Network Coding-Aware Queue Network Coding Aware Queue Management for Unicast Flows over Coded Wireless Networks Coded Wireless Networks Hülya Sefero ğ lu, Athina Markopoulou University of California Irvine University of California, Irvine
Wireless Mesh Networks Wireless Mesh Networks o Y. Wu, P. A. Chou, S. Y. Kung, “Information exchange in wireless network coding and physical layer broadcast” CISS ’05 coding and physical layer broadcast , CISS 05. o S. Katti, H. Rahul, W. Hu, D. Katabi, M. Medard, J. Crowcroft "XORs In The Air: Practical Wireless Network Coding, (COPE)“, ToN ’08. o o Throughput increases by mixing packets Throughput increases by mixing packets Gateway Intermediate Node Intermediate Node Internet
One-hop network coding One hop network coding Output Output queue Output queue queue a 1 b 1 a 3 a 2 b 1 b 2 b 3 A B a 1 b 1 a 1 b 1 I D C a 1 + b 1 a 1 b 1 Virtual Virtual queue queue
Motivation Motivation o Problem: o TCP over COPE does not fully exploit the NC potential o Intuition: o N t Not enough coding opportunities due to TCP burstiness h di t iti d t TCP b ti o Coded flows do not compete for resources TCP TCP x 1 x 1 x 2 x 2 A B a 1 b 1 a 1 b 1 b 1 I I D C C max{x 1 , x 2 } Packets in the ? buffer ☺
Motivation Motivation o Problem: o TCP over COPE does not fully exploit the NC potential o Intuition: o N t Not enough coding opportunities due to TCP burstiness h di t iti d t TCP b ti o Coded flows do not compete for resources o A Possible Solution: o A Possible Solution: o Artificially delay packets at intermediate nodes Y. Huang, M. Ghaderi, D. Towsley, and W. Gong, “TCP performance in coded wireless mesh networks,” SECON ‘08. o Throughput increases with small delay, but decreases with large delay o Optimal delay depends on the network topology and the background traffic, and may change over time b k d t ffi d h ti o Not practical
Motivation Motivation o Problem: o TCP over COPE does not fully exploit the NC potential o Intuition: o Not enough coding opportunities due to TCP burstiness N t h di t iti d t TCP b ti o Coded flows do not compete for resources o Proposed Solution: o Network Coding-Aware Queue Management (NCAQM) o o No changes to TCP and MAC No changes to TCP and MAC o Formulate network utility maximization (NUM) problem o TCP+NCAQM doubles the network coding benefit of TCP COPE TCP+COPE
Previous Work I Intra-session Network Coding i N k C di o Minimum cost multicast for wired and wireless: o D. S. Lun, N. Ratnakar, M. Medard, R. Koetter, D. R. Karger, T. Ho, E. Ahmed, and F. Zhao, “Minimum-cost multicast over coded packet networks,” ToIT'06. o L. Chen, T. Ho, S. Low, M. Chiang, and J. C. Doyle, “Optimization based rate control for multicast with network coding,” Infocom'07. o Minimum cost unicast with for wireless: o B. Radunovic, C. Gkantsidis, P. Key, P. Rodriguez, and W. Hu, “Toward Practical Opportunistic Routing with Intra-session N Network Coding for Mesh Networks,” ToN’09. k C di f M h N k ” T N’09
Previous Work Inter-session Network Coding I i N k C di o Optimal Scheduling and Routing: p g g o P. Chaporkar and A. Proutiere, “Adaptive network coding and scheduling for maximizing througput in wireless networks,” Mobicom'07. o S. Sengupta, S. Rayanchu, and S. Banarjee, “An Analysis of Wireless Network Coding for Unicast Sessions: The Case for Coding-Aware Routing,” Infocom'07. o Energy efficient network coding: E ffi i t t k di o T. Cui, L. Chen, and T. Ho, “Energy Efficient Opportunistic Network Coding for Wireless Networks,” Infocom'08. o End2end pairwise network coding: o A. Khreishah, C. C. Wang, and N. B. Shroff, “Cross-layer optimization for wireless multihop networks with pairwise i intersession network coding,” JSAC'09. i k di ” J C'09
Motivation Motivation o Proposed Solution: o Proposed Solution: Network Coding-Aware Queue Management (NCAQM) o o No changes to TCP and MAC o F Formulate network utility maximization (NUM) problem l k ili i i i (NUM) bl o TCP+NCAQM doubles the network coding benefit of TCP+COPE o Our work in perspective: o Multiple unicast flows over wireless with given network coding scheme and pre-determined routing paths scheme and pre determined routing paths o Connection between optimization and protocol design o Intuition for practical implementation
Outline Outline o Introduction o Network Utility Maximization (NUM) o Network Coding-Aware Queue Management (NCAQM) o Performance Evaluation o Extensions & Summary
Network utility maximization y Formulation Optimize ∑ total utility max ( ) U x s s α τ U(x 1 ) U(x 2 ) , , x ∈ s S x 1 x 1 x 2 x 2 ∑ ∑ A B α ≤ ≤ τ ∀ ∀ ∈ , , s s k k s s k k . . max { { } } , s t t H H x R R h h A A a 1 b 1 h h s h h ∈ s S ∈ k k K h ∑ ∑ α = ∀ ∈ ∈ , s k a 1 1 , , s S i P b 1 b 1 h s I I ∈ ∈ ∈ ∈ ∈ ∈ h h ( ( J J )| )| h h A A k k K K | | s s S S h h k k ∑ τ ≤ γ ∀ ⊆ , C q A h Capacity D ∈ h C C max{x 1 , x 2 } q constraint Flow Conservation Interference Interference
Network utility maximization y Solution I: , α ∑ ∑ , , s s k k s s k k s s k k max H H x m h h s h m α ∈ , , s k s k ≡ s S max { } H x k h h s ∑ ∈ = s S , s k . . 1 k s t m h ∈ k s S ∑ ∑ max max ( ( ) ) U U x x s s α τ , , x ∈ s S ∑∑ α ≤ τ ∀ ∈ s , k s , k s , k * . . ( ) , s t H x m R h A h h s h h h ∈ ∈ k K s S h k ∑ ∑ α = ∀ ∈ ∈ , s k 1 , , s S i P h s ∈ ∈ ∈ ( )| | h J h A k K s S h k ∑ ∑ τ τ ≤ ≤ γ γ ∀ ∀ ⊆ ⊆ , , C C q A A h h ∈ h C q
Network utility maximization y Solution II: Rate Rate ⎛ ⎛ ⎞ ⎞ ∑ ∑ ⎜ ⎜ ⎟ ⎟ − = α 1 , , , * s k s k s k ( ' ) ( ) x U q H m Control ⎜ ⎟ s s h h h h ⎝ ⎠ ∈ ∈ ∈ h A k K | s S h k + Queue Qu u ⎧ ⎧ ⎫ ⎫ ⎡ ⎡ ⎤ ⎤ ⎪ ⎪ ⎪ ⎪ ∑∑ + = + α − τ , , , * ⎨ s k s k s k ⎬ ⎢ ⎥ ( 1 ) ( ) ( ) q t q t c H m x R Size h h t h h h s h h ⎪ ⎪ ⎣ ⎦ ⎩ ⎭ ∈ ∈ k K s S k k ∑ ∑ ∑ ∑ α s , k k s , k k * * s , k k min ( ) q H m Traffic h h h h α ∈ ∈ ∈ h ( J )| h A k K | s S h k Splitting ∑ ∑ α = ∀ ∈ , s k . . 1 , s t i P h s ∈ ∈ ∈ ∈ ∈ ∈ ( ( )| )| | | h h J J h h A A k k K K s s S S h h k k ∑ τ max q R h h h τ Scheduling Scheduling ∈ h A ∑ τ ≤ γ ∀ ⊆ , C q A h ∈ h C q
Network Coding-Aware Queue Management (NCAQM) P Protocol modifications, mimicking the optimal solution t l difi ti i i ki th ti l l ti Implementation Summary Implementation Summary Queue management • Network coding (NCAQM) ( Q ) • Packet dropping pp g TCP No change (TCP-SACK) MAC MAC No change (802 11) No change (802.11) Minimal and intuitive
NCAQM Q Maintaining queues and packet transmission + + Queue ⎧ ⎧ ⎫ ⎫ ⎡ ⎤ { } ⎪ ⎪ ∑ + = + α , − τ s , k s k ⎨ ⎢ ⎥ ⎬ ( 1 ) ( ) max q t q t c H x R Size h ⎪ h t h h s h h ⎪ ∈ ⎣ ⎦ ⎩ s S ⎭ ∈ k k K k Modification I o Q i is the output queue at node i A B o o Store network coded packets (when an Store network coded packets (when an Q Q i i opportunity arises) instead of uncoded packets I o o Keep track of hypearc queues Keep track of hypearc queues D o Estimate traffic splitting parameters C o Packet scheduling is according to FIFO FIFO queue
NCAQM Q Rate control and packet dropping − 1 ⎛ ⎛ ⎞ ⎞ Optimal Rate l S Sum of network coded queue sizes f k d d i = ∑ ⎜ ⎟ s x q ⎜ ⎟ across all nodes on the path Control ( ) s h i ⎝ ⎠ ∈ h ( i ) P s Modification II Modification II o Upon congestion, compare Q i s for all x A B flows s. Drop an uncoded packet Q i from the largest flow from the largest flow s ? o How to calculate Q i I o Determine hyperarc queues that flow s is dominating (has the fl s is d min tin (h s th D C largest number of packets) o Sum the number of packets of flow s over these hyperarc flow s over these hyperarc queues
NCAQM Q Implementation Summary o Problem: P bl m: o TCP over COPE does not fully exploit the NC potential o Intuition: o Flows coded together do not compete for resources o Not enough coding opportunities due to TCP burstiness o Modifications o Store network coded packets (when an opportunity arises) instead of uncoded packets instead of uncoded packets. o Compare Q i s for all s. Drop an uncoded packet from the “largest” flow.
Outline Outline o Introduction o Network Utility Maximization (NUM) o Network Coding-Aware Queue Management (NCAQM) o Performance Evaluation o Extensions & Summary
Performance evaluation Scenarios [Glomosim + NC] X Topology A & B Topology p gy Grid Topology l Cross Topology
Recommend
More recommend
