My research page
Utilization of FPGA co-processor in Wireless Sensor platform
Wireless sensor networks which firstly were designed for specific domain such like military field now have been extended into a wide variety of applications: healthcare, security, smart homes, intelligent robots, entertainment, nature observation, etc. As various and complicated applications have been introduce to the WSNs, it becomes increasingly necessary to enable sensor node with both higher performance and flexibility to meet those vastly varying requirements and characteristics.
Our research of utilizing FPGA as coprocessor in Wireless sensor platform is focusing on exploring “best” SW/HW co-design for WSN application. “Best” means providing a performance enhancement with acceptable power consumption, cost and complexity.
First, we did survey on existing FPGAs-based/FPGAs-assisted sensor platform. FPGA-assisted platform uses FPGAs and MCU at the same time. FPGAs-based sensor platform uses only FPGA and implements soft-core processor and some peripherals in it. Considering the cost and energy, we choose FPGAs-assisted model. Then, for evaluating the speed up, power consumption, we use AVRORA with its components to estimate time in cycles and power consumption of Atmel microprocessor and XPower or VPR for FPGAs. Evaluating the cost of communication is to be achieved later.
Now, we are identifying the existing applications could be accelerated and algorithm was physically restricted on Microprocessor based platform. After this, we will do research based on those to find suitable SW/HW partitioning to implement and accelerate those applications.
Reading List
SOPC design for sensor platforms:
Title: Nokia Wrist Attached Sensor Platform: Wearable FPGA Based Wireless Sensor Platform
Year: Feb. 2008
Author: Tom Ahola from Nokia Research Center, Finland
Citation: http://opensource.nokia.com/NWSPTitle: Atific Helicopter High Performance Multi-Radio WSN Platform
Year: May 2006
Author: Mikko Kohvakka from Atific co., Finland
Citation: http://atific.fi/en/index.php?page=CustomWirelessComments:
This is a whitepaper from Atific co., which introduced the first commercial FPGA-based platform for Wireless Sensor Network. The name helicopter comes from four removable independent Nordic Semiconductor nRF2401A 2.4 GHz radio transceivers with antennas which could be used to complete intensive applications on multi-hop video and audio streaming.
As a development kit, the FPGA used in this platform provides valuable reconfigurability. Just in one Altera Cyclone EP1C20, we can change the implemented NIOS II processer count as we need to realize parallel processing for high performance. Moreover, the high flexibility enables us to connect or change various sensor modules easily. Indeed, Atific Helicopter development kit gives a fast and easy start for the wireless sensor network development. Now the only problem is, this platform obviously is a performance-oriented but an energy-efficient one which may become an obstacle for several applications.
Applications and related algorithm:
Smart Home and Health care:
Title: A Prototype on RFID and Sensor Networks for Elder Healthcare: Progress Report
Year: Aug. 2005
Author: Loc Ho from Venturi Wireless, Melody Moh & Zachary Walker from San Jose State University, Takeo Hamada & Ching-Fong Su from Fujitsu Laboratories of America.
Citation: SIGCOMM’05 WorkshopsComments:
Title: The Design and Implementation of Smart Sensor-based Home Networks
Author: Changsu Suh .etc from Ajou University, Republic of Korea
Comments:
Positioning and Tracking algorithm:
Title: IMPLEMENTATION STRATEGIES FOR PARTICLE FILTER BASED TARGET TRACKING
Year: May 2007
Author: Rajbabu Velmurugan from Georgia Institute of Technology
Citation: Thesis
Comments: Particle filters provide robust positioning and tracking of moving objects. But for most processors in sensor platform, it is too intensive to run. So we want to figure out how to implement PF in HW that may make it more practical. This PhD thesis tries to implement PF in FPGA, DSP and Analog which could be a good theoretical and experimental reference.
Title: Object Tracking Circuit using Particle Filter with Multiple Features
Year: Oct. 2006
Author: Jung Uk Cho .etc from Sungkyunkwan University, Korea
Citation: SICE-ICASE International Joint Conference 2006
Comments:
Temperature and Humidity:
Gyroscope and Accelerometer:
*This website is being updated