|
INTRODUCTION
ELPL (Embedded Low-Power Laboratory) is one of the leading groups in power and energy optimization for embedded systems. ELPL has introduced innovative techniques for device- and system-level energy measurement and estimation, LCD power reduction, low-power SDRAM and flash memory systems, FPGA power minimization, practical issues on dynamic voltage scaling, fuel cell and battery hybrid power source for portable systems, and so on. ELPL has been regularly contributing to relevant journals, magazines, conferences, symposia and workshops since 2000, publishing over 50 technical papers in this area. ELPL aims at practically applicable low-power design techniques, and all the developed techniques have been demonstrated with working prototypes at ISLPED design contest and DAC University Booth.
[Notice] (Aug. 10, 2011) Application Information for Foreign Students
|
|
 Publication (Randomly selected)
|
|
|
|
 |
|
|
[ABSTRACT]
Recent wireless sensor nodes, equipped with ultra-low-power (ULP)
RISC microcontrollers, do not generally support DVS (dynamic
voltage scaling), though the ULP microcontrollers have ideal energyvoltage-
frequency characteristics for DVS. In general, an outputadjustable
DC-DC converter is hardly affordable in such sensor
nodes, and surprisingly light current consumption makes the DCDC
converter operate in a very inefficient region.
This paper introduces a new supply voltage scaling namely PVS
(passive voltage scaling) that eliminates the use of a DC-DC converter.
While a battery is directly connected to a ULP microcontroller,
PVS monitors the battery voltage drop (50% in alkaline batteries)
and scales down the clock frequency accordingly for a reliable
operation. Opposite to DVS, PVS does not achieve the optimal
energy consumption of the microcontroller, but it minimizes
the loss of the power delivery system, peripherals and battery, and
achieves ultimate longer operational lifetime.
Along with such energy gain, the throughput of a PVS-enabled
sensor node continuously decreases by the battery state of charge
loss, which makes a traditional performance-driven routing also
energy-aware. We applied PVS to a wireless sensor network, and
demonstrated 34% more network lifetime, and 26% shorter average
latency over a modern sensor network such as Telos [1],
64% more network lifetime and 3% shorter average latency over a
traditional sensor network with a DC-DC converter.
|
|
|
|
|
|
 |
 |
|
|
 |
|
|
|
 |
|
|
|
 |
| [Conference] |
|
| Naehyuck Chang, Wook Hyun Kwon and Seungkweon Jeong, "Response time driven scheduling for real-time programmable controllers with network-based I/O systems," in Algorithm and Architecture for Real Time Control Cancun, Mexico (AAFTC), pp. 41-46, 1998. |
|
|
 |
|
|
