From Low-power systems to Design Automation of Things 

Dr. Chang is pursuing the following initiatives based on his in-depth and variety interdisciplinary work ranging from chemistry to mechanical engineering, but not limited to:

  • Design automation of things 
    • Design automation of power-aware electric vehicles
    • Large-scale battery management
    • Smart buildings and Smart Grids
    • Large-scale energy harvesting including solar and thermal energy
  • Application-specific cyberphysical system design and optimization
    • Electric vehicle conversion of extreme off-road vehicles
    • Energy harvesting for IoT
    • IoT applications of UV skin damage protection
  • Embedded low-power systems
    • Power management of computing systems
    • Power management of memory and I/O subsystems

Fig. 1. Embedded, Cyberphysical and low-power systems built by Dr. Chang from 1997 (bottom) to 2015 (top). Cybersystem side only for demonstration purpose.
 Power consumption has become one of the most critical issues in modern and future Cyberphysical systems. Most low-power system research in the past focused on the CPU and memory, and system-level design was not the mainstream concern of low-power systems until the early 2000s. Abstraction has been a fundamental philosophy that has been driving the evolution of computing systems and also extended to Cyberphysical systems. However, abstraction ends up with layered optimization and is not always appropriate for low-power systems because aggregation of the best low-power components does not always result in the most efficient low-power systems. 

Dr. Chang is one of the earliest researchers to have made a significant, fundamental, and lasting impact on system-level low-power systems. He pioneered and led holistic low-power design and management methodologies for the last 17 years and initiated several ground-breaking design methodologies that have had substantial practical and academic impact. 

Dr. Chang has shown power consumption due to the I/O (input and output) subsystem and in-system power conversion/delivery loss actually dominate the overall system power consumption especially for portable and personal computing systems. Dr. Chang has provided fundamental solutions for low-power display, memory, audio, buses and thus the whole embedded systems. 

From the mid 2000, Dr. Chang extended the low-power design focus from power consumers (CPU, memory, I/O, etc.) to suppliers (power sources, power converters, etc.) and their joint optimization. Dr. Chang initiated joint power optimization of embedded systems with fuel cells, Photovoltaic cells, and primary batteries. Dr. Chang also introduces power management considering the DC-DC converter efficiency that largely varies by the input/output voltage difference and the load current. 

Through the above mentioned research, Dr. Chang has been building innovative concept of a series of new interdisciplinary systems. Dr. Chang i) designs and implements the baseline prototype from the beginning stage of the research to correctly identify the problem. Dr. Chang ii) creates system models including power, performance, temperature and other important state variables by measurement of the prototypes, iii) applies design automation methodologies for fast and quality solutions, and iv) finally verifies the results again with the optimized prototypes. 

Fig. 1 visualizes Dr. Chang’s system design and implementation results since 1997. He demonstrated the prototypes at Design Automation Conference University Booth since 2000. His contributions to the University Booth were introduced in People of ACM on March 19, 2015. Dr. Chang won the ACM/IEEE ISLPED Design Contest Awards an impressive six times. 






The World’s First Dynamic Backlight Scaling of Liquid Cristal Display

Dr. Chang’s these research streams are all based on in-depth interdisciplinary research starting from the domain-specific characterization and systematic modeling. For instance, the world’s first dynamic backlight scaling (DLS) of color TFT LCD (thin-film transistor liquid crystal display) is based on the principle of operation of the color TFT LCD panels (Fig. 2 left.) This is a truly innovative attempt that keeps the same screen brightness even with a dimmed  backlight for power saving, by the use of dynamic color compensation of the pixels with a cross-layer hardware support. This methods commonly allows up to 60% saying from the total system power of portable electronics [1]. There are 480+ Google citations on DLS (first three papers), and he won the ISLPED Design Contest Awards in 2002 and 2004. His innovation has been practically applied in industry, including Freescale’s XEC-LCD, the MPEG-21 Standard Digital Item Adaptation, and modern smartphones. In addition, Dr. Chang’s LCD variable refresh rate became the fundamental of the iPad Pro’s variable refresh rate and Nvidia’s G-sync as well.


          

Fig. 2. Dynamic LCD backlight scaling, Bluetooth handsfree for severe disables, ocean bottom seismometer, hydrogen fuel cell, direct methanol fuel cell and a battery hybrid (from left to right.) 


Fuel Cell and Battery Hybrid Architecture and Joint Power Management with the Load Systems

Dr. Chang introduced the a totally new concept of a fuel cell and battery hybrid for portable computing. This work is one of the earliest joint optimization of the power source and computing power management. He showed how dynamic power management should be altered when the power source is not ideal, demonstrating up to 40% fuel savings [4].

The fuel cell and battery hybrid systems are based on the characteristics of hydrogen and direct methanol fuel cell stacks, and their BOP (balance of plant) as well as Li-ion battery characteristics (Fig. 2 right.) Both hydrogen and direct methanol fuel cells cannot dynamically follow the load current, especially for embedded systems because their power consumption of electronics circuits varies orders of magnitude faster compared with the fuel cell frequency response. Therefore, the system should be equipped with an unnecessarily big fuel cell to prepare for the worst-case power demand. 

Dr. Chang introduced fuel cell and battery hybrid such that the fuel cell is the the primary power source and directly supplies power to the load while managing a small-capacity battery as an energy buffer. Dr. Chang dramatically reduced the battery size as well as increased the fuel cell efficiency with this innovative concept. Such a hybrid architecture was new to the fuel-cell society in 2005, where a fuel cell was generally used as a battery charger, the battery is the primary power source,  and thus the systems had to maintain a high-capacity battery. 

Dr. Chang also introduces joint optimization of the battery charing current management and the load-side power management. This is a flagship showcase of a power supply and a load device joint power management; how conventional DVFS (Dynamic Voltage and Frequency Scaling) and DPM (Dynamic Power management) policies should be changed for the fuel-cell and battery hybrid power sources to maximize the overall system efficiency. Dr. Chang demonstrated actual efficiency enhancement with a real system demonstration. He transferred a real-time software control for the constant-current operation of direct methanol fuel cell to Samsung Electronics, and the prototype won the 2007 ISLPED Design Contest Award.


The World’s First Computer-Memory-Hierarchy-Inspired Hybrid Energy Storage Systems (HESS)

Dr. Chang is a pioneer of energy efficiency enhancement for beyond embedded systems based on legacy power management practices. He discovered a truly interdisciplinary research paradigm that contributes to both electronics/computer communities and the chemistry application community. Dr. Chang improved the fuel-cell and battery hybrid idea to a general system design framework for multiple heterogeneous power source systems, which is the world’s first computer-memory-hierarchy-inspired hybrid energy storage system (HESS) [6] (see Fig. 3.)

In the past, ESS (energy storage systems) have been driven by chemists and physicists who mainly contributed to the new battery technologies. People anticipated the ultimate battery technology for high performance HESS (high density, high power capacity, low-cost, long life, etc.) However, battery technologies almost hit the theoretical limits just like semiconductor memory devices. It takes a long time to commercialize a new storage technology even the proof of concept has been confirmed. 

Computer systems has been using a memory hierarchy and memory management composing the memory systems with heterogenous memory technologies such as cache, main memory and the secondary storage, over decades. The HESS idea is to compose an ESS with multiple heterogenous storage technologies, connecting them via a change transfer interconnect (CTI), and exploit advantages and hide disadvantages of each storage technology. Dr. Chang accomplished the whore HESS design framework ranging from the CTI architecture to charge management policies.


            
                                                                                              
Fig. 3. The world’s first computer-memory-hierarchy-inspired hybrid energy storage system (HESS.)

      
                     

Fig. 4. Modules of the HESS (the main controller to and from the AC power line, the supercapacitor bank front, the supercapacitor bank back, and the charge transfer interconnect (CTI) module - from top right to clockwise.)


Dr. Chang started from precise and dynamic power characterization of various storage technologies and power converters reflecting their operating conditions. He built real working HESS prototypes and confirmed the proposed methodologies. Dr. Chang finally introduced the complete design and management framework for HESS.

Dr. Chang won the Mid-Career Research Grant from National Research Foundation (NRF) of Korean Government from 2010 to 1015 with the amount of $1.3M, which is the second biggest personal research grant from NRF (NRF gives a few biggest personal grant per year for all the natural since and engineering area.) 

Dr. Chang brought the whole HESS prototype from Korea to the US and demonstrated at the Design Automation Conference (DAC) University Booth. HESS prototype won the 2012 ISLPED Design Contest Award.


Dynamic Reconfiguration of Non-Ideal Power Source Arrays

Dr. Chang demonstrated how a reconfigurable interconnect network can resolve fundamental limitations of the solar panels. He also introduced its automotive applications. This work won the ISLPED 2014 Best Paper Award. He also introduced solar energy harvesting without a power converter nor an energy storage for extreme low-cost and small form-factor wearable computing devices, which is being commercialized. This work won the ISLPED Design Contest Awards 2014

Fig. 5 shows a novel reconfigurable switch network for non-ideal power sources. He introduced a reconfigurable photovoltaic cell array to overcome partial shading and permanent cell damage. He also invented a linear complexity switch network for Solid Oxide fuel cells (SOFC) to extend the array lifetime against imbalanced cell aging. 

The reconfiguration switch network can dynamically modify the array dimension (series and parallel connections including irregular arrays) including the number of rows and columns. Some damaged or shaded cells can be bypassed. The two-level hierarchy switch network for SOFC array can compose multiple regular arrays against aged cells in arbitrary locations with a linear switch complexity. Dr. Chang transferred the SOFC reconfiguration technology to SK Innovation Co. Ltd.

Dr. Chang also applied the reconfigurable Photovoltaic cell array to partial solar-powered electric vehicles. Fig. 5 bottom right shows a wireless sensor node for solar irradiance measurement that is attached a vehicle panel with a magnet mount. Dr. Chang collected solar irradiance change over time and location attaching the sensor node to a vehicle hood, a rooftop, quarter panels, and door panels. He finally devised a method to deploy Photovoltaic cell array to all the vehicle panels and perform dynamic reconfiguration against to the solar irradiance changes on each panel (Fig. 5 bottom left.) 


        

Fig. 5. Dynamically configurable Photovoltaic (PV) array prototype back, front, wireless solar irradiance profiler for vehicles, and a partial solar powered electric vehicle with dynamic PV reconfiguration (from top right to clockwise.)   



Initiating and Leading the Design Automation of Things (DAoT)

Dr. Chang has been leading a new initiative of the electronics design automation (EDA) community as the ACM SIGDA (Special Interest Group on Design Automation) and currently the Past Chair, the Editor-in-Chief of ACM Transactions on Design Automation of Electronics Systems (TODAES), and Technical Program Chair of ASP-DAC (Asia and South Pacific Design Automation Conference) 2015 and DAC 2016, as well as an active individual researcher. The EDA community is actively expanding the application areas of design automation, aka Design Automation of Things (DAoT.) Dr. Chang initiated and leads DAoT where the Thing is low-power and low-energy electric vehicles [7]. 


  
  
  

Fig. 6. A custom electric vehicle for Design Automation of low-power electric vehicle research.


Dr. Chang attempts the same systematic system-level low-power design methods for legacy low-power systems. Dr. Chang introduced novel power power modeling, power estimation, remaining range estimation, design-time power and energy optimization, and runtime power and energy optimization, and eventually low-power and low-energy electric vehicle synthesis. This framework is the first representative work of Design Automation of Things (DAoT). 

Dr. Chang again applied his own research style for the DAoT of electric vehicles. Dr. Chang started from actual fabrication of a custom electric vehicle for power modeling and performance verification (Fig. 6.) This vehicle enables Dr. Chang to derive accurate instantaneous power modeling of electric vehicle with a function of the weight, instantaneous velocity, instantaneous acceleration, road slop, regenerative braking current, and so forth. 


        
 

Fig. 7. A platform of a medium to high speed electric vehicles for education and research.


Dr. Chang won a new three-year (3+3, which is a revised format of the previous 5 year format) NRF mid-career grant for this research immediately after the NRF five-year mid-career grant for HESS, which is very rare. Dr. Chang is going to expand the DAoT of electric vehicles to a smart mobility infrastructure combined with ESS and Smart Grid. Dr. Chang believes his low-power and low-energy design framework will be an essential component of future autonomous driving.

Dr. Chang is also fabricating the second-generation of custom electric vehicles (Fig. 7.) This electric vehicle has up to a 70+mph top speed. The first version shown in Fig. 6 is a low-speed electric vehicle and thus cannot accurately explain the aerodynamics resistance. Dr. Chang gives a strong emphasis to help other research groups who want to join the electric and smart cars but do not have an adequate platform nor a fabrication technology. Dr. Chang fabricates and provides custom electric vehicles, which are completely transparent, fully customizable and easy to use, and provide them to two university research groups, a group at Seoul National University of Technology and Jeju National University who are working on motor control and battery charging, respectively. Dr. Chang is going to develop and distribute a complete assembly kit of an electric vehicle for K9 students to graduate school research. 

Dr. Chang is also working on application-specific electric vehicle design and optimization. Dr. Chang is focusing on extreme off-road electric vehicles, which require totally different criteria compared with common on-road electric vehicles. Dr. Chang recently performed an electric vehicle conversion of a heavily modified Jeep JK Wrangler with long-arm suspension lift and 37” tires (Fig. 8.) Dr. Chang removed the engine, transmission and related fuel and exhaust subsystems, but installed electric traction motors, motor controllers, a high-voltage 28 kWh Nickel Lithium Cobalt Oxide (NMC) battery pack, a battery management system, an on-board charger, an electric power steering pump, an electric brake vacuum bolster, a custom instrument panel with a data logger via CAN (control area network) as shown in Fig. 8. The electric extreme off-road vehicles give distinct advantages over internal combustion engine vehicles beyond going green. They provide stealth operation capability in terms of sound and heat, which is ideal for natural habitats and military operations. More importantly, the flat and fast torque response of electric motors enable superior off-road driving (rock crawling) capability (Fig. 9.)




Fig. 8. Electric vehicle conversion of Jeep JK Wrangler with an extreme off-road modification.


   

Fig. 9. Demonstration of the off-road capability of the electric conversion of a Jeep JK Wrangler.


Other Interdisciplinary System Design Activities

Together with the above research activities, Dr. Chang introduced an extremely low overhead system-level clock-cycle-accurate power characterization tools. He introduced the unique cycle-accurate energy measurement principle and operation and developed working tools [2]. The memory power tool has been delivered to Samsung Electronics. 

In addition, Dr. Chang worked on low-power unmanned aerial aircraft, ocean bottom seismometers, advanced automotive electrical control units, a self-powered battery-less cumulative UV exposure detect sensor system to prevent skin damage, etc. 


New Initiatives in Interdisciplinary System Design Activities

Dr. Chang recently became a Co-Pi of a new Engineering Research Center (ERC) in Korea for flexible Thermoelectric Generators (TEG). Dr. Chang is leading the application group of the ERC and working for vehicle thermal energy harvesting, which mandates complete knowledge about the vehicle exhaust systems. Dr. Chang is installing TEG arrays to the vehicle exhaust system.

Dr. Chang recently started collaboration with a battery company, Kokam Co. Ltd. (www.kokam.com) for battery pack integration, thermal management, cell balancing, and protection for a large-scale applications including electric vehicles, electric boats, ESS (energy storage systems) for vessels, etc.