Upload
erica-short
View
217
Download
1
Embed Size (px)
Citation preview
Architectures and Algorithms for Future Wireless Local Area Networks 110987654321Chapter
Architectures and Algorithms for Architectures and Algorithms for Future Wireless Local Area Future Wireless Local Area
NetworksNetworks
PhD Defense2012-12-14Peter Dely
Architectures and Algorithms for Future Wireless Local Area Networks 210987654321Chapter
High speeds require small distances
1
10
100
1000
10000
100000
1995 1998 2001 2004 2006 2009 2012 2014 2017
Year
PHY
Rate
(Mbi
t/s)
802.11n 802.11ac
802.11ac802.11ad
802.11g802.11a
802.11b802.11
60 GHz5 GHz2.4 GHz
2 Mbit/s
7000 Mbit/s
2
Carrier: 2.4 GHzModulation: BPSKChannel: 20 MHz
Carrier: 2.4 GHzModulation: BPSKChannel: 20 MHz
Carrier: 60 GHzModulation: QAM256Channel: 1800 MHz
Carrier: 60 GHzModulation: QAM256Channel: 1800 MHz
Architectures and Algorithms for Future Wireless Local Area Networks 310987654321Chapter
AP selection with small WLAN “cells”
Case 1: AP selection for mobility management
2
Architectures and Algorithms for Future Wireless Local Area Networks 410987654321Chapter
AP selection with small WLAN “cells”
Case 2: AP selection for load balancing
2
Architectures and Algorithms for Future Wireless Local Area Networks 510987654321Chapter
Research questions
1. Which AP should a station select at a given time?2. How to design and implement an architecture
that enables fair resource distribution and mobility management?
3. How to estimate the quality of an AP?
Architectures and Algorithms for Future Wireless Local Area Networks 610987654321Chapter
Greedy is not always best
Handover Handover
6 Mbit/s
54 Mbit/s
Given a certain “cost” for a handover, when is it beneficial
to perform that handover?
Given a certain “cost” for a handover, when is it beneficial
to perform that handover?
4
Architectures and Algorithms for Future Wireless Local Area Networks 710987654321Chapter
Dynamic model finds best AP for the known future
Mixed Integer Linear Program
Solver
1. Which station uses which AP at what point in time2. Which rate it can use for downloading
Compute
Describe system state
Describe system characteristics
Important constraints1. Capacity of the wireless channel2. Data transfer only after a station has been connected to an AP for a certain time
Important parameters1. Cost of a handover2. Link rates between APs and stations3. Interference
Maximize minimum average throughput
Objective
4
Architectures and Algorithms for Future Wireless Local Area Networks 810987654321Chapter
Evaluation scenario Comparison of
– Greedy – Hysteresis– k-Handover
Mobile stations “walk” on the corridors
Random waypoint with fixed waypoints
Comparison metric:– Normalized throughput– Relative to optimal solution with
dynamic model– 1 = Optimum
4
Architectures and Algorithms for Future Wireless Local Area Networks 910987654321Chapter
Mobility and handover costs reduce performance
Impact of mobility Impact of handover cost
More mobility =Less performance
Higher handover costs ~Less performance
4
Architectures and Algorithms for Future Wireless Local Area Networks 1010987654321Chapter
Internet
AP selection in wireless mesh networks
5
Architectures and Algorithms for Future Wireless Local Area Networks 1110987654321Chapter
Max-min fair rate allocations and single path routing cannot be found through a simple MILP
Solution algorithm
1. Which traffic is sent via which route2. Rates between the stations and the Internet3. Which station uses which AP
Compute
Describe system state
Describe system characteristics
Important constraints1. As in the pure AP scenario2. Flow conservation constraints3. Single-path routing constraints
Important parameters1. Network graph2. Interference set3. Association opportunities
Max-min fair rate allocation
ObjectiveMixed Integer Linear Program
Solver
Solve partial problem
ReformulateModel
5
Max-min fair optimization is multi-objective optimizationMax-min fair optimization is multi-objective optimization
Standard algorithms are very slow Design of
heuristic MESHMAX-FAST(*)
Standard algorithms are very slow Design of
heuristic MESHMAX-FAST(*)
Architectures and Algorithms for Future Wireless Local Area Networks 1210987654321Chapter
MESHMAX-FAST* is better than a straight forward linear relaxation
In 50% of cases, the performance is greater
than 95% of the optimum
In 50% of cases, the performance is greater
than 95% of the optimum
5
Better
Architectures and Algorithms for Future Wireless Local Area Networks 1310987654321Chapter
MESHMAX-FAST is suitable for online optimization
5
Optimal solution:> 1000 seconds
Optimal solution:> 1000 seconds
MESHMAX-FAST*:< 10 seconds with
comparable results
MESHMAX-FAST*:< 10 seconds with
comparable results
Architectures and Algorithms for Future Wireless Local Area Networks 1410987654321Chapter
Research questions
1. Which AP should a station select at a given time?2. How to design and implement an architecture
that enables fair resource distribution and mobility management?
3. How to estimate the quality of an AP?
6 7
Architectures and Algorithms for Future Wireless Local Area Networks 1510987654321Chapter
Internet
A software defined mesh network
OpenFlow Control Server
Monitoring and Control Server
Build network graph Run MESHMAX-FAST Trigger actions at the OpenFlow control server
Build network graph Run MESHMAX-FAST Trigger actions at the OpenFlow control server
Configure routes Configure rate shapers Trigger handovers
Configure routes Configure rate shapers Trigger handovers
Programmable forwarding unit (OpenFlow) Legacy routing protocol in a virtual network
Programmable forwarding unit (OpenFlow) Legacy routing protocol in a virtual network
6
Architectures and Algorithms for Future Wireless Local Area Networks 1610987654321Chapter
MESHMAX performance in real networksTestbed setup
IEEE 802.11a links, fixed PHY Download from “Internet” Gateways are connected via “DSL”
Results with 6 Mbit/s GW links
SNR = Use AP with best signal Hop-count = Use AP closest to GW MESHMAX = Use optimal AP
6
1
2
3
4
5
6
8
9
7
10
Architectures and Algorithms for Future Wireless Local Area Networks 1710987654321Chapter
Internet
CloudMAC distributes MAC processing
OpenFlow Controller
Policy
Application
Application
Virtual AP
Virtual WLAN NIC
Virtual AP
Virtual WLAN NIC
Virtual AP
Virtual WLAN NIC
Programswitch
7
Architectures and Algorithms for Future Wireless Local Area Networks 1810987654321Chapter
Internet
CloudMAC enables simple handovers
OpenFlow Controller
Policy
Application
Application
Virtual AP
Virtual WLAN NIC
Virtual AP
Virtual WLAN NIC
Virtual AP
Virtual WLAN NIC
ProgramswitchChange flow table
7
Architectures and Algorithms for Future Wireless Local Area Networks 1910987654321Chapter
CloudMAC reduces data loss during handovers
Median reduction in packet lossfrom approx. 10000 to 3.5
Standard IEEE 802.11: scan and re-association CloudMAC: Association state in Virtual AP no reassociation
7
Architectures and Algorithms for Future Wireless Local Area Networks 2010987654321Chapter
Research questions
1. Which AP should a station select at a given time?2. How to design and implement an architecture
that enables fair resource distribution and mobility management?
3. How to estimate the quality of an AP?
98
Architectures and Algorithms for Future Wireless Local Area Networks 2110987654321Chapter
Finding the AP with the best link is hard
8
Interference Cross traffic
Architectures and Algorithms for Future Wireless Local Area Networks 2210987654321Chapter
RSSI is not a good indicator for throughput
100% loss
0% loss
8
Architectures and Algorithms for Future Wireless Local Area Networks 2310987654321Chapter
BEST-AP uses regular traffic for performance estimation
BEST-AP Server
Internet
BW Estimation
Handover
BW Estimation
8
Architectures and Algorithms for Future Wireless Local Area Networks 2410987654321Chapter
Dynamic AP selection with stationary users
8
Testbed setup
Two APs Interference according to real traces
Interference Interference
Results
Architectures and Algorithms for Future Wireless Local Area Networks 2510987654321Chapter
Time
BufferLevel
Quality forcurrent AP starts
to decrease
Connection completely
breaks
Video playout freezes
Scan for new APs
Connect to new AP
Video playout resumes
Evaluation with mobile users
9
Video freeze duration
Architectures and Algorithms for Future Wireless Local Area Networks 2610987654321Chapter
BEST-AP reduces video freezesTestbed setup
9
Results
BEST-AP has fewer and shorter freezes than Linux
A dedicated scan card is not necessary
0%
5%
10%
15%
20%
25%
30%
% o
f tim
e in
free
ze m
ode
BEST-AP/BW-Estimation/No scan cardBEST-AP/RSSI/No scan card BEST-AP/RSSI/Scan card Linux/Optimized scanning Linux/Standard scanning
Architectures and Algorithms for Future Wireless Local Area Networks 2710987654321Chapter
Summary of contributions
1. Theoretical study of AP selection problem– Considering handover costs– Wireless mesh backhauls
2. Architecture proposals and their implementation– For the centralized management of wireless mesh
networks– For distributing MAC layer processing
3. Quality estimation of APs– Bandwidth estimation method– Improved video streaming
10
Architectures and Algorithms for Future Wireless Local Area Networks 2810987654321Chapter
Future directions Use other modeling frameworks
– Convex optimization – Robust optimization
Unify architectures– Use CloudMAC in mesh networks– Integrate bandwidth estimation in system
Comprehensive tests in large networks– More realistic traffic loads– Scalability tests
10
Architectures and Algorithms for Future Wireless Local Area Networks 2910987654321Chapter
Thank you!
Watch this talk on youtube.com/kaumesh
Architectures and Algorithms for Future Wireless Local Area Networks 3010987654321Chapter
Backup slides
Architectures and Algorithms for Future Wireless Local Area Networks 3110987654321Chapter
… but this comes at a cost From 2.4 GHz to 60 GHz
Higher path loss Higher attenuation by walls etc.
From BSPK to QAM256 More constellations More susceptible to noise
From 20 MHz to 160 MHz channels Less spatial reuse Higher energy use
2.4 GHz60 GHz
BSPK QAM256
36 40 44 48 52 56 60 64
To achieve high speeds short distances between the AP and
the stations are necessary.
To achieve high speeds short distances between the AP and
the stations are necessary.
2
Architectures and Algorithms for Future Wireless Local Area Networks 3210987654321Chapter
Main idea: use predictions of the future for optimization
Optimize each time slot using the estimate of the future
Using predictions to improve performance
More estimationerrors
4
Architectures and Algorithms for Future Wireless Local Area Networks 3310987654321Chapter
MESHMAX-FAST improves performance
5
Architectures and Algorithms for Future Wireless Local Area Networks 3410987654321Chapter
Key ideas of MESHMAX-FAST Observation: APs are traffic aggregation points for many
stations Algorithm sketch:
1. Compute a multi-path solution between the APs and the stations
2. Assign stations to APs according to the capacity of the APs3. Re-compute the multi-path solution and give priority to APs
with many stations4. Move stations to other APs if it increases the minimum rate5. Repeat steps 3-4 until the minimum rate cannot be increased
anymore6. Finally, find a single-path for each station
5