Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
A. Mishra, V. Shrivastava, D.Agrawal, S. Banerjee, S. Ganguly Distributed Channel Management in Uncoordinated Wireless Environments ACM MobiCom, 2006
Advisor: Dr. Jenq-Shiou Leu Student: Cheng-Wei Tsai Date: May. 28, 2008
Reference
n
A. Mishra, V. Brik,, S. Banerjee, A. Srinivasan, W. Arbaugh A client-driven Approach for Channel Management in Wireless LANs IEEE Infocom, 2006 A. Akella, G. Judd, S. Seshan, and P. Steenkiste, Selfmanagement in chaotic wireless deployments, in ACM Mobicom, 2005. A. Mishra, V. Brik, S. Banerjee, A. Srinivasan, and W. Arbaugh, Efficient strategies for channel management in wireless lans, UMD, CS Tech. Rep. CS-TR-4729, 2005.
Outline
n n n n n
Introduction
Least
Proposed Scheme
Distributed
Related Work
Client-driven
Simulation Conclusion
Introduction
Chaotic
Proposed Scheme
Distributed
Algorithm Algorithm
Related Work
Client-driven
Simulation Conclusion
Chaotic Deployment(1/3)
n
nControlled
by a single organization nCarefully planned to optimize coverage and minimize cell overlap nOptimize global metric nLoad balancing of users among APs in transmission range
by multiple organizations nHighly variable densities nPriorities conflict nConfiguring and managing difficult
Chaotic Deployment(2/3)
Channel
% AP
1 6 2 11
Most users dont change default channel Channel selection must be automated
Chaotic Deployment(3/3)
n
Chaotic deployment
Unplanned
n
Unmanaged
n
Motivation
Vendor % AP
Linksys (Cisco) Aironet (Cisco) Agere D-Link Apple Netgear ANI Communications Delta Networks Lucent Acer Others
33.5 12.2 9.6 4.9 4.6 4.4 4.3 3 2.5 2.3 16.7
n n
Detailed RF site surveys by using spectrum analyzers and manually assign specific channels to each AP Each AP periodically checks other data transmissions in the channel it is using. If the volume of traffic in that channel is greater than a threshold, then the AP tries to move over to a less congested channel.
Assign a fixed channel to an AP for a relatively long duration of time (such as hours or days). Using the same channel for a long duration of time shows that the corresponding APs suffer unfairly when compared to other APs
AP1 AP4 AP1 AP4 AP1 AP4
AP2
AP3
AP2
AP3
AP2
AP3
Unfair!!
PDF created with pdfFactory Pro trial version www.pdffactory.com
n n n n n
Introduction
Least
Proposed Scheme
Distributed
Related Work
Client-driven
Simulation Conclusion
APs use a sequence of channels and transition between them over time. AP1 AP2 AP1 AP2
AP1 = { 1, AP2 = { 2, AP3 = { 3, AP4 = { 3, 2, 1, 1, 3, 2, 2, 3, 1, 3, 2, 1, 3, 2, 1, 3, 1, 2} 1} 2} 3} AP3 AP1 AP4 AP2 AP3 AP1 AP4 AP2
AP3 AP1
AP4 AP2
AP3 AP1
AP4 AP2
AP3
PDF created with pdfFactory Pro trial version www.pdffactory.com
AP4
AP3
AP4
Two APs interfere with each other if either the APs interfere directly or clients associated to these APs interfere with each other.
PDF created with pdfFactory Pro trial version www.pdffactory.com
Distributed Algorithm(1/7)
Channel = 1 Channel = 2 AP7 AP1 = { 2, AP2 = { 1, AP3 = { 2, AP4 = { 1, AP5 = { 3, AP6 = { 3, 1 1 1 1 1 1
PDF created with pdfFactory Pro trial version www.pdffactory.com
Channel = 3 1 1 1 1 2 2
1, 3, 3, 1, 2, 2,
2, 1, 2, 3, 3, 1,
3, 3, 1, 1, 2, 2,
2, 3, 3, 2, 1, 1,
1} 2} 1} 2} 3} 3}
1 2 1 2 1 1
2 1 2 1 1 1
Distributed Algorithm(2/7)
Channel = 1 Channel = 2 AP7 2 1, 3, 3, 1, 2, 2, 3 2 3 2 2 2 2, 1, 2, 3, 3, 1, 3, 3, 1, 1, 2, 2, 2, 3, 3, 2, 1, 1, 1} 2} 1} 2} 3} 3} 4 2 3 3 2 2 3 2 3 2 3 3 Channel = 3 3 3 4 2 2 2
Distributed Algorithm(3/7)
Channel = 1 Channel = 2 AP7 2 2 1, 3, 3, 1, 2, 2, 3 2 3 2 3 3 2, 1, 2, 3, 3, 1, 4 3 4 3 4 4 3, 3, 1, 1, 2, 2, 2, 3, 3, 2, 1, 1, 1} 2} 1} 2} 3} 3} 4 4 4 3 4 5 5 3 5 3 4 4 Channel = 3 4 3 4 4 5 4
Distributed Algorithm(4/7)
Channel = 1 Channel = 2 AP7 2 2 1, 3, 3, 1, 2, 2, 3 2 3 2 3 3 3 2, 1, 2, 3, 3, 1, 4 3 4 4 5 4 3, 3, 1, 1, 2, 2, 5 4 5 5 6 5 2, 3, 3, 2, 1, 1, 1} 2} 1} 2} 3} 3} 5 4 6 6 6 5 5 4 5 5 7 6 Channel = 3 6 5 5 5 6 5
Distributed Algorithm(5/7)
Channel = 1 Channel = 2 AP7 2 2 1, 3, 3, 1, 2, 2, 3 2 3 2 3 3 3 2, 1, 2, 3, 3, 1, 4 3 4 4 5 4 3 3, 3, 1, 1, 2, 2, 6 5 5 5 6 5 2, 3, 3, 2, 1, 1, 7 6 6 6 7 6 1} 2} 1} 2} 3} 3} 7 6 6 6 8 7 8 6 6 7 7 6 Channel = 3 7 7 7 6 7 6
Distributed Algorithm(6/7)
Channel = 1 Channel = 2 AP7 2 2 1, 3, 3, 1, 2, 2, 3 2 3 2 3 3 3 2, 1, 2, 3, 3, 1, 4 3 4 4 5 4 3 3, 3, 1, 1, 2, 2, 6 5 5 5 6 5 3 2, 3, 3, 2, 1, 1, 7 7 7 6 7 6 1} 2} 1} 2} 3} 3} 8 8 8 7 8 7 9 8 9 7 8 7 8 9 8 8 8 7 Channel = 3 8 8 8 7 9 8
Distributed Algorithm(7/7)
Channel = 1 Channel = 2 AP7 2 2 1, 3, 3, 1, 2, 2, 3 2 3 2 3 3 3 2, 1, 2, 3, 3, 1, 4 3 4 4 5 4 3 3, 3, 1, 1, 2, 2, 6 5 5 5 6 5 3 2, 3, 3, 2, 1, 1, 7 7 7 6 7 6 2 1} 2} 1} 2} 3} 3} 8 8 8 7 8 7 9 8 9 7 8 7 8 9 8 8 8 7 Channel = 3 8 8 8 7 9 8
n n n n n
Introduction
Least
Proposed Scheme
Distributed
Related Work
Client-driven
Simulation Conclusion
Motivation(1/2)
AP2
AP2
C1 AP1 C2
AP4
AP4
Motivation(1/2)
C2 AP2
AP2
C1
AP1
AP3
C3
AP1
AP3
AP4
AP4
C4
Client-driven Approach
n
Each client periodically monitors its range set and interference set and sends this information to the central entity. The central entity uses this information to compute an appropriate channel assignment and communicates this assignment to each AP.
Range set: the set of all APs such that a client lies within the transmission range of each such AP Interference set: the set of all APs not in the range set of a client but interfere the such client
C2 AP1 C3
C1 AP2
There is exactly one AP in its channel from its range set (this is the AP the client will be associated to), and there is no other AP in its range or interference set using the same channel.
Range Set C1 C2 C3 {AP3} {AP2, AP3} {AP1} Interference Set {AP1, AP2} {AP1} {AP3}
Assume: 2 channels available Assign channel 1 to AP1 Assign channel 1 to AP2 Assign channel 2 to AP3
PDF created with pdfFactory Pro trial version www.pdffactory.com
Inter-AP Interference
n
Inter-AP interference occurred at a client when the same channel was used by more than one AP in the range or interference set.
Assume: 2 channels available Assign channel 1 to AP1 Assign channel 1 to AP2 Assign channel 2 to AP3
PDF created with pdfFactory Pro trial version www.pdffactory.com
If many clients are already associated to an AP, such clients would suffer intra-AP interference.
Assume: 3 channels available Assign channel 1 to AP1 Assign channel 2 to AP2 Assign channel 3 to AP3
PDF created with pdfFactory Pro trial version www.pdffactory.com
Algorithm Complexity
n
Invoke the algorithm multiple times with different random permutations and obtained the best solution across these runs. Increases the chances of converging to a better optima and possibly the global optimum over multiple executions.
KN N*k
N : number of APs k : number of channels
Client-driven Algorithm(1/6)
AP2 1. Generate a random permutation of APs 2. Improve the interference iteratively by different channel assignments 3. If the refinement stays the same, terminate!! AP2 AP = {AP1, AP2, AP3}, 2 channels available A permutation of APs = {AP2, AP3, AP1} AP3 AP1 AP3
AP1
Client-driven Algorithm(2/6)
A permutation of APs = {AP1, AP2, AP3, AP4} Assuming 2 channels available
Range Set C1 C2 C3 C4 {AP1, AP3} {AP2, AP4} {AP1, AP2, AP4} {AP3, AP2} Interference Set {AP2, AP4} {AP1} {AP3} {AP1, AP4}
{ 1, 2, 1, 2 } { 1, 2, 1 } { 1, 1, 2, 2 } { 2, 1, 1, 2 }
1+0=1 0+0=0 1+0=1 1+0=1
C1 AP1, inter-AP = 1, intra-AP = 0 C2 AP4, inter-AP = 0, intra-AP = 0 C3 AP2, inter-AP = 1, intra-AP = 0 C4 AP3, inter-AP = 1, intra-AP = 0
Client-driven Algorithm(3/6)
Assuming 2 channels available
Range Set C1 C2 C3 C4 {AP1, AP3} {AP2, AP4} {AP1, AP2, AP4} {AP3, AP2} Interference Set {AP2, AP4} {AP1} {AP3} {AP1, AP4}
{ 2, 2, 1, 2 } { 1, 2, 2 } { 2, 1, 2, 2 } { 2, 1, 2, 2 }
2+0=0 0+2=2 0+2=2 0+2=2
C1 AP1, inter-AP = 2, intra-AP = 0 C2 AP2, inter-AP = 0, intra-AP = 2 C3 AP2, inter-AP = 0, intra-AP = 2 C4 AP2, inter-AP = 0, intra-AP = 2
Client-driven Algorithm(4/6)
Assuming 2 channels available
Range Set C1 C2 C3 C4 {AP1, AP3} {AP2, AP4} {AP1, AP2, AP4} {AP3, AP2} Interference Set {AP2, AP4} {AP1} {AP3} {AP1, AP4}
{ 1, 2, 2, 2 } { 2, 2, 1 } { 1, 2, 2, 2 } { 2, 2, 1, 2 }
0+1=1 1+0=1 0+1=1 2+0=2
C1 AP1, inter-AP = 0, intra-AP = 1 C2 AP2, inter-AP = 1, intra-AP = 0 C3 AP1, inter-AP = 0, intra-AP = 1 C4 AP3, inter-AP = 2, intra-AP = 0
Client-driven Algorithm(5/6)
Assuming 2 channels available
Range Set C1 C2 C3 C4 {AP1, AP3} {AP2, AP4} {AP1, AP2, AP4} {AP3, AP2} Interference Set {AP2, AP4} {AP1} {AP3} {AP1, AP4}
{ 1, 1, 1, 2 } { 1, 2, 1 } { 1, 1, 2, 1 } { 1, 1, 1, 2 }
2+0=2 0+1=1 0+1=1 2+0=2
C1 AP1, inter-AP = 2, intra-AP = 0 C2 AP4, inter-AP = 0, intra-AP = 1 C3 AP4, inter-AP = 0, intra-AP = 1 C4 AP2, inter-AP = 2, intra-AP = 0
Client-driven Algorithm(6/6)
Assuming 2 channels available
Range Set C1 C2 C3 C4 {AP1, AP3} {AP2, AP4} {AP1, AP2, AP4} {AP3, AP2} Interference Set {AP2, AP4} {AP1} {AP3} {AP1, AP4}
{ 1, 2, 1, 1 } { 1, 1, 1 } { 1, 1, 1, 2 } { 2, 1, 1, 1 }
0+1=1 2+0=2 2+0=2 0+1=1
C1 AP3, inter-AP = 0, intra-AP = 1 C2 AP2, inter-AP = 2, intra-AP = 0 C3 AP1, inter-AP = 2, intra-AP = 0 C4 AP3, inter-AP = 0, intra-AP = 1
n n n n n
Introduction
Least
Proposed Scheme
Distributed
Related Work
Client-driven
Simulation Conclusion
Performance Metrics
n
The impact of channel assignment algorithms on the overall network capacity. Measure the fairness in throughput by calculating its Jains n fairness index 2 i i =1 1 2 n n 2 i i =1
f ( x , x ,..., x ) =
( x ) n x
unfair 0
fair 1
Simulation Settings
San Francisco City
Simulation Result(1/3)
Simulation Result(2/3)
Simulation Result(3/3)
n n n n n
Introduction
Least
Proposed Scheme
Distributed
Related Work
Client-driven
Simulation Conclusion
Conclusion
n n n n
How to choose the objective function? Centralized or distributed? Design your algorithm Prove your proposal will get a convergent solution.
Appendix(1/2)
Distributed Algorithm
Appendix(2/2)
Client-driven Algorithm
Interfering APs
AP1
C1
AP2
C2
C1
AP1
AP2
C2
AP1
C1
C2
AP2