Ada's research interests are electromagnetic theory, integrated circuits, and information theory with applications in the following areas.
CHIC (CHip In Cell)
In-situ detection of chemical changes in human body at the cellular level can bring enormous benefits in diagnosis and in therapeutic monitoring. We are developing techniques to place micron-sized sensor chip inside each cell. It might revolutionize biochemical imaging by introducing the idea of replacing ''passive'' radiotracers with ''active'' IC chips. This may open up an array of new biomedical applications from novel medical diagnostic and therapeutic tools that operate at single cell level to a novel class of autonomously operating intrabody nanobiosensors.
Focused Microwave Ablation
Funded by CVI seed grant
Atrial fibrillation (AF) is the most common cardiac arrhythmia, responsible for approximately one?third of all hospital admissions with a cardiac rhythm disturbance and affects 1% to 1.5% of the population in the developed world. Myocardial ablation to interrupt electrical foci that may induce AF has been shown to be feasible but a range of technical shortcomings limit its applicability. Radiofrequency catheter?based ablation requires transseptal catheterization and direct energy delivery to the inside of the heart, steps that increases complications and the technical expertise required for the procedure. Ablation is directed towards the endocardial surface of the left atrium. The development of techniques that permit the delivery of energy for ablation outside the heart will likely decrease the risk associated with catheterization and endocardial ablation. The ideal system of the future for ablation of the atrium would use external energy delivery. Alternative energy sources such as laser and cryoablation have been used for endocardial ablation but are unable to achieve external ablation. High intensity focused ultrasound (HIFU) has limited penetration into air?filled viscera and bone.
Electromagnetic waves in the microwave spectrum has good propagation properties inside tissue as compared to ultrasound. It is based on non-ionized radiation and therefore, repeated exposures are feasible and there are very few side effects of treatment. In this project, we study focused microwave as an alternative energy source for ablation.
Atrial fibrillation (AF) is the most common cardiac arrhythmia, responsible for approximately one?third of all hospital admissions with a cardiac rhythm disturbance and affects 1% to 1.5% of the population in the developed world. Myocardial ablation to interrupt electrical foci that may induce AF has been shown to be feasible but a range of technical shortcomings limit its applicability. Radiofrequency catheter?based ablation requires transseptal catheterization and direct energy delivery to the inside of the heart, steps that increases complications and the technical expertise required for the procedure. Ablation is directed towards the endocardial surface of the left atrium. The development of techniques that permit the delivery of energy for ablation outside the heart will likely decrease the risk associated with catheterization and endocardial ablation. The ideal system of the future for ablation of the atrium would use external energy delivery. Alternative energy sources such as laser and cryoablation have been used for endocardial ablation but are unable to achieve external ablation. High intensity focused ultrasound (HIFU) has limited penetration into air?filled viscera and bone.
Electromagnetic waves in the microwave spectrum has good propagation properties inside tissue as compared to ultrasound. It is based on non-ionized radiation and therefore, repeated exposures are feasible and there are very few side effects of treatment. In this project, we study focused microwave as an alternative energy source for ablation.
Locomotive Micro Implants with Active Electromagnetic Propulsion
Funded by DARPA and FCRP C2S2
Locomotive implantable devices have numerous applications including sensing, imaging, minimally invasive surgery, and research. Many techniques have been used to generate motion, including mechanical solutions and passive magnetic solutions. Power sources dominate the size of existing active implant technologies, and this size constaint (typically in the cm-range) limits the potential for propulsion. Additionally, mechanical propulsion is inherently inefficient at the scale of interest. Passive locomotion schemes have circumvented the power and efficiency issues, but require large field gradients and usually cannot generate vertical motion. Recent improvements in wireless power transmission efficiency allow us to devise a new propulsion method that can exploit the advantages of both approaches.
The new propulsion method is a simple application of the Lorentz force on specific wire arrangements. In addition to forward propulsion, these wire arrangements can generate torques to steer the device through the fluid. Although the magnetic field only exerts perpendicular forces on the wires, altering the orientation of the device can generate lift due to the fluid drag force, resulting in vertical motion. To fully realize the forces and torques necessary for this control, DC magnetic shielding must be used to prevent certain currents from experiencing forces. In this way, the design achieves full 3D motion by a simple manipulation of currents.
[1] D. Pivonka, A. S. Y. Poon, and T. H. Meng, "Locomotive micro-implant with active electromagnetic propulsion," Proc. IEEE Engineering in Medicine and Biology Society Annual International Conference (EMBC), Minneapolis, Mn, Sept. 2009.
Locomotive implantable devices have numerous applications including sensing, imaging, minimally invasive surgery, and research. Many techniques have been used to generate motion, including mechanical solutions and passive magnetic solutions. Power sources dominate the size of existing active implant technologies, and this size constaint (typically in the cm-range) limits the potential for propulsion. Additionally, mechanical propulsion is inherently inefficient at the scale of interest. Passive locomotion schemes have circumvented the power and efficiency issues, but require large field gradients and usually cannot generate vertical motion. Recent improvements in wireless power transmission efficiency allow us to devise a new propulsion method that can exploit the advantages of both approaches.
The new propulsion method is a simple application of the Lorentz force on specific wire arrangements. In addition to forward propulsion, these wire arrangements can generate torques to steer the device through the fluid. Although the magnetic field only exerts perpendicular forces on the wires, altering the orientation of the device can generate lift due to the fluid drag force, resulting in vertical motion. To fully realize the forces and torques necessary for this control, DC magnetic shielding must be used to prevent certain currents from experiencing forces. In this way, the design achieves full 3D motion by a simple manipulation of currents.
[1] D. Pivonka, A. S. Y. Poon, and T. H. Meng, "Locomotive micro-implant with active electromagnetic propulsion," Proc. IEEE Engineering in Medicine and Biology Society Annual International Conference (EMBC), Minneapolis, Mn, Sept. 2009.
Wireless Delivery of Power and Data to Medical Implants
Funded by CIS seed fund and FCRP C2S2
An increasingly important problem in biomedicine and biomimetics is the contactless monitoring of physiological processes. Both power and data are transferred wirelessly to the implanted stimulators and sensors. While the concept of wireless power delivery itself is not new, harvesting meaningful power from incident electromagnetic waves in current wireless power delivery approaches requires large receive antennas that are impractical for use in these implants. We have recently shown that, through a full-wave analysis [1,2] and innovative circuit design techniques [3], the antenna area of the receiver can be reduced by two orders of magnitude while maintaining the same power transfer efficiency. Next, we will focus on the power transmitter, the data link, and the sensing circuits to track the implants and to optimize both power and data delivery. The goal of this project is to develop a core technology for the wireless delivery of power and data with a mm-scale implant and a cm-scale external transceiver. This core technology will serve as a foundation for existing and emerging medical implants.
[1] A. S. Y. Poon, S. O'Driscoll, and T. H. Meng, "Optimal operating frequency in wireless power transmission for implantable devices," Proc. IEEE Engineering in Medicine and Biology Society Annual International Conference (EMBC), Lyon, France, pp. 5673ñ5678, Aug. 2007.
[2] A. S. Y. Poon, S. O'Driscoll, and T. H. Meng, "Optimal frequency for wireless power transmission into dispersive tissue," submitted to IEEE Trans. Antennas and Propagation, Mar. 2008, revised Jan. 2009.
[3] S. O'Driscoll, A. S. Y. Poon, and T. H. Meng, "A mm-sized implantable power receiver with adaptive link compensation," IEEE International Solid-State Circuits Conference (ISSCC), Feb. 2009.
[4] S. O'Driscoll, A. S. Y. Poon, and T. H. Meng, "A mm-sized implantable power receiver with adaptive link compensation," submitted to IEEE Journal of Solid-State Circuits.
An increasingly important problem in biomedicine and biomimetics is the contactless monitoring of physiological processes. Both power and data are transferred wirelessly to the implanted stimulators and sensors. While the concept of wireless power delivery itself is not new, harvesting meaningful power from incident electromagnetic waves in current wireless power delivery approaches requires large receive antennas that are impractical for use in these implants. We have recently shown that, through a full-wave analysis [1,2] and innovative circuit design techniques [3], the antenna area of the receiver can be reduced by two orders of magnitude while maintaining the same power transfer efficiency. Next, we will focus on the power transmitter, the data link, and the sensing circuits to track the implants and to optimize both power and data delivery. The goal of this project is to develop a core technology for the wireless delivery of power and data with a mm-scale implant and a cm-scale external transceiver. This core technology will serve as a foundation for existing and emerging medical implants.
[1] A. S. Y. Poon, S. O'Driscoll, and T. H. Meng, "Optimal operating frequency in wireless power transmission for implantable devices," Proc. IEEE Engineering in Medicine and Biology Society Annual International Conference (EMBC), Lyon, France, pp. 5673ñ5678, Aug. 2007.
[2] A. S. Y. Poon, S. O'Driscoll, and T. H. Meng, "Optimal frequency for wireless power transmission into dispersive tissue," submitted to IEEE Trans. Antennas and Propagation, Mar. 2008, revised Jan. 2009.
[3] S. O'Driscoll, A. S. Y. Poon, and T. H. Meng, "A mm-sized implantable power receiver with adaptive link compensation," IEEE International Solid-State Circuits Conference (ISSCC), Feb. 2009.
[4] S. O'Driscoll, A. S. Y. Poon, and T. H. Meng, "A mm-sized implantable power receiver with adaptive link compensation," submitted to IEEE Journal of Solid-State Circuits.
Fundamentals of wireless power transfer and backscatter sensing/modulation
Coming Soon...
Millimeter-wave MAC and network protocols
Funded by CISCO
In recent years, we have used to walking around with our laptops and enjoying the convenience of Internet access via WiFi, and checking email with our cell phones on the road. However, there is still a considerable gap in speed and security level between the wired and wireless connections. We propose to consider millimeter-wave wireless connectivity to close this gap. Higher carrier frequency allows the use of wider signal spectrum. For example, there are 7 GHz of unlicensed spectrum available in the 60 GHz band in the United States. The multi-gigahertz spectrum allows multi-gigabit of data rate with ease. This is in contrast to current practice of cramming many bits per Hertz of spectrum in the low GHz-range which is expensive and hard to implement. At millimeter-wave frequencies, antenna is small and more antennas can be placed on the same device area. This results in links having very narrow beamwidths. Connections are therefore directional. Only intended receivers can listen to the transmitted message. Adjusting the phase and magnitude on each antenna element can change the direction of each link. As a result, each link is like a wired connection while the placement of this “wire” can be changed quickly.
As point-to-point millimeter-wave physical radios with electronically steerable arrays have been successfully implemented in CMOS with low power and small form factor, it is now possible to bring it to another level and implement a fully infrastructureless network of these radios. This project focuses on the investigation of the MAC and network protocols for millimeter-wave networks that can potentially support multi-gigabit data link connectivity.
In recent years, we have used to walking around with our laptops and enjoying the convenience of Internet access via WiFi, and checking email with our cell phones on the road. However, there is still a considerable gap in speed and security level between the wired and wireless connections. We propose to consider millimeter-wave wireless connectivity to close this gap. Higher carrier frequency allows the use of wider signal spectrum. For example, there are 7 GHz of unlicensed spectrum available in the 60 GHz band in the United States. The multi-gigahertz spectrum allows multi-gigabit of data rate with ease. This is in contrast to current practice of cramming many bits per Hertz of spectrum in the low GHz-range which is expensive and hard to implement. At millimeter-wave frequencies, antenna is small and more antennas can be placed on the same device area. This results in links having very narrow beamwidths. Connections are therefore directional. Only intended receivers can listen to the transmitted message. Adjusting the phase and magnitude on each antenna element can change the direction of each link. As a result, each link is like a wired connection while the placement of this “wire” can be changed quickly.
As point-to-point millimeter-wave physical radios with electronically steerable arrays have been successfully implemented in CMOS with low power and small form factor, it is now possible to bring it to another level and implement a fully infrastructureless network of these radios. This project focuses on the investigation of the MAC and network protocols for millimeter-wave networks that can potentially support multi-gigabit data link connectivity.
Polarization Degrees of freedom
The extra degrees of freedom from polarization effectively increase the number of communication channels between transceivers without increasing the space occupied by transceiver arrays. Exploitation of polarization in wireless systems therefore represents a formidable opportunity, particularly for space-constrained mobiles. Our understanding of the polarization degrees of freedom in multipath fading channels, however, is not adequate. This stems from the inadequacy of physical abstractions in existing channel models.
From physics, the six components of electric and magnetic vector fields are not independent and two scalar equations are sufficient to capture the electrodynamic phenomenon. This suggests that there are only two degrees of freedom from polarization. However, the literature of triaxial induction in geophysical remote sensing, vector-sensor array in signal processing, and MIMO antenna systems in wireless communication speculate the independence of the 6 field components under certain channel conditions. Thus, it is concluded that polarization can potentially provide a six-fold increase in the number of degrees of freedom. In this research, we attempt to understand the different perspectives and unify the conclusions on the polarization degrees of freedom.
[1] A. S. Y. Poon and D. N. C. Tse, "Polarization degrees of freedom," in Proc. IEEE Intl. Symp. Inform. Theory (ISIT), Toronto, Canada, July 2008.
[2] A. S. Y. Poon and D. N. C. Tse, "Degree-of-freedom gain from using polarimetric antenna elements," submitted to IEEE Trans. Information Theory, April 2009.
[3] A. S. Y. Poon and D. N. C. Tse, "Degree-of-freedom gain from using polarimetric antenna elements," Allerton, Oct 2009.
From physics, the six components of electric and magnetic vector fields are not independent and two scalar equations are sufficient to capture the electrodynamic phenomenon. This suggests that there are only two degrees of freedom from polarization. However, the literature of triaxial induction in geophysical remote sensing, vector-sensor array in signal processing, and MIMO antenna systems in wireless communication speculate the independence of the 6 field components under certain channel conditions. Thus, it is concluded that polarization can potentially provide a six-fold increase in the number of degrees of freedom. In this research, we attempt to understand the different perspectives and unify the conclusions on the polarization degrees of freedom.
[1] A. S. Y. Poon and D. N. C. Tse, "Polarization degrees of freedom," in Proc. IEEE Intl. Symp. Inform. Theory (ISIT), Toronto, Canada, July 2008.
[2] A. S. Y. Poon and D. N. C. Tse, "Degree-of-freedom gain from using polarimetric antenna elements," submitted to IEEE Trans. Information Theory, April 2009.
[3] A. S. Y. Poon and D. N. C. Tse, "Degree-of-freedom gain from using polarimetric antenna elements," Allerton, Oct 2009.
Millimeter-wave 3D active imaging system
Funded by FCRP C2S2
Current state-of-the-art millimeter-wave imaging systems reconstruct a 2D image. This project focuses on imaging system that is able to recover both angle-of-arrival and depth information of the target. Holographic algorithms will be applied to reconstruct the 3D image from the angular and the depth information obtained from a 2D wideband antenna array. RF beamformer at 60 GHz will be built in CMOS utilizing the mathematical idea of over-complete expansion and coarse quantization. Sub-array concept will be applied to reduce the number of analog/RF components.
Current state-of-the-art millimeter-wave imaging systems reconstruct a 2D image. This project focuses on imaging system that is able to recover both angle-of-arrival and depth information of the target. Holographic algorithms will be applied to reconstruct the 3D image from the angular and the depth information obtained from a 2D wideband antenna array. RF beamformer at 60 GHz will be built in CMOS utilizing the mathematical idea of over-complete expansion and coarse quantization. Sub-array concept will be applied to reduce the number of analog/RF components.
MIMO Antenna Systems
MIMO antenna systems are usually analysed with an abstract mathematical model that does not capture physical constraints. As a result, theoretical studies do not provide much insight for system design and implementation. In this work, we incorporated electromagnetics to develop a mathematical model [1] that captures the area limitation on the physical placement of antennas and an intuitive way of parameterizing the scattering condition of channel. Based on this model, we performed analyses on the number of spatial degrees of freedom [2], channel capacity [3], and spatial diversity. The mathematical model was validated through experimental channel measurements [4]. The results help assess practical uses of multiple antennas, and shed light on transceiver architectures that would utilize channel scattering more efficiently. A lot of the intuitions developed from this research were applied directly to the design of MIMO antenna systems at 60 GHz in a startup company.
[1] A. S. Y. Poon, R. W. Brodersen, and D. N. C. Tse, "A spatial channel model for multiple-antenna systems," in Proc. IEEE Antennas and Propagat. Symp., vol. 4, Monterey, CA, June 2004, pp. 20ñ25.
[2] A. S. Y. Poon, R. W. Brodersen, and D. N. C. Tse, "Degrees of freedom in multiple-antenna channels: a signal space approach," IEEE Trans. Information Theory, vol. 51, no. 2, pp. 523ñ536, Feb. 2005.
[3] A. S. Y. Poon, D. N. C. Tse, and R. W. Brodersen, "Impact of scattering on the capacity, diversity, and propagation range of multiple-antenna channels," IEEE Trans. Information Theory, vol. 52, no. 3, pp. 1087ñ1100, Mar. 2006.
[4] A. S. Y. Poon and M. Ho, "Indoor multiple-antenna channel characterization from 2 to 8 GHz," in Proc. IEEE Intl. Conf. Commun., vol. 5, Anchorage, AK, May 2003, pp. 3519ñ3523.
[1] A. S. Y. Poon, R. W. Brodersen, and D. N. C. Tse, "A spatial channel model for multiple-antenna systems," in Proc. IEEE Antennas and Propagat. Symp., vol. 4, Monterey, CA, June 2004, pp. 20ñ25.
[2] A. S. Y. Poon, R. W. Brodersen, and D. N. C. Tse, "Degrees of freedom in multiple-antenna channels: a signal space approach," IEEE Trans. Information Theory, vol. 51, no. 2, pp. 523ñ536, Feb. 2005.
[3] A. S. Y. Poon, D. N. C. Tse, and R. W. Brodersen, "Impact of scattering on the capacity, diversity, and propagation range of multiple-antenna channels," IEEE Trans. Information Theory, vol. 52, no. 3, pp. 1087ñ1100, Mar. 2006.
[4] A. S. Y. Poon and M. Ho, "Indoor multiple-antenna channel characterization from 2 to 8 GHz," in Proc. IEEE Intl. Conf. Commun., vol. 5, Anchorage, AK, May 2003, pp. 3519ñ3523.
"Multicore" Baseband Processor
The popularity of wireless communication leads to the proliferation of many air interface standards such as the IEEE 802 families and the 3G families. This evolution of standardization continues in an accelerated manner. Flexible radio architectures that can support not only multiple standards but also upcoming ones, become an important area of research. For the digital part of the radio, domain-specific reconfigurable processors offer the advantage of flexibility as general-purpose processors, and low power by exploiting parallelism in baseband algorithms and providing a direct spatial mapping from algorithms to the architecture, hence reducing the memory and control overhead associated with general-purpose processors. Most existing techniques focus on the computation model of the reconfigurable processor. They, in general, compose of an array of heterogeneous coarse-grained configurable units controlled by either RISC (reduced instruction set computer), VLIW (very long instruction word), or both instruction sets. The granularity of configurable units is usually a word-level operation such as a multiplier, an ALU, or a register.
As the granularity of configurable units directly impacts the energy efficiency of the hardware, this work focuses on increasing the granularity of the configurable units without compromising flexibility. This is carried out by matching the granularity to the degree-of-freedom (DOF) processing in wireless systems. A wireless channel is built upon multiple signal dimensions: time, frequency, and space. The sophistication of a transceiver is measured by its resolvability along these dimensions. For example, a system with bandwidth W, transmission interval T, and number of antennas N has a resolution of 2WTN degrees of freedom. Baseband algorithms are collectively performing channel estimation over these degrees of freedom, and modulation (demodulation) of data symbols onto (from) these degrees of freedom such as DS-CDMA, OFDM, and space-time coding schemes. We abstract operations that are typically performed as per degree-of-freedom
and put them into a single dominant configurable unit. In our design, this unit consists of 4 multipliers, 5 adders, 2 accumulators, 2 shifters, 8 two's complement operations, and 2 multiplexers 
. To ease programming, all pairs of inputs and outputs of the unit have the same latency. Meeting these specifications as well as achieving the throughput requirement requires a design flow that accelerates the exploration of trade-offs among various architectures for the configurable unit. We use the Chip-in-a-Day design flow developed at U. C. Berkeley 
.
The macro-architecture of the baseband processor is shown in
. A prototype of the processor is implemented using the Intel 0.13 um CMOS standard cell library at 1.2 V. The energy efficiency is 95 MOP/mW, which is in the same order as dedicated hardware implementations and is 2 orders of magnitude more power efficient than digital signal processors (DSP).
[1] A. S. Y. Poon, "An energy-efficient reconfigurable baseband processor for wireless communications," IEEE Trans. VLSI Systems, vol. 15, no. 3 pp. 319ñ327, Mar. 2007.
[2] A. S. Y. Poon, "An energy-efficient reconfigurable baseband processor for flexible radios," Proc. IEEE Workshop on Signal Processing Systems, Banff, AB, Canada, pp. 393ñ398, Oct. 2006.
As the granularity of configurable units directly impacts the energy efficiency of the hardware, this work focuses on increasing the granularity of the configurable units without compromising flexibility. This is carried out by matching the granularity to the degree-of-freedom (DOF) processing in wireless systems. A wireless channel is built upon multiple signal dimensions: time, frequency, and space. The sophistication of a transceiver is measured by its resolvability along these dimensions. For example, a system with bandwidth W, transmission interval T, and number of antennas N has a resolution of 2WTN degrees of freedom. Baseband algorithms are collectively performing channel estimation over these degrees of freedom, and modulation (demodulation) of data symbols onto (from) these degrees of freedom such as DS-CDMA, OFDM, and space-time coding schemes. We abstract operations that are typically performed as per degree-of-freedom
(see Figure 1) 
and put them into a single dominant configurable unit. In our design, this unit consists of 4 multipliers, 5 adders, 2 accumulators, 2 shifters, 8 two's complement operations, and 2 multiplexers (see Figure 2)
. To ease programming, all pairs of inputs and outputs of the unit have the same latency. Meeting these specifications as well as achieving the throughput requirement requires a design flow that accelerates the exploration of trade-offs among various architectures for the configurable unit. We use the Chip-in-a-Day design flow developed at U. C. Berkeley (see Figure 3)
.The macro-architecture of the baseband processor is shown in
(see Figure 4)
. A prototype of the processor is implemented using the Intel 0.13 um CMOS standard cell library at 1.2 V. The energy efficiency is 95 MOP/mW, which is in the same order as dedicated hardware implementations and is 2 orders of magnitude more power efficient than digital signal processors (DSP).[1] A. S. Y. Poon, "An energy-efficient reconfigurable baseband processor for wireless communications," IEEE Trans. VLSI Systems, vol. 15, no. 3 pp. 319ñ327, Mar. 2007.
[2] A. S. Y. Poon, "An energy-efficient reconfigurable baseband processor for flexible radios," Proc. IEEE Workshop on Signal Processing Systems, Banff, AB, Canada, pp. 393ñ398, Oct. 2006.
A "Digital-like" RF Beamformer
Funded by FCRP C2S2
We can find a lot of applications for antenna arrays, for example, high-rate point-to-point data link, medical imaging, and ranging. Electronic implementation of these systems always uses the phased-array architecture where only the phase of the signal paths are adjusted. There are several shortcomings of this approach: antennas are needed to be placed ?/2 apart; it cannot deal with multipath; and it requires calibration for both phase and amplitude mismatch. An analog beamformer architecture can solve these problems since both phase and amplitude of the signal paths are adjustable. In this project, we apply the mathematical concept of overcomplete expansion and coarse quantization to derive an analog beamformer that is ìdigital-likeî.
Noticing that beamforming coefficients ui's need not have to expand in only 2 (I and Q) components, we expand the beamforming coefficients over a unit-norm tight frame [1]. The example architecture shown below corresponds to an oversampling rate of 2. As the beamforming coefficients are expanded in a redundant basis, the expansion on each basis element can be coarsely quantized. In the example shown below, we use only 1 bit quantization. With the coarse quantization, the new beamformer architecture can be implemented efficiently in CMOS. In addition, the state of the switches (?1) is controlled by a quantization algorithm running in baseband DSP. The algorithm can be adapted easily to take into account any phase and amplitude mismatch. The whole beamformer involves 3 major building blocks: multiphase generation, mixer bank, and combiners. A prototype receiver at 90 nm IBM CMOS is undertaken.
[1] R. Tseng, A. S. Y. Poon, and Y. Chiu, "A mixed-signal vector modulator for eigen-beamforming receivers," IEEE Trans. Circuits and Systems II, vol. 55, no. 5, pp. 479ñ483, May 2008.
[2] R. Tseng, A. S. Y. Poon, and Y. Chiu, "A mixed-signal MIMO beamforming receiver," Proc. IEEE Radio and Wireless Symposium, Orlando, FL, pp. 327ñ330, Jan. 2008 (invited).
We can find a lot of applications for antenna arrays, for example, high-rate point-to-point data link, medical imaging, and ranging. Electronic implementation of these systems always uses the phased-array architecture where only the phase of the signal paths are adjusted. There are several shortcomings of this approach: antennas are needed to be placed ?/2 apart; it cannot deal with multipath; and it requires calibration for both phase and amplitude mismatch. An analog beamformer architecture can solve these problems since both phase and amplitude of the signal paths are adjustable. In this project, we apply the mathematical concept of overcomplete expansion and coarse quantization to derive an analog beamformer that is ìdigital-likeî.
Noticing that beamforming coefficients ui's need not have to expand in only 2 (I and Q) components, we expand the beamforming coefficients over a unit-norm tight frame [1]. The example architecture shown below corresponds to an oversampling rate of 2. As the beamforming coefficients are expanded in a redundant basis, the expansion on each basis element can be coarsely quantized. In the example shown below, we use only 1 bit quantization. With the coarse quantization, the new beamformer architecture can be implemented efficiently in CMOS. In addition, the state of the switches (?1) is controlled by a quantization algorithm running in baseband DSP. The algorithm can be adapted easily to take into account any phase and amplitude mismatch. The whole beamformer involves 3 major building blocks: multiphase generation, mixer bank, and combiners. A prototype receiver at 90 nm IBM CMOS is undertaken.
[1] R. Tseng, A. S. Y. Poon, and Y. Chiu, "A mixed-signal vector modulator for eigen-beamforming receivers," IEEE Trans. Circuits and Systems II, vol. 55, no. 5, pp. 479ñ483, May 2008.
[2] R. Tseng, A. S. Y. Poon, and Y. Chiu, "A mixed-signal MIMO beamforming receiver," Proc. IEEE Radio and Wireless Symposium, Orlando, FL, pp. 327ñ330, Jan. 2008 (invited).