Transcript Slides

Green WLAN: On-Demand WLAN
Infrastructures
By: A.P.Jardosh (monet 2009)
Presented by: Rojeena Bajracharya
1
Content
• Introduction
• SEAR
– Green clustering;
– User demand estimation;
– Topology management;
– User management.
• Evalutation
• Conclusion
2
Introduction
• Wireless Local Area Networks :
– essential tool for a flexible service;
– centralized management;
•
•
•
•
•
Dense WLANs with redundant layers
Objective: to meet user demand during peak time.
Fact/Problem: peak times rarely occur.
APs remain idle energy waste 
Solution: adoption of highly efficient resource
management strategies that depend on user demand.
• In practice: SEAR.
3
Introduction
• Resource on-demand strategy: efficiently
manage WLANs resources, provide a high
quality service.
Demand-driven<-> schedule-driven.
• RoD strategies must fulfill three requirements:
– Ensure coverage;
– Maintain client performance;
– Avoid frequent client re-associations.
4
SEAR
• Survey ,Evaluate ,Adapt and Repeat.
• SEAR resides on the central controller of a
centralized WLAN.
– it has a complete knowledge of the physical position
and state of APs.
– it controls efficiently all APs.
• Four steps:
1. Green clustering;
2. User demand estimation;
3. Topology management;
4. User management.
5
SEAR
• Components :
6
Green clustering
• Objective: form clusters
of APs.
• Idea: only one AP within
each cluster can ensure
basic coverage.
• Two steps:
1.
2.
Neighborhood discovery
Cluster formation
• One AP within each
cluster is named as the
cluster's head and
powered on all the time.
7
Green clustering
1.
Neighborhood discovery
where Nthresh and Sthresh are pre-determined thresholds for the median number of beacons
per second and the median signal strength of the beacons, respectively sets to 9 and -50dB,
respectively
2.
Cluster formation:
 d-cluster condition :
 use the AP with the maximum ni add i as the first member of the cluster, and simultaneously
remove i from all neighborhood sets of other APs.
 Once i is added and Cv is formed, step through all the APs j in the
neighborhood set Ri and add them to Cv as long as every new j that is added to Cv is in the
neighborhood set of every other AP already added to Cv.
APs that form a cluster are removed from the neighborhood sets of all other APs; their
updated ni values.
8
Demand Estimation
• Objective: Estimate user demand within each cluster.
• Idea: helps SEAR making strategic decisions.
• In practice: channel utilization because it encompasses
user count as well as data-rates.
– Channel utilization is defined as the percentage of time the
medium remains busy due to the transmission of bytes in the
network or due to inter-frame spacing.
– Each AP in the WLAN continuously sniffs MAC layer data and
control frames transmitted by all the clients and APs in its
vicinity on the same channel, and computes both the aggregate
channel utilization of the medium in the vicinity of each AP, and
the channel utilization per client connected to that AP
9
Topology management
• Objective: reconfigure the network by
powering on or off Aps.
• In practice: done at regular time intervals with
the threshold value of channel utilization as a
decision parameter.
10
User management
• Objective: reduce excessive roaming between
APs.
– re-associations (incre)performances (decre)
• determines the quality of service.
• In practice:
1. Load Balancing :
– Client with greatest traffic load in list are moved first.
– move clients from AP i to AP j until: U′i ≤ 1/2 × Ui and
Uj ≤ Tthresh.
2. Enforcing user Association: blacklist or 802.11v
11
Evaluation
Does SEAR satisfies each requirements?
Two WLANs infrastructures:
• 15 APs and 9 clients, two adjacent doors;
evaluate green clustering on coverage,
throughput
determine energy savings.
• 3 APs and 9 clients, same room;
evaluate user association management;
• What are the results ?
12
Evaluation
Client connectivity:
each client receives connectivity from at least one
AP.
each client receives the same throughput
whatever AP is powered on.
one AP per cluster is sufficient to provide basic
coverage.
13
Evaluation
Throughput of each client with the 15 APs in a IEEE 802.11bWLAN. The bars
indicate throughput between 0 and 6 Mbps. The vertical dotted lines delineate
the seven clusters of APs
14
Evaluation
• Throughput:
 small decrease in the average throughput.
Nine clients’ load in a 15 AP WLAN. (b)Average throughput received by
the nine clients, without and with SEAR.
15
Evaluation
• Aggregate offered load
and throughput
received by nine clients
during the 1-hour
(a) low traffic interval, and
(b) high traffic interval
16
Evaluation
• Power savings:
– Depends on a few parameters (especially a
channel utilization threshold).
– threshold of 60%  cut in energy Consumption
of 46%.
17
Conclusion
• What are the conclusions?
1. powering off APs has minimal impact on client
performances;
2. extra APs are only necessary during peak times;
Resource on-demand strategies lead to
energy savings.
18