1. Energy-aware cross-layer scheduler
for reliable MPEG4 video transmission
over HCF 802.11e WLANs
IMEC’s Cross-Layer Technology Program
IMEC’s cross-layer (XL) Technology Program aims at drastically improving the
quality of experience (QoE) for future mobile multimedia services, by offering Gap
to the users the best trade-off between energy consumption and multimedia Energy requirement
quality. Handheld devices being battery-powered, the performance
Energy available
requirements of new applications and technologies are indeed coupled with
in battery
severe constraints on energy efficiency. This is becoming a key concern: there
exists a continuously growing gap between the available energy, resulting Time
from battery technology evolution, and the steeply increasing energy
requirements of emerging radio systems (top figure on the right).
Propagation
requirements
conditions
To bridge this energy gap, the XL Technology Program designs cross-layer
Application
control solutions for QoS-energy management of reconfigurable wireless
multimedia systems, The goal of the designed control solutions is to minimize
t t
the total energy consumption in the system, while providing the required QoS
expected by the user. To that purpose, the scalability in the reconfigurable Cross-layer optimized
radio terminal is exploited by a cross-layer controller that follows at run-time Run-time Performance/
Energy manager
the dynamics in the application requirements and propagation conditions to
achieve low power operation (bottom figure on the right). Energy-scalable SDR
Baseband Front-end
WBA project: Instantiation in a video-surveillance scenario
In the WBA project, the proposed approach was instantiated in a video-surveillance scenario. The
considered scenario consists more specifically of multiple cameras sending MPEG4 video traffic to
the access point (AP) of an 802.11e wireless local area network (WLAN), considering the hybrid
coordination function (HCF). Different QoS levels (here in terms of video quality, at fixed resolution
and frame rate) can be required for the different cameras.
AP-controlled MPEG4 Data
simple profile
TXOP
UDP/IP
Node 1 Node 4
Energy-Aware
AP
AP MAC
Scheduler
QoE Manager
QoE Manager 802.11e
PHY
802.11a
Node 2 Node 3
(Channel Access Grant)
Based on the approach introduced above, a cross-layer multi-user video scheduler was developed.
The goal of this run-time cross-layer scheduler is twofold:
to ensure the reliable and timely delivery of the different real-time video streams
to minimize overall energy consumption in the different camera devices.
The proposed solution is a controller located at the AP, which steers the whole WLAN network,
including the camera devices. This run-time adaptation is carried out with respect to the dynamics
present in the system:
the instantaneous wireless link attenuation between the different cameras and the AP.
the instantaneous data rate demands of the different cameras.
Based on this information and capitalizing on appropriate design-time performance-energy modeling
of the whole system, the proposed controller can then decide at run-time on a regular basis how to
allocate
channel resources to the different devices
available knobs (i.e., run-time controllable parameters) in these devices (e.g., constellation order,
code rate, output power)

2. Design-time/run-time
approach
Design-time optimization Run-time control Design-time: system-level
performance-energy
Pathloss
Channel State
QoE manager
characterization, leading to
Pareto optimal tradeoff curves
Request Alg or ith m un it
between energy consumption
Netw ork cell
state metrics
Selected
configuration
and performance, as a function
Configuration Cost Resource Metric Cost & resource
of the system configuration. This
Access metrics,
Configuration Cost Resource Metric
Pool
Configuration
is done based on an end-to-end
Configuration Cost Resource I n te rfa ce un it D a ta b a se un it Load
database network simulation framework.
Cost
e.g. e.g.
Run-time: Low-overhead run-
Channel state,
Requests,
...
baseband
configuration,
front-end
configuration,
time QoE-power manager
...
selecting optimal operation
Resource points on the tradeoff curves
taking into account run-time
application/user requirements
and environment conditions.
Demonstration framework
In order to support the validation of the
XL approach, an end-to-end real-time Camera
demonstration framework was built. Wireless
It consists of a full video data path Network
(from the capture and encoding of
video data at the transmitter side, to
the decoding and playing at the Control Room
receiver side), with transmission over
an NS-2 simulated wireless network
(Figure on the right). GUI
Ns-2
The proposed solution integrates the
NS-2 simulated wireless network into a
XL Manager Network
real-life IP stack, supporting real-time simulator
Encoders
network traffic.
Benefits
Conventional energy management
techniques in the present context
belong to two major categories,
namely sleeping (MAC-centric) and
scaling (PHY-centric). They are
considered as reference point to
benchmark our approach.
The XL approach enables to jointly
take advantage of these two
contradictory approaches, which
XL present a tradeoff in minimizing
the overall system energy .
By combining these two
approaches and leveraging
scalability in the devices, the XL
approach was shown to save up to
a factor 3 in energy consumption
(i.e., battery lifetime) when
compared to state-of-the art-
approaches.