuzzification. A typical MIMO (multiple-input multipleoutput) fuzzy system of this category is depicted in Fig. 3. ler, which is a typical rule-based fuzzy system. We assume that the controller is simple and only has two input variables and one output variable. To put Regardless of the type, pure or hybrid, the role that fuzzy mathematics plays in a fuzzy system can ultimately be viewed as a practical, simple and intuitive way to incorporate nonlinear characteristics/effects to the system so that it can outperform its linear counterpart. Simply put, fuzzy systems are just nonfuzzy systems with some intrinsic, peculiar and advantageous nonlinear features. More specifically, fuzzy systems are nonlinear systems with variable gains/coefficients constantly changing with the input state of the systems. It is these variable gains/coefficients that enable the fuzzy systems to perform better, if properly designed. We now graphically illustrate this important point using a fuzzy control- Simply put, fuzzy systems are just nonfuzzy systems with some intrinsic, peculiar and advantageous nonlinear features. the demonstration in the perspective of fixed-gain systems, we first show a constant gain (Fig. 4a) and the resulting output of a simple linear system (Fig. 4b). Figures 5a and 5b give, re. . . continued on Page 32 (a) (b) 8 6 Gain 4 2 0 -10 5 10 Output 5 0 5 Input 1 -5 0 -5 0 0 Input 2 -5 Input 2 -5 10 0 -5 -10 Input 1 5 5 10 -10 10 -10 Figure 4. (a) An example of constant gain of a simple linear system/controller; (b) the resulting system output. (a) (b) 2 Gain 1.75 1.5 1.25 2 1 0 Input 2 -4 -2 10 5 0 -5 -10 -10 10 5 0 -5 -1 0 Input 1 Output 2 4 -2 Input 1 Input 2 -5 0 5 10 -10 Figure 5. An example simple fuzzy system/controller: (a) its variable gain around the origin; (b) its output in the entire input space. 31 (a) Figure 6. Three more examples of nonlinear gain variation characteristics. They are generated by another simple fuzzy system/controller at three different sets of parameter values. 1 0.75 Gain 0.5 0.25 0 -10 10 5 0 -5 Input 2 -5 0 Input 1 Fuzzy Systems Overview … continued from Page 31 5 10 -10 (b) 1 0.75 Gain 0.5 0.25 0 -10 10 5 0 -5 -5 0 Input 1 Input 2 5 10 -10 (c) 1 0.75 Gain 0.5 0.25 0 -10 10 5 0 -5 -5 0 Input 1 5 10 -10 32 Input 2 spectively, the variable gain for the input state around the origin region and the output of the fuzzy controller in the entire input space. Clearly, the variable gain causes the system output to be nonlinear. It can be proved that the gain variation leads to better performance of the fuzzy controller relative to its linear counterpart, which is a linear PI (proportional-integral) controller. The more detailed coverage however is beyond the scope of this short article [1]. Figures. 6a to 6c provide more examples of nonlinear gain variation characteristics, all of which are inherently built into a different fuzzy controller at three different parameter settings. By changing the parameter values, an infinite number of different gain variations can be generated. Many commercial products have been produced by fuzzy systems technology, especially in the past 10 years or so [3]. The technology has been used to enhance the processing of digital images and signals. The results include the autofocus system for Canon cameras, and the autofocus, autoexposure, and autozoom systems for Minolta cameras. Fuzzy logic has also made Sanyo and Canon camcorders better by making their autofocus, autoexposure, and auto-white-balancing systems more intelligent. Other examples include the image stabilizer for camcorders from Matsushita, and TV sets made by Sanyo. In the latter case, fuzzy inference was used to im- Fuzzy Systems Technology: A Brief Overview prove the image quality. In addition, fuzzy systems technology has improved the electrophotography process of photocopying machines from Canon and Ricoh, and boosted the image quality of Sanyo copies by improving their toner supply control and Matsushita copies by introducing more precise autoexposure and toner control. Other successful applications include voice recognition, and handwritten language recognition. Fuzzy control is the most successful and active branch of fuzzy systems technology, in terms of both theoretical research and practical applications. The primary thrust of this novel control paradigm, created in the early 1970s, is to utilize a human control operator’s knowledge and experience to intuitively construct controllers so that the resulting controllers are able to emulate human control behavior to a certain extent. Compared to the traditional control paradigm, the advantages of fuzzy control paradigms are twofold. First, a mathematical model of the system to be controlled is not required (an impossible assumption for most, if not all, other modern control methodologies); and (2) a satisfactory nonlinear controller can often be developed empirically without complicated mathematics. The core value of these advantages is the practicality, leading to less system development time and cost. Industrial automation and commercial products have been successfully developed worldwide using fuzzy control. And Japan has led the way. Its success includes Hitachi’s automated train operation for the Sendai subway Industrial automation and commercial products have been successfully developed worldwide using fuzzy control. And Japan has led the way. Its success includes Hitachi’s automated train operation for the Sendai subway system in Japan that has been in daily operation since about 1987. The trains, controlled by a fuzzy predictive controller, consume less electric energy, and ride more comfortably than the ones controlled by the nonfuzzy controllers. system in Japan that has been in daily operation since about 1987. The trains, controlled by a fuzzy predictive controller, consume less electric energy, and ride more comfortably than the ones controlled by nonfuzzy controllers. Another Hitachi product is the group fuzzy control operation for elevators. The waiting time and idle . . . continued on Page 34 33 Fuzzy Systems Overview … continued from Page 33 time of the elevators are reduced during the rush hour; and riding and stopping are smoother. Fuzzy control started to be built into consumer products around the turn of the 1990s. Home electronics/ appliance products include fuzzy controlled rice cookers, vacuum cleaners, A major application area has been in the automotive industry, where most of the major automobile manufacturers in the U.S., Japan and Europe have actively been pursuing the concept. Fuzzy control has been used to control the engine system, automatic transmission system, suspension system, antilock brake system, and climate system. These systems are either regulated individually or jointly to make the vehicles better, more efficient and safer. washing machines, and home climate control systems. A major application area has been in the automotive indus34 try, where most of the major automobile manufacturers in the U.S., Japan, and Europe have actively been pursuing the concept. Fuzzy control has been used to control the engine system, automatic transmission system, suspension system, anti-lock brake system, and climate system. These systems are either regulated individually or jointly to make the vehicles better, more efficient and safer. While fuzzy systems technology has been applied to many different industries, its larger impact is expected to be on the healthcare industry. At present, however, the number of biomedical applications is relatively small, partially due to the inherent complexity and uncertainty of the systems as well as the risks involved. Biomedicine is much more an art than a science in that human knowledge, experience and skills play a vital role in the diagnosis and treatment of diseases. Biomedical systems are the most difficult systems to control because they are intrinsically nonlinear, time-varying and have time delay. In the late 1980s, a real-time fuzzy control drug delivery system was successfully developed and clinically implemented to regulate blood pressure in postsurgical open-heart patients at the Cardiac Surgical Intensive Care Unit [2]. This is the world’s first realtime fuzzy control in medicine. Fuzzy systems have also been applied to control of muscle immobility and hypertension during general anesthesia, assessment of cardiovascular dynamics during ventricular assistance, diagnosis of artery lesions and coronary stenosis, support for seriodiagnosis, intelligent medical alarms, and multineuron studies. Other successful medical applications are the detection of coronary artery disease, classification of tissue and structure in electrocardiograms, and classification of normal and cancerous tissues in brain magnetic resonance images. Fuzzy systems can be realized in a number of different ways. In many commercial products, they are embedded systems implemented via generalpurpose microcontrollers, such as those made by Motorola, or via dedicated fuzzy logic/inference processors that are special VLSI chips. They can also be integrated with other hardware components of the products. A number of fuzzy system development software packages are on the market to facilitate the development of fuzzy systems, especially fuzzy controllers. They include MATLAB Fuzzy Logic Toolbox TM, Mathematica Fuzzy LogicTM, SieFuzzyTM, fuzzyTech TM, TILShell TM, FIDE TM, RT/Fuzzy TM, Fuzzy Knowledge Builder TM, and Fuzz-CTM. These packages provide friendly graphical user interfaces to make the fuzzy system development easier and more efficient. Better yet, Fuzzy systems have also been applied to control of muscle immobility and hypertension during general anesthesia, assessment of cardiovascular dynamics during ventricular assistance, diagnosis of artery lesions and coronary stenosis, support for seriodiagnosis, intelligent medical alarms, and multineuron studies. once a fuzzy system is completed and needs to be deployed, some of the packages can automatically generate optimized assembly code, C code, or microcontroller code to be directly downloaded to the target hardware. Theory of fuzzy systems has advanced significantly along with the rapid success of the practical applications. Most fuzzy controllers are used as black-box controllers in that their internal mathematical structures are unknown. Since the late 1980s, significant progress has been made to mathematically explore the structure of various types of fuzzy controllers [1]. Fuzzy control has been related to PID control, sliding mode control, adaptive . . . continued on Page 36 Hao Ying is associate professor in the Department of Electrical and Computer Engineering, Wayne State University, after leaving the faculty of the University of Texas Medical Branch at Galveston. He is also advisory professor of Dong Hua University, Shanghai, China. He received the B.S. and M.S. degrees in electrical engineering from Dong Hua University in 1982 and 1984, respectively, and the Ph.D. degree in biomedical engineering from the University of Alabama at Birmingham in 1990. He holds one U.S. patent and has published one research monograph and 44 journal papers. He served as program chair for a major international intelligent systems conference in 1994, was publication chair for the 2000 IEEE International Conference on Fuzzy Systems, and has been program committee member for many international conferences. He is guest editor for three journals and has been invited to review papers for 24 journals, including six different IEEE Transactions. 35 Fuzzy modeling is a new modeling paradigm, and fuzzy models are nonlinear dynamic models. Compared with the conventional blackbox modeling techniques that can only utilize numerical data, the uniqueness of a fuzzy modeling approach lies in its ability to utilize both qualitative and quantitative information.... Qualitative information is human modeling expertise and knowledge.... The expertise and knowledge are actually nonlinear structures of physical systems, and the structures are represented in an implicit and linguistic form instead of an explicit and analytical form. Fuzzy Systems Overview … continued from Page 35 control, relay control, etc. in conventional control, resulting in insightful understanding of fuzzy control in the context of classical control. The results have also been used to analyze some important aspects of fuzzy control systems, including system stability and control performance, and to better design fuzzy controllers. System modeling and system control are two closely related problems. A common form of the system model is a differential equation for continuous-time systems or a difference equation for discrete-time systems. Nonlinear systems are complex and, worse 36 yet, no general theory exists for modeling them. Different nonlinear system modeling techniques have been developed, including the Volterra and Wiener theories of nonlinear systems. Such nonlinear system models are called black-box models because they only attempt to mimic a system’s input-output relationship with the measurement data and hence cannot provide any insight on internal structure of the system. Nonlinear system modeling is complicated because there exist an infinitive number of possible model structures. Correctly assuming a nonlinear model structure is a very hard problem. Fuzzy modeling is a new modeling paradigm, and fuzzy models are nonlinear dynamic models. Compared with the conventional black-box modeling techniques that can only utilize numerical data, the uniqueness of a fuzzy modeling approach lies in its ability to utilize both qualitative and quantitative information. This advantage is practically important and even crucial in many circumstances. Qualitative information is human modeling expertise and knowledge, which are captured and utilized in the form of fuzzy sets, fuzzy logic and fuzzy rules. The expertise and knowledge are actually nonlinear structures of physical systems, and the structures are represented in an implicit and linguistic form instead of an explicit and analytical form. Fuzzy models are often intuitive because fuzzy sets, fuzzy logic and fuzzy rules are intuitive and meaningful. However, fuzzy models are not as simple as those models that can be expressed in mathematical formulas. In general, fuzzy models are black-box models. Nevertheless, under certain conditions, analytical structure of some fuzzy models can be derived. When this is the case, the fuzzy mod- els are no long black-boxes. Recent theoretical analysis shows that some fuzzy models are actually nonlinear ARX (Auto Regressive with eXtra input) models with variable model parameters. A system capable of uniformly approximating any continuous function is called either a functional approximator or a universal approximator. The issue of universal approximation is crucial to fuzzy systems. In the context of control, the question is whether a fuzzy controller can always be constructed to uniformly approximate any desired continuous, nonlinear control solution with enough accuracy. For modeling, the question is whether a fuzzy model can always be established which is capable of uniformly approximating any continuous, nonlinear physical system arbitrarily well. Recent theoretical work has led to affirmative answers to these qualitative questions. Furthermore, the latest theoretical progress has provided quantitative answers to the following issues: (1) necessary and sufficient conditions for universal approximation, and (2) given a continuous function, how to design a fuzzy system to uniformly approximate it with a required approximation accuracy. The establishment of the necessary conditions has also provided insight on the strengths and limitations of fuzzy systems as functional approximators. On one hand, even if a required approximation accuracy is very small, a small number of fuzzy rules may suffice to uniformly approximate those continuous functions that have a complicated formulation but a relatively small number of extrema. On the other hand, a large number of fuzzy rules are necessary for uniform approximation of continuous functions that are simple but have a lot of extrema (i.e., periodic functions). The interested reader is referred to [1] for an up-to-date, comprehensive, and deep theoretical treatment of fuzzy control, fuzzy modeling and fuzzy approximation. References [1] H. Ying, Fuzzy Control and Modeling: Analytical Foundations and Applications, IEEE Press, 2000. [2] H. Ying, M. McEachern, D. Eddleman, and L. C. Sheppard, “Fuzzy Control of Mean Arterial Pressure in Postsurgical Patients with Sodium Nitroprusside Infusion”, IEEE Transactions on Biomedical Engineering, 39:1060-1070, 1992. [3] J. Yen, R. Langari, and L. A. Zadeh (eds.), Industrial Applications of Fuzzy Logic and Intelligent Systems, IEEE Press, 1995. The issue of universal approximation is crucial to fuzzy systems. In the context of control, the question is whether a fuzzy controller can always be constructed to uniformly approximate any desired continuous, nonlinear control solution with enough accuracy. For modeling, the question is whether a fuzzy model can always be established which is capable of uniformly approximating any continuous, nonlinear physical system arbitrarily well. Recent theoretical work has led to affirmative answers to these qualitative questions. 37 ty e ci et y Awards Nominations 2001 ty S y t ie So c S y ie ty et i S ie S ty So ie oc ci the CAS ty For further y y information, c y S t osee t e y t o i t y e e y S y y e t i i t c e t t i S c e S c c c at http://www.ieee-cas.org ie homepage ie ci ie ie So ci y So So So So So oc oc et oc oc So So i ty be mailed to Bing y S S S S Nominations should be submitted electronically to the 2001 Awards Chair, Bing Sheu, at b.wehner@ieee.org. Forms may also i oc Sheu c/o Barbara Wehner, 15 W. Marne Ave., P.O. Box 265, Beverly Shores, IN 46301–0265. Tel: (219) 871–0210; Fax: (219) 871–0211. ◊◊ Guillemin-Cauer Award ◊◊ ◊◊ Industrial Pioneer Award ◊◊ Purpose: To honor a person or persons with outstanding and pioneering contributions in developing academic and industrial research results into industrial applications and/or commerical products. The award is to be presented annually together with the other awards, and given by the Awards Committee on the basis of quality, originality, and significance of contribution. Prize: Plaque and a $500 cash award that will be divided if there is more than one recipient. ◊◊ Education Award ◊◊ ◊◊ Darlington Award ◊◊ Purpose: To honor a person with outstanding contributions to education in a field within the scope of the CAS Society as documented by publications of textbooks, research supervision of graduate and undergraduate students, development of short courses and participation in adult education. The award is based on general quality and originality of contributions and continuity of effort. Anyone who is a member of the CAS Society is eligible. Prize: Plaque and $500 check. Purpose: To recognize the best paper bridging the gap between theory and practice published in the IEEE Transactions on Circuits and Systems. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of papers bridging the gap between theory and practice published in the IEEE Transactions on Circuits and Systems during the two calendar years preceding the award is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). ◊◊ CAD Transactions Best Paper Award ◊◊ ◊◊ Technical Achievement Award ◊◊ Purpose: To honor a person with outstanding technical contributions over a period of years within the scope of the CAS Society as documented by publications (including patents). The award is based on the general quality and originality of contributions and continuity of effort. Anyone who is a member of the CAS Society is eligible. Prize: Plaque and $500 check. Purpose: To recognize the best paper published in the IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of a paper published in the IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems during the two calendar years preceding the award is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). ◊◊ Mac Van Valkenburg Award ◊◊ ◊◊ VLSI Transactions Best Paper Award ◊◊ Purpose: To honor a person with outstanding technical contributions in a field within the scope of the CAS Society and outstanding leadership in the field. The award is based on quality and significance of contribution and continuity of technical leadership. Anyone who is a member of the CAS Society is eligible. Prize: Plaque and $1,000 Check. Purpose: To recognize the best paper published in the IEEE Transactions on Circuits and Systems for Very Large Scale Integration (VLSI) Systems. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of a paper published in the IEEE Transactions on Circuits and Systems for VLSI Systems during the two calendar years preceding the award is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). ◊◊ Chapter-of-the-Year Award ◊◊ ◊◊ CSVT Transactions Best Paper Award ◊◊ Purpose: To recognize the CAS Society Chapter with the best yearly activities. The award is based on best yearly activities in the categories of Chapter-sponsored technical activities, increase in membership and participation in BOG meetings. Anyone who is a member of the CAS Society Chapters is eligible. Prize: Certificates to Chapter Officers. Purpose: To recognize the best paper published in the IEEE Transactions on Circuits and Systems for Video Technology. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of papers published in the IEEE Transactions on Circuits and Systems for Video Technology during the two calendar years preceding the award is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). ◊◊ Meritorious Service Award ◊◊ Purpose: To honor a person with outstanding long-term service to the welfare of the CAS Society. The award is based on dedication, effort and contributions. Anyone who is a member of the CAS Society is eligible. Prize: Plaque and $500 check. 38 Purpose: To recognize the best paper published in the IEEE Transactions on Circuits and Systems. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of a paper published in the IEEE Transactions on Circuits and Systems during the two calendar years preceding the award is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). ◊◊ Outstanding Young Author Award ◊◊ Purpose: To honor an especially meritorious paper published in any one of the CAS Society's Transactions whose author at the date of submission is less than 30 years of age. The award is based on general quality, originality, contributions, subject matter and timeliness. Anyone who is an author of papers published in any one of the CAS Society Transactions during the two calendar years preceding the award, who at the date of submission of the paper shall be less than 30 years of age is eligible. Prize: Certificate and $250 check for each author (maximum of $1,000 per award). Use these forms as a guide to electronically submit nominations to Bing Sheu (see previous page). DUE BY FEBRUARY 1, 2001 IEEE Circuits and Systems Society 2000 Outstanding Paper Award Nomination Outstanding Young Author Award—CAD Transactions Best Paper Award Guillemin-Cauer Award—CSVT Transactions Best Paper Award Darlington Award—VLSI Transactions Best Paper Award Paper Award Recommended: _________________________________________________________________ Paper Title: _____________________________________________________________________________ ______________________________________________________________________________________ Paper Authors: ______________________________ ______________________________ ______________________________ ______________________________ Paper Listing: Name of Transactions: _______________________________________________________ Month: ________________________ Year: ________________ Pages: _________________ Nominator: Name: _________________________ Tel (day): ___________________ Tel (home-opt.): __________________ Address: _______________________ Fax: _______________________ E-mail: _________________________ ______________________________ ______________________________ Basis for Nomination: Please give the reasons you believe this paper is deserving of the outstanding paper award. Judging is based upon general quality, originality, contribution, subject matter, and timeliness. Continue on additional page(s). IEEE Circuits and Systems Society 2000 Society/Achievement Award Nomination Mac Van Valkenburg Award—Meritorious Service Award Industrial Pioneer Award—Chapter-of-the-Year Award Education Award—Technical Achievement Award Name of Award: _____________________________Date: _________________________________________ Nominee: Name: _____________________________ Address: _____________________________ Present Employment Position(s): ____________________ _____________________________ ______________________________________________ _____________________________Highest Degree Attained: __________________________ Telephone (day): ________________________________ Nominator: Name: Address: _____________________________ _____________________________ _____________________________ _____________________________ Telephone (day): ________________________________ Tel (home-opt.): __________________________________ Fax: __________________________________________ E-mail: ________________________________________ 1. Proposed Citation: Provide a brief statement, not exceeding 50 words, giving the major accomplishments for which the award is being made. This will be used if the nominee is selected as the awardee. Continue on separate page(s). 2. Basis for Nomination: Prepare a statement not exceeding 750 words on why the candidate is being nominated for the award. This statement should then be followed by the record of accomplishments of the candidate as an educator, and/or as a researcher, and/or as an administrator, and/or as an industrial pioneer, as appropriate. Continue on separate page(s). 3. Short Biography (Not exceeding 2 pages): Include degrees earned (list universities and granting dates); other postgraduate study; record of all positions held (chronologically starting with the most recent position); IEEE activities and offices; other society memberships and offices; awards, honors, patents, inventions and other relevant contributions. Continue on separate page(s). 4. Publications: List all books, book chapters, and journal papers as well as 10 of the most important publications stating the engineering significance of each. Continue on separate page(s). 5. References: No more than five brief supporting letters from colleagues (and former students for the CAS Society Education Award) should be included with each award nomination. List the names of the references on the nomination form. The reference letters can either be collected by the nominator and forwarded unopened to Bing Sheu, or the references can be instructed to forward their recommendations directly to Dr. Sheu. All reference letters must be received by the due date of the nominations, February 1, 2001. In the case of a re-nomination, references and other materials can be re-used. 39 The Midwest Symposium on Circuits and Systems: An Update T he 43rd meeting of the Midwest Symposium on Circuits and Systems (MWSCAS) has just concluded, and so it seems useful to recall some of the history of this symposium, record the events that took place at this meeting, and list future venues. Historical Overview The MWSCAS is the oldest continuously operating conference devoted to circuits and systems in the United States [1] and possibly in the Americas. It was started by Professors Myril B. Reed and Ray M. Wainright in 1955, and the first meeting was sponsored by the University of Illinois at Urbana in cooperation with the Professional Group on Circuit Theory of the Institute of Radio Engineers (IRE). The second meeting was held in 1956 at Michigan State University, where the symposium recently returned for its 43rd meeting. Although the symposium was originally designed as a regional conference, it has gradually extended its area of venues so that now it is a national or international conference. It has been held in states from Michigan to Texas and from California to New York, and it has been held in Canada three times (Waterloo in 1973, Quebec in 1975, and Calgary in 1990), Mexico once (Puebla in 1983), and Brazil once (Rio de Janeiro in 1995). A common quip at the meetings is that the MWSCAS is always held in the midwest of some city, county, state or country. The MWSCAS was an independent conference until 1989 when more formal ties were established with the Circuits and Systems Society and the IEEE. While it was an independent conference, its proceedings could be obtained from Western Periodicals, North Hollywood, California, except for the proceedings of the meetings in 1987 and 1988 which can be obtained from Elsevier Science Publishing Company, New York. Proceedings of 40 meetings from 1989 to the present time can be obtained from the IEEE. In 1977, the first paper to receive the Myril B. Reed Best Paper Award was selected. The award was presented at the meeting in 1978 to Prof. Ken Jenkins for his paper “Techniques for High-Precision Digital Filtering with Multiple Processors” [2]. This award has become one of the traditions of the MWSCAS and is presented every year. Other traditions and activities that have always been part of the MWSCAS or which have been added over the years are pre-conference tutorial sessions, a conference reception, a keynote address, plenary sessions, Janie Fouke, dean of engineering at Michigan State University and IEEE Division 4 director, addressing the audience at the Awards Banquet. panel sessions, tours of local points of interest, and, of course, a conference banquet. More recently, the conference has encouraged exhibitors, and although the number has been small, interest in the exhibits has been strong. Since its inception, the MWSCAS has been a student-friendly conference. The presence of students at this conference is appreciated, and the MWSCAS is always considering ideas to increase student attendance. It is at this conference that graduate students often present the results of their research for the first time in a public forum. At the same time, established researchers and other wellknown members of the engineering profession use this conference to communicate new results and ideas to their peers. The presence of graduate students, established researchers, and other well-known members of the engineering profession makes for a good mix. The 43rd Meeting The most recent meeting of the MWSCAS was held August 8–11 in Lansing, Michigan, and was sponsored by Michigan State University and the Circuits and Systems Society. About 350 registrants were in attendance. Welcoming remarks were provided by M. Peter McPherson, president of Michigan State University, Janie Fouke, dean of engineering, and Fathi Salam, professor of ECE and general chair. Professor Ken Jenkins of Pennsylvania State University was the program chair, and Professor Hoda Abdel-Aty-Zohdy of Oakland University was the program co-chair. There were three plenary sessions this year. The first was given by Professor Mohsen Kavehard of Pennsylvania State University and was entitled “Next Generation Wireless Communications System”. Professor Wasfy Michael of the University of Central Florida presented the second plenary session, and it had the intriguing title “Filtering and Signal Processing: From Weight Reduction to Evolutionary Learning”. The third plenary session, “Signal Processing by Floccular and Ventral Cells”, was presented by Dr. Stephen Highstein of the Washington University School of Medicine. Generally, the plenary sessions at the MWSCAS are focussed on new topics that are related in some way to network theory or to the design of new circuits or to other fields of science that may contribute to these areas in the future; and this year’s plenary sessions were no exception. Two informative panel sessions were included in this year’s meeting. In the first, moderated by Professor Robert W. Newcomb of the University of Maryland, the discussion centered on “Internet Based Design and Test”. Professor Rui de Figueiredo of the University of California, Irvine, and past president of the Circuits and Systems Society, moderated the second panel session on “The Role of Circuits and Systems Technologies in Emerging Telecommunications Scenarios”. The topics of both panel sessions are important to engineers in the circuits and systems community and were exciting to most of the attendees. Over the years, the scope of the MWSCAS has broadened, although there were papers presented even in the first meeting discussing digital circuits and digital circuit design philosophies. The 43rd meeting included technical tracks in analog circuits and signal processing, general circuits and systems, neural networks and systems, computer-aided design, digital signal processing, VLSI circuits and systems, multi-media and communication, and sensors, MEMS and industrial applications. These tracks contained a total of 50 sessions with each session consisting of at least six or seven papers. There were sessions entitled Recent Advances in Communication Systems, Robotics and Control, Bio-Inspired Signals and Systems, and Topics on Fuzzy Logic in addition to sessions such as Topics on Analog Circuits, Solid State Circuits and Devices, and Current Issues in CMOS Circuit Design. A unique event at this year’s symposium was a luncheon to honor Professor Robert Newcomb and his wife Sally. Dr. de Figueiredo announced that Dr. Newcomb was being honored for the beneficial impact his work has had on the MWSCAS, on the Circuits and Systems Society, and on electrical engineering at the national and international levels. He emphasized that Dr. Newcomb was one of the inventors of microelectronics, a pioneer in neural systems, and a pioneer in models for biological systems. Dr. Fathi Salam stated that Dr. Newcomb puts a human face on all his technical ac- tivities, and he treats everyone with warmth and depth. Dr. Ken Jenkins remarked that Dr. Newcomb is known for working with and helping students, accompanying them to conferences and introducing them to other people in their fields. Dr. M. N. S. Swamy mentioned that he has known Dr. Newcomb for 35 years not only for his contributions to circuit theory but also as a wonderful human being and recalled that Dr. Newcomb is also a poet and has a long-term interest in poetry and literature. Dr. Majid Ahmadi then presented Dr. Newcomb with an album containing notes and signatures from well-wishers who were present at the luncheon. Dr. Igor Filanovsky asserted that Dr. Newcomb’s many papers and books were always clear and even withstood Russian translations. He recalled an early paper in active synthesis by Robert W. Newcomb, honored for the beneficial impact his work has had on the MWSCAS, on the Circuits and Systems Society, and on electrical engineering. Kerwin, Huelsman, and Newcomb [3–4] in which the KHN circuit was introduced and stated, “When your work is reduced to one letter, you have become a classic researcher.” He conveyed the thanks of thousands of Russian radio engineers to Dr. Newcomb. Dr. Mona Zaghloul thanked Dr. Newcomb on behalf of all the students he has mentored and for being such a good mentor for her. Dr. Newcomb was then prevailed upon to say a few words. He commented that he had roots in Michigan from his grandfather on his mother’s side. He ended the proceedings by stating that he has taken great pride and joy in working with students and seeing them progress in life. The banquet at the MWSCAS is usually an occasion for attendees to hear an important address. This year, Dr. de Figueiredo introduced the main banquet speaker, Dean Janie Foulke, who spoke more in her capacity as director of Division 10 (Systems and Controls) of the IEEE rather than as dean of engineering at Michigan State University. Dr. Foulke based her remarks on the book she edited entitled Engineering Tomorrow [5] in which 50 engineers and scientists share their thoughts about engineering in the future. She conveyed a brief synopsis of some of the ideas in this book to the audience and emphasized that she is optimistic about the future. The conference banquet is also utilized to announce the winners of the best paper award and other prizes. This year the Myril B. Reed Best Paper Award was given to Samuel L. SanGregory, Charles Brothers, and David Gallagher of the Air Force Institute of Technology, WrightPatterson Air Force Base and to Raymond Siferd of Wright State University for their paper “A Fast, LowPower Logarithm Approximation with CMOS VLSI Implementation” which was presented at the MWSCAS held in Las Cruces in 1999 [6]. In recent years, a student paper contest has been held as one means to encourage student participation, and the names of the winners are announced at the banquet. This year there were about forty entries in this contest, split almost evenly between papers that addressed analog topics and those that were devoted to digital topics. Since the total number of papers submitted was quite an increase from the previous year, the contest was split into two contests corresponding to the natural split in the topics addressed by the papers. Each contest had a separate group of judges and a separate set of prizes. Prizes consisted of checks for $300 for first place, $200 for second place, and $100 for third place for each con. . . continued on Page 42 41 MWSCAS Report … continued from Page 41 test. Tables I and II list the winners for each contest. Table I Winning Student Papers in Analog Contest at 43rd Meeting First Place: “CMOS 5-10 GHz Oscillators for Low Voltage RF Applications” by Ahmed H. Mostafa, Mourad N. El-Gamal and Ramez A. Rafla, McGill University. Second Place: “A New Charge Redistribution D/A and A/D Converter Technique - Pseudo C2C Ladder” by Lin Cong and William C. Black, Iowa State University. Third Place (tie): students who participated in the contests as well as their faculty advisors are to be congratulated for making these paper competitions lively and strong. Future Venues The 44th MWSCAS is scheduled for August 13–17, 2001, in Dayton, Ohio. Dr. Robert L. Ewing (robert. ewing@wpafb.af.mil) is the general chair. Professor Harold Carter is cochair, and Gary B. Lamont chairs the Technical Program Committee with aid from Professors Belle Shenoi, Hoda Abdel-Aty-Zohdy, and Mohammed Ismail. The conference is being sponsored by the College of Engineering at Wright State University, the Dayton IEEE Section, and the IEEE “Charge-Mode Parallel Architecture for Matrix-Vector Multiplication” by Roman Genov and Gert Cauwenberghs, Johns Hopkins University. “A Novel Linear Tunable MOS Transconductor” by Ko-Chi Kuo and Adrian Leuciuc, SUNY at Stony Brook. Table II cal program chair is Dr. Keith Teague (teague@okstate.edu). Dr. Soderstrand is one of the “old hands” of the MWSCAS, having chaired the very successful 40th meeting at Sacramento, California. Conclusions The MWSCAS is probably one of the better kept secrets of the circuits and systems community. However, it has continued to serve the purpose of disseminating research results, not only in the circuits and systems areas, but also in a wide variety of other areas, for almost half a century. So in addition to the attendees ranging from graduate students to well-established researchers, the range of technical interests of the attendees is quite broad. It is a good conference to attend to see what is happening in your field, to find out what is happening in other fields that may affect your field, to introduce students to people they ought to know in their technical areas, and to enjoy the camaraderie that this conference seems to foster. References Winning Student Papers in Digital Contest at 43rd Meeting [1] First Place: “Quality of Data Reconstruction Using Stochastic Encoding and an Integrating Receiver” by Alyssa Apsel and Andreas Andreou, Johns Hopkins University. Second Place: “Efficient Polyphase Decomposition of Comb Decimation Filters in SigmaDelta Analog-to-Digital Converters” by H. Aboushady, Y. Dumonteix, M. M. Louerat, and H. Mehrez, University of Paris. Third Place (tie): “A VLSI Architecture for Soft-Output PR4 Detection” by Warren J. Gross, Vincent C. Gaudet, and P. Glenn Gulak, University of Toronto. “A Digital Frequency Synthesizer for a 2.4 GHz Fast Frequency Hopping Transceiver” by Riku Uusikartano and Jarkko Niittylahti, Tampere University of Technology. In both contests, there was a tie for third place. These students and all the 42 [2] Professor Fathi Salam, general chair of MWSCAS 2000, welcoming the Symposium audience. Circuits and Systems Society. Support for this meeting is provided by the University of Cincinnati, Ohio State University, and the Air Force Institute of Technology. Proposed papers for this conference can be submitted electronically to the web site (see www.mwscas.org) or five hard copies can be sent to Dr. Mohammed Ismail, Ohio State University, 205 Dreese Laboratory, 2015 Neil Avenue, Columbus, OH 43210–1272. In 2002, the 45th MWSCAS is scheduled to be held in Tulsa, Oklahoma under the auspices of Oklahoma State University. The general chair is Dr. Michael A. Soderstrand (sodersm @okstate.edu), chairman of the ECE Department at OSU; and the techni- [3] [4] [5] [6] Peter B. Aronhime and Donald J. Scheer, “The MWSCAS: A 42-Year History”, IEEE Circuits and Devices Magazine, vol. 13, no. 4, pp. 42–48, July 1997. W. K. Jenkins, “Techniques for High-Precision Digital Filtering with Multiple Microprocessors”, Proceedings of the MWSCAS, Lubbock, Texas, pp. 58–62, August 15–17, 1977. W. J. Kerwin, L. P. Huelsman, and R. W. Newcomb, “State-Variable Synthesis for Insensitive Integrated Circuit Transfer Functions”, IEEE Journal of Solid-State Circuits, vol. SC–2, pp. 87–92, September 1967. L. P. Huelsman and P. E. Allen, Introduction to the Theory and Design of Active Filters. New York: McGraw-Hill, pp. 214–232, 1980. Janie Fouke, ed., Engineering Tomorrow: Today’s Technology Experts Envision the Next Century. Piscataway, NJ: IEEE Press, 2000. Samuel L. SanGregory, Charles Brothers, David Gallagher, and Raymond Siferd, “A Fast, Low-Power Logarithm Approximation with CMOS VLSI Implementation”, IEEE Proceedings of the MWSCAS, vol. 1, Las Cruces, New Mexico, pp. 388–391, August 8–11, 1999. Peter Aronhime ECE Dept. University of Louisville The CNN Young Researcher Contest 2000 by Bertram Shi, Csaba Rekeczky, Marco Gilli and Mamoru Tanaka T he CAS Society Technical Committee on Cellular Neural Networks and Array Computing recently initiated a CNN Young Researcher Contest. The organizing committee consisted of Csaba Rekeczky, Bert Shi, Marco Gilli and Mamoru Tanaka. It was held in conjunction with the 6th IEEE International Workshop on Cellular Neural Networks and Their Applications (CNNA 2000), which was held in Catania, Italy between 23–25 May 2000. The goal was to provide students and young researchers worldwide with an opportunity to familiarize themselves with the capabilities of the CNN Universal Machine and to hone their analogic programming skills by competing against others. Cash prizes of US$500, US$300, US$200 and certificates were awarded to the top three entries. Student winners were also given fee offsets to assist them in attending CNNA 2000 to present their results. Participants in the contest were given the task of deriving an algorithm to segment and classify watermarks embedded into a set of images. The algorithm was required to be executable on a Cellular Neu- ral Network Universal Machine. There were three classes of watermarks (sharks, turtles and eagles). Each class contained 12 binary images: 3 different sizes of a prototype image in 4 different orientations (rotated by 90 degrees). The watermarks were embedded into 21 training images of natural seashore via an undisclosed sequence of nonlinear operations. To ensure that all entries would be treated fairly, carefully defined numerical evaluations were released to all participants in advance. Although the criteria were known in advance, they were computed based on test images which were similar, but not identical to those in the training set and which were not released to the participants. The entries were evaluated in terms of their quality of the watermark segmentation and classification, their computational (time) complexity and their hardware complexity. These three individual scores were then combined into a global figure of merit, which was used to determine the winning entries. Figure 1 shows example input/output images from the segmentation algorithm of one of the contest entries. . . . continued on Page 44 (input.tif) (output.tif) (reference.tif) (a) (b) (c) Figure 1. (a) Sample input image, (b) output image of the segmentation from one of the contest entries, (c) the reference image used to generate the watermarked input image and to evaluate the segmentation quality of the output image. 43 CNNA Contest … continued from Page 43 A web page (located at http:// lab.analogic.sztaki.hu/cnna2000) was set up which contained the description of the contest and a description of the quantitative evaluation criteria used to decide the winning entries. After registering through the web page, participants were able to download a student version of a commercially available analogic CNN algorithm THE ADVENTURES OF … …THE 'UMBLE OHM …Shlomo Karni AMPHi(t)HEATER simulator and programming environment, the watermark images and the training images. Participants were also able to track the performance of their submitted algorithms on testing images through this web page. Judging by the response, the contest achieved its goal. Before the contest deadline, the web page registered over 900 hits from over 250 unique IP addresses. In total, 23 users registered for the contest from a variety of countries in North America, Europe and Asia. Authors of the top three winning entries were invited to present their results at the 6th IEEE International Workshop on Cellular Neural Networks and Their Applications (CNNA 2000) held 44 in Catania, Italy on 24 May 2000. The top three winning entries were 1. Péter Földesy (Focal Plane Array Processors Research Group, IMSE-CNM, Seville, Spain) 2. Dávid Bálya (Vision Research Laboratory, MCB-UCB, Berkeley, USA) 3. David Monnin (French-German Research Institute of Saint-Louis, SaintLouis, France) Based upon the success of this contest, the IEEE CAS Technical Activities Board has agreed to sponsor a biannual CNN Young Researcher Contest. The contest will be held in conjunction with every IEEE International Workshop on Cellular Neural Networks and Their Applications. The next contest will be held in 2002 and the results will be announced in Frankfurt, Germany. Keep your eyes posted for announcements, which should be coming out in 2001. Readers who cannot wait that long are encouraged to point their browsers to http:// lab.analogic.sztaki.hu/cnna2000, which still contains all necessary information, so that they can try the contest task for themselves and see how they would have done. CHUA RECEIVES NEURAL NETWORKS PIONEER AWARD P rofessor Leon Chua is the recipient of the year 2000 Neural Networks Pioneer Award. The award was presented at the International Joint Conference on Neural Networks (IJCNN 2000), held in Como, Italy, on July 24–27. The Neural Networks Pioneer Award is the most prestigious award given annually by the IEEE Neural Networks Council. Previous winners include Teuvo Kohonen, Stephen Grossberg, Bernard Widrow, Shun-Ichi Amari, Paul Webos, Michael Arbib, John Hopfield, and Geoffrey Hinton. The award consists of a plaque and $2,000 travelling expenses for the winner and spouse. Places Places Places Places 2000 IEEE Asia-Pacific Conference on Circuits and Systems Electronic Communication Systems December 4–7, 2000 Crystal Palace Hotel, Tianjin, China December 17–20, 2000 http://www.icecs2k.polymtl.ca Technical Program Co-Chair Professor Yoji Kajitani Tokyo Institute of Technology, Japan Professor Runtao Ding Tianjin University, Tianjin, China Tutorials Dr. Ibrahim Hajj University of Illinois, USA E-mail: i-hajj@uiuc.edu Special Sessions Dr. Fadi Kurdahi University of California, Irvine, USA E-mail: kurdahi@ece.uci.edu Requests for information: ICECS’2K Secreteriat School of Engineering and Architecture The Lebanese American University P.O. Box 36, Byblos, Lebanon FAX: 961 9 944 851 E-mail: genasr@lau.edu.lb sawan@vlsi.polymtl.ca Le Places na hi C ng Ko ng Places Places Jo sé Ho Places an 45 Web site: http://www.eee.hku.hk/misc/ieeecas/ Places Places Places Places S Places Places Places Visit our web site at http://www.iccad.com Places Program Vice Chair Lawrence T. Pileggi CMU pileggi@ece.cmu.edu Places Program Chair Rolf Ernst Tech. Univ. of Braunschweig ernst@ida.ing.tu-bs.de For further information, please contact: Professor Tony T. S. Ng Department of Electrical & Electronic Engineering The University of Hong Kong Pokfulam Road, Hong Kong Tel: +852-28592710 Fax: +852-25598738 E-mail: tsng@eee.hku.hk Places Conference Manager Kevin Lepine MP Associates, Inc. 5305 Spine Rd., Ste. A Boulder, CO 80301 303-530-4562 303-530-4334 kevin@dac.com Organization Committee Tony T. S. Ng (Chairman) Qiuting Huang George Moschytz (Co-Chairman) Markus Halfenstein Ruey-Wen Liu Magdy Bayoumi Hari Reddy Chris Toumazou Yrjo Neuvo Geert De Veirman Places General Chair Ellen Sentovich Cadence Berkeley Labs 2001 Addison St. 3rd Floor Berkeley, CA 94704-1103 510-647-2807 510-486-0205 ellens@cadence.com This three-day workshop addresses implementation and application issues for 3rd Generation Mobile Communication. The program will combine presentations by experts in the field from industry and academia, with panel and informal discussions. Places Places Hong Kong Convention & Exhibition Center HONG KONG Places DoubleTree Hotel San José, California Places November 29–December 1, 2000 Places ICCAD–2000 Places 4th IEEE CAS Workshop on Emerging Technologies in Circuits and Systems Third Generation Mobile Technologies and Applications The rapid progress in process technology development combined with the expanding diversity of information processing applications is placing enormous demands on CAD tool development. The International Conference on Computer Aided Design 2000 (ICCAD) offers a place for CAD developers and VLSI designers to meet and exchange ideas about the problems and solutions in the era of system-on-a-chip. Places APCCAS2000 Conference Secretariat School of Electronic Information Engineering Tianjin University, Tianjin 300072, China Fax : 86-22- 2740-1471 Tel : 86-22- 2740-5623 E-mail : hywang@tju.edu.cn n General Co-Chair Professor Zhenming Chai Institute of Electronics Academia Sinica Beijing, China Technical Program Chair Dr. Mohamad Sawan Ecole Polytechnique de Montreal Email: sawan@vlsi.polymtl.ca no Technical Program Chair Professor Yih-Fang Huang University of Notre Dame U.S.A. General Chair Dr. Abdallah Sfeir Lebanese American Univ., Lebanon Email: asfeir@lau.edu.lb Places General Chair Professor Yongshi Wu Tianjin University, Tianjin, China The IEEE International Conference on Electronics, Circuits and Systems (ICECS) will celebrate the year 2000 in Beirut. For the first time an IEEE International Conference on Electronics, Circuits and Systems will be held in Lebanon. The venue will be in the continuously expanded and modern region of Lebanon located at the north of Beirut—the heart of a lively multicultural and Mediterranean country. ba The 2000 IEEE Asia-Pacific Conference on Circuits and Systems (APCCAS2000) is the fifth in the series of biennial Asia-Pacific Conferences sponsored by the IEEE Circuits and Systems Society, Chinese Institute of Electronics (CIE), IEEE Beijing Section, IEEE CAS Beijing Chapter and Tianjin University of China. It will be held at the Crystal Palace Hotel, Tianjin, China on December 4–7, 2000. The conference will be devoted to all aspects of theory, design, modeling, simulation, and applications of circuits and systems. Plenary sessions, special sessions, invited talks, and tutorials on specific advanced topics will also be included in the program. Places Kaslik, LEBANON Places Places Places Places Places Places Places Places Places Places Places Places Places APCCAS2000 http://www.tju.edu.cn/news/apccas2000 Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places November 13–17, 2000 Registration Information AMSC, Texas A&M University ATTN: Ella Gallagher Dept. of Electrical Engineering College Station, TX 77843-3128 USA Tel: (979) 845–9587 Fax: (979) 845–7161 E-mail: ella@ee.tamu.edu October 16, 2000 December 4, 2000 January 8, 2001 D al la s Registration deadline is October 24 Internet Information http://amsc.tamu.edu/shortcourses Places 2001 SOUTHWEST SYMPOSIUM ON MIXED-SIGNAL DESIGN 8th IEEE International Conference on Electronics, Circuits and Systems Places February 25–27, 2001 Austin, Texas, U.S.A. Places Places Places Places Places tin Places Places Places Places Places Places Places Places M al ta R om an ia Places us Other information: http://www.tuiasi.ro/events/scs2001 http://www-imt.unine.ch/scs2001 lgoras@etc.tuiasi.ro ( Prof. Liviu Goras ) Places Secretariat: Phone: +40 32 142283 FAX: +40 32 217720 or +40 32 278628 The fifth edition of the International Symposium on Signals, Circuits and Systems will be held in Iasi at the Faculty of Electronics and Telecommunications, Technical University “Gh. Asachi”. Iasi, the oldest academic center of Romania, is located in the North-Eastern part of the country and can be reached from Bucharest by plane or by train. A wine tasting and a one day post symposium trip in the neighboring region, well known for its wonderful monasteries, will be organized. Places Address for correspondence: SCS’2001 International Symposium Faculty of Electronics & Telecommunications “Gh. Asachi” Technical University of Iasi Bd. Carol 11, Iasi, 6600, ROMANIA Iasi, Romania Places A July 10–11, 2001 Places Places The Fifth International Symposium on Signals, Circuits and Systems Places Places Technical Program Chair Prof. Franco Maloberti Department of Electrical Engineering Texas A&M University College Station Texas 77843 USA Email: franco@ele.unipv.it Places Further Information For details on the electronic submission procedure, please go to http://www.ece.arizona.edu conferences/ssmsd ICECS 2001 Secretariat Department of Microelectronics Faculty of Engineering University of Malta Msida MSD 06 Malta Europe E-mail: icecs01@eng.um.edu.mt Tel: (+356) 346760 Fax: (+356) 343577 Places Author’s Schedule Submission of three-page summary: October 13, 2000 Notification of acceptance: December 15, 2000 Final camera-ready paper & registration due: January 15, 2001 General Chair Dr. Joseph Micallef Department of Microelectronics University of Malta Msida MSD06 Malta Europe Email: jjmica@eng.um.edu.mt Places Overview The 2001 Southwest Symposium on Mixed-Signal Design is sponsored by the University of Arizona Dept. of ECE in cooperation with the IEEE SolidState Circuits Society. The symposium will bring together researchers in academia, industry, and government from the areas of CAD tools and mixedsignal circuit design. This will allow not only the presentation of the latest developments in each field, but also interaction between the areas in order to expedite the development of improved integrated systems and “systems on a chip”. Of particular interest is the integration of mixed-signal types, such as analog, digital, RF, optical, and microwave circuits and systems. Schedule for Authors: Proposals for special sessions, plenary sessions, and short courses: October 15, 2000 Deadline for submission of draft papers: February 15, 2001 Places General Chair General Co-Chair John Choma, Jr. Jo Dale Carothers Univ. of Southern Univ. of Arizona California Program Chair Franco Maloberti Texas A&M Univ. Places Places September2–5, 2001 The Westin Dragonara Resort, Malta Places Places http://www.eng.um.edu.mt/microelectronics/icecs2001 ICECS 2001, the 8th IEEE International Conference on Electronics, Circuits and Systems is a major international conference which includes regular, special and poster sessions on topics covering analogue circuits and signal processing, general circuits and systems, digital signal processing, VLSI, multimedia and communication, computational methods and optimization, neural systems, control systems, industrial and biomedical applications, and electronic education. Places 46 ICECS’01 ICECS’01 ICECS’01 Places Places < < CALL FOR PAPERS > > Places * * First Call for Papers * * Places http://ewh.ieee.org/soc/cas/dallas/wks2 Places For further information visit: Places Author’s Schedule: Deadline for submission of papers Notification of paper acceptance Deadline for camera-ready paper as Places Analog and Mixed-Signal Testing Short Course Who Should Attend This course is intended for new and experienced electric and electronic engineers who need intensive concentrated information on the state-of-the-art of Analog and Mixed-Signal Circuits Testing. In particular this course is intended for: • IC testing engineers • Analog and Digital Circuit Designers • Application engineers. The IEEE 2nd Dallas CAS Workshop on Low Voltage and Low Power Mixed-Signal Circuits & Systems is being conducted under the aegis of the IEEE Dallas Section, Circuits & Systems Society. The workshop will include invited lectures and poster sessions. Leading experts Dr. John Choma, Dr. Randall Geiger, Dr. Franco Maloberti, and Dr. Jaime Ramírez-Angulo will present the latest techniques and challenges involved in low power RF design, low voltage analog circuits, low voltage sigma-delta and nyquist rate data converters. Prospective authors are invited to submit a one-page summary of their posters reporting original work in the area of low voltage/low power mixed-signal design. Registration Chair: Dr. Oscar Moreira-Tamayo omoreira@ti.com Places TEXAS A&M UNIVERSITY Department of Electrical Engineering Monday, 26th March 2001 Dallas, Texas Technical Chair: Dr. Joseph Varrientos joseph.varrientos@dalsemi.com Places Analog and Mixed Signal Center 2000 IEEE 2nd Dallas CAS Workshop on Low Voltage Mixed Signal Circuits and Systems General Chair: Dr. Gabriele Manganaro gabriele.manganaro@ieee.org Places Te x Places Places ANNOUNCEMENT AND CALL FOR PAPERS Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Places Wireless Networks … continued from Page 24 [5] 3GPP, Radio Interface Architecture, 3GPP TS 25.301, version 3.1.0, July 1999. [6] E. Dahlman, P. Beming, J. Knutsson, F. Ovestjo, M. Persson, and C. Roobol, “WCDMA- The Radio Interface for Future Mobile Multimedia Communications”, IEEE Transactions on Vehicular Technology, vol. 47, no. 4, 1998. [7] Michael Meyer, “TCP Performance over GPRS”, Proceedings of the Wireless Communications and Networking Conference, pp. 1248–1252, 1999. [8] M. Allman, et al., “Ongoing TCP Research Related to Satellites”, Internet-Draft: http://www.ietf.org/internet drafts/ draft-ietf-tcpsat-res-issues-11.txt, September 28, 1999. [9] S. Dawkins, et al., “End-to-End Performance Implications of Links with Errors”, Internet-Draft: http://www.ietf.org/ internet-drafts/draft ietf-pilc-error-02.txt, October 26, 1999. [10] S. Dawkins, et al., “End-to-End Performance Implications of Slow Links”, Internet-Draft: http://www.ietf.org/ internet-drafts/draft-ietf-pilc-slow-02.txt, October 21, 1999. [11] S. Gruhl, A. Echihabi, T. Rachidi, M. Link, and M. Söllner, “A Demonstrator for Real-Time Multimedia Sessions over 3rd Generation Wireless Networks”, to be published in Proceedings of ICME 2000, IEEE Press. [12] “Packet-Based Multimedia Communications Systems”, International Telecommunications Union, Geneva, Switzerland, ITU-T Recommendation H. 323, 1998. [13] “Call Signaling Protocols and Media Stream Packetization for Packet-Based Multimedia Communication Systems,” International Telecommunications Union, Geneva, Switzerland, ITU-T Recommendation H. 225.0, 1998. [14] “Control of Communications between Visual Telephone Systems and Terminal Equipment,” International Telecommunications Union, Geneva, Switzerland, ITU-T Recommendation H.245, 1998. [15] H. Schulzrinne, et al., “RTP: a Transport Protocol for RealTime Applications,” Request for Comments (Proposed Standard) RFC 1889, Internet Engineering Task Force, January 1996. [16] L. Zang, et al., “RSVP: A New Resource Reservation Protocol,” IEEE Network, pp. 8–18, September 1993. [17] H. Schulzrinne, A. Rao, and R. Lanphier, “Real-Time Streaming Protocol (RTSP)”, Internet Draft, Internet Engineering Task Force, March 1997. IEEE CAS Newsletter The Circuits and Systems Society’s homepage web site is: http:// www.ieee-cas.org Society Officers B.J. Sheu, President H.C. Reddy, President-Elect M.E. Zaghloul, Vice President, Technical Activities G.A. De Veirman, Vice President, Conferences G. De Micheli, Vice President, Publications A. Dunlop, Vice President, Regions 1–7 A. Davies, Vice President, Region 8 J. E. Cousseau, Vice President, Region 9 T.S. Ng, Vice President, Region 10 I.N. Hajj, Vice President, Administration G. Moschytz, Past President Board of Governors W. Black R.J. Marks, II J.A. Nossek W. Wolf E. Yoffa E. Friedman N. Fujii I. Galton M. Green C. Toumazou G.E. Gielen R. Gupta M. Hasler P. Pirsch H. Yasuura Representatives E. Friedman, Solid-State Circuits Society C.Y. Wu, J. Zurada, IEEE Neural Networks Council M.E. Zaghloul, Sensors Council Standing Committees G. Moschytz, Awards L.O. Chua, Fellows R.J.P. de Figueiredo, Nominations K. Thulasiraman, Constitution/Bylaws Transformers with High Power Density and High Efficiency”, IEE Electronics Letters, vol. 36, no. 11, May 2000, pp. 943–944. [16] S. Y. R. Hui, S. C. Tang, and H. S. H. Chung, “Some Electromagnetic Aspects of Coreless PCB Transformers”, IEEE Transactions on Power Electronics, vol. 15, no. 4, July 2000. [17] S. Y. R. Hui and S. C. Tang, “Coreless Printed Circuit Board Transformers”, US patent pending [18] Analog Devices Manual: Isolation Amplifier AD215. Editors M.K. Sain, IEEE CAS Newsletter M.N.S. Swamy, IEEE CAS I Transactions C. Toumazou, IEEE CAS II Transactions G. De Micheli, IEEE CAD Transactions W. Li, IEEE Transactions on Video Technology M.T. Sun, Multimedia Transactions W. Wolf, IEEE Transactions on VLSI Systems M.E. Zaghloul, Circuits & Devices Magazine Corresponding Editor A.J. Payne, IEEE CAS II Transactions Dist. Lecturer Program Conference Chairs E. Yoffa D. Skellern, G. Hellestrand, 2001 ISCAS Co-Chairman E.M. Sentovich, 2000 ICCAD General Chairman C. Chase, 2000 ICCD General Chairman J. Rabaey, 2001 DAC General Chairman Y.-S. Wu, 2000 APCCAS General Chairman F.M. Salam, 2000 MWSCAS General Chairman A. Sfeir, 2000 ICECS General Chairman Technical Committees Coreless PCB Transformers … continued from Page 15 G. Ron Chen, Nonlinear Circuits and Systems A. Ioinovici, Power Systems & Power Electronics Circuits G. Barrows, Sensors & Micromachining J. Ostermann, Visual Signal Processing R. Sridhar, VLSI Systems and Applications G. Cauwenberghs, Analog Signal Processing T. Roska, Cellular Neural Networks & Array Computing M.A. Bayoumi, CAS for Communications R. Gupta, Computer-Aided Network Design P. Diniz, Digital Signal Processing C.S. Li, Multimedia Systems and Applications M. Ahmadi, Neural Systems and Applications Administrator B.Wehner 47 I E E E Circuits and Systems S O C I E T Y N E W S L E T T E R A 2001 IEEE International Symposium on Circuits and Systems May 6–9, 2001 Sydney Convention and Exhibition Center Darling Harbor, Sydney, Australia SYSTEMS of CIRCUITS and MIXED TECHNOLOGY ELEMENTS The 2001 IEEE International Symposium on Circuits and Systems will be held in Sydney, Australia. The Symposium is sponsored by the IEEE Circuits and Systems Society. The Symposium will include: regular sessions; plenary sessions on advanced aspects of theory, design and applications of circuits and systems; and short courses/tutorials linked with special sessions on wireless, mixed technology systems engineering, high speed devices and modelling, signal and video processing, and low power high speed VLSI design. Web Site: http://www.elec.mq.edu.au/iscas01/ CONFERENCE CO-CHAIRMEN: Professor David Skellern Electronics Department (E6A 247) Division of Information and Communication Sciences Macquarie University NSW 2109 Australia daves@elec.mq.edu.au Tel: +61 2 9850 9145; Fax No: +61 2 9850 918 http://www.elec.mq.edu.au/ Professor Graham Hellestrand Address: VaST Systems Technology Corporation 1230 Oakmead Parkway Suite 314, Sunnyvale, CA 94806 USA Tel: +1 408 328 0909 Fax No: +1 408 328 0945 g.hellestrand@vastsystems.com http://www.vastsystems.com THE INSTITUTE OF ELECTRICAL & ELECTRONICS ENGINEERS, INC. 445 HOES LANE PISCATAWAY, NJ 08855 48