3 channel dependency-based power model for mobile AMOLED displays

Active matrix organic light-emitting diode (AMOLED) displays are being adopted in increasing number of smartphones. Most applications are now based on consistent user interfaces, such as games and instant messaging, and it is therefore crucial to efficiently manage the power consumption of the AMOLE...

Full description

Saved in:
Bibliographic Details
Published in:2017 54th ACM/EDAC/IEEE Design Automation Conference (DAC) pp. 1 - 6
Main Authors: Seongwoo Hong, Suk-Won Kim, Young-Jin Kim
Format: Conference Proceeding
Language:English
Published: IEEE 01.06.2017
Subjects:
Accuracy
Active matrix organic light emitting diodes
AMOLED Display
Channel
Dependency
Error analysis
Mobile communication
Power demand
Power measurement
Power Model
Smart phones
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Active matrix organic light-emitting diode (AMOLED) displays are being adopted in increasing number of smartphones. Most applications are now based on consistent user interfaces, such as games and instant messaging, and it is therefore crucial to efficiently manage the power consumption of the AMOLED display. To this end, an accurate power model for this type of display is needed. The prior work available in the literature did not consider the dependencies between the red, green, and blue (RGB) channels, or only accounted for limited dependencies between the channels. In this paper, we develop a novel accurate power model for AMOLED displays that considers all possible dependencies between the three channels. We applied this model to four panels from the Galaxy S1, Galaxy S3, and Galaxy S5 from SAMSUNG, and a CHIMEI panel with different customizations. Using four well-known image datasets, the proposed model shows that the smallest error rate is on average 1.13% for the Galaxy S5 panel, while both the simple model and the existing dependency models represent average error rates of 9.69% and 5.53%, respectively. As a result, the proposed model is shown to be correct and generally usable. In these tests, this model was implemented using the vector floating point support of the Cortex-A9 on an Android smartphone at the operating system level. The computation time required for each test image that had a resolution of 720 × 1280 was 3 seconds.
DOI:10.1145/3061639.3062181