Energy consumption models for smart-camera networks
[Abstract][Validation experiments][Sample results][Downloads][References]
This page provides additional material and software
for the paper:
Energy
consumption models for smart-camera networks [link]
J. SanMiguel (show email) and A. Cavallaro
IEEE Transactions on Circuits and
Systems for Video Technology (accepted 2016)
Abstract Camera
networks require heavy visual-data processing and high-bandwidth communication.
In this paper, we identify key factors underpinning the development of
resource-aware algorithms and we propose a comprehensive energy consumption
model for the resources employed by smart-camera networks, which are composed
of cameras that process data locally and collaborate with their neighbours. We account for the main parameters that
influence consumption when sensing (framesize and
framerate), processing (dynamic frequency scaling and task load) and
communication (output power and bandwidth) are considered. Next we define an
abstraction based on clock frequency and duty cycle that accounts for active,
idle and sleep operational states. We demonstrate the importance of the
proposed model for a multi-camera tracking task and show how one may
significantly reduce consumption with only minor performance degradation when
choosing to operate with an appropriately reduced hardware capacity. Moreover,
we quantify the dependency on local computation resources and on bandwidth
availability. The proposed consumption model can be easily adjusted to account
for new platforms, thus providing a valuable tool for the design of
resource-aware algorithms and further research in resource-aware camera
networks.
Validation with power measurements
To validate our model we have performed a set of
experiments using measurements obtained from the battery discharge of a
smart-camera system running video applications on an Ubuntu OS 14.04 64bit.
We emulated a high-end smart camera with the
following devices (see Fig. 1): ·
USB QuickCam Ultra Vision for sensing ·
Toshiba Portege R-700
i5-450M for processing ·
AC600 Wireless Dual
Band USB Adapter for communication. Battery readings are obtained from the OS kernel
system file /sys/class/power_supply/BAT0/uevent
where the discharge over time has been extracted each 5 seconds. |
|
These
experiments have been divided in two phases:
1.
Get the measurements and model fitting using
Eq. (4) using three UNIX C++ programs for sensing, processing and communication
2.
Design and development of a video application for
multi-target people tracking based on the upper-body parts of people [1] which
transmits visual descriptors that can be used for people re-identification
purposes [2]. The OpenCV library was used to
implement the visual analysis algorithms.
A detailed
description of the protocol to get the measurements is available here.
Selected results that can
be extracted from the measurements obtained: ·
Sensing power (runs): [Active]
[Idle] [Comparison] ·
Processing power (runs): [Active]
[Idle] [Comparison] ·
Communication power (runs): [Comparison] ·
Video application (runs): [Active] Other statistics that can be obtained from the
extracted data: ·
Sensing utilization (runs): [Active] ·
Communication bitrate: [Active] ·
Video application: [CPU
utilization] [Overall
activation time] |
Comparison with related state-of-the-art:
|
Downloads (Please cite this paper if you use this material)
The software employed
in this section can be freely used for research purposes. [Github]
The Unix programs employed and the measurements
obtained:
(tested on Ubuntu OS 14.04 64bit and OpenCV 3.1.01)
·
Software & bash scripts to execute the experiments: [github]
[local
copy] 1.8 MB (18/07/16)
·
Data measurements: [github]
[local copy]
1.3 MB
(18/07/16)
Matlab
scripts to generate the figures presented in the paper [github] [local copy] 8.5 MB
(18/07/16)
1You can get the OpenCV Library here
[1] R. Vezzani, D. Baltieri, and R. Cucchiara, “People reidentification
in surveillance and forensics: A survey,” ACM Computing Surveys (CSUR), vol.
46, no. 2, p. 29, 2013.
[2] R. Mazzon, S. Tahir, and A. Cavallaro, “Person
re-identification in crowd,” Pattern Recognition Letters, vol. 33, no. 14, pp.
1828–1837, 2012.
[3] Y. Sharrab and N. Sarhan, “Aggregate
power consumption modeling of live video streaming Syst.” in ACM Conf. on
Multimedia Syst. (MMSys), Feb. 2013, pp. 60–71.
[4] A. Wang and
A. Chandrakasan, “Energy-efficient DSPs for wireless
sensor networks,” IEEE Signal Process. Mag., vol. 19, no. 4, pp. 68–78, Jul.
2002.
[5] Z. He and D.
Wu, “Resource allocation and performance analysis of wireless video sensors,”
IEEE Trans. Circuits Syst. Video Technol.,vol. 16,
no. 5, pp. 590–599, May 2006.
[6] A. Redondi, D. Buranapanichkit, M. Cesana, M. Tagliasacchi, and Y. Andreopoulos, “Energy Consumption of Visual Sensor
Networks: Impact of Spatio-Temporal Coverage Based on
Single-Hop Topologies,” IEEE Trans. Circuits Syst. Video Technol., vol. 24, no.
12, pp. 2117–2131, Dec. 2014.