Distributed Spectrum Assignment for Home WLANs Julien Herzen (EPFL) - PowerPoint PPT Presentation

Distributed Spectrum Assignment for Home WLANs Julien Herzen (EPFL) Ruben Merz (Swisscom) Patrick Thiran (EPFL) April 17th, 2013 1 / 14

Context Interfering neighboring wi-fi home/office networks www.wigle.net • Several possible channels (center frequencies) • Variable bandwidth (5 → 20 → 40 → 160 MHz), limited spectrum • Non-heterogeneous density • No central control 2 / 14

Goal Joint allocation of channel center frequencies and bandwidths Conflicting goals: • Bandwidth ր Capacity ր ⇒ • Bandwidth ր ⇒ Interference likelihood ր 3 / 14

Goal Joint allocation of channel center frequencies and bandwidths Conflicting goals: • Bandwidth ր Capacity ր ⇒ • Bandwidth ր ⇒ Interference likelihood ր f f 1 f f 1 Capacity ր 3 / 14

Goal Joint allocation of channel center frequencies and bandwidths Conflicting goals: • Bandwidth ր Capacity ր ⇒ • Bandwidth ր ⇒ Interference likelihood ր f f 1 f f 1 Capacity ր f f 1 f 2 f f 1 f 2 Capacity � ? 3 / 14

Design Goals • Decentralized algorithm • Global convergence guarantees • Online for adaptivity to time-varying conditions • Transparent to user traffic • Practical for implementation on off-the-shelf 802.11 hardware Main contribution The first decentralized algorithm for joint center frequency and bandwidth adaptation with global convergence guarantees 4 / 14

Interference Model time power → l k f l f k frequency Interference produced by k on neighbor l : I l ( k ) = airtime( k ) · overlap( k , l ) 5 / 14

Interference Model time power → l k f l f k frequency Interference produced by k on neighbor l : I l ( k ) = airtime( k ) · overlap( k , l ) For two BSSs A and B : � � I A ( B ) = I l ( k ) l ∈ A k ∈ B 5 / 14

Optimization Objective Explicit interference vs. bandwidth trade-off: � � � minimize E := I A ( B ) + cost A ( b A ) B ∈N A A A � �� � � �� � Sum of bandwidth ”costs” Total interference • cost A ( b A ) is the cost that BSS A attributes to using bandwidth b A • E.g., cost A ( b A ) ∝ 1 / b A 6 / 14

Algorithm at BSS A Initialization: Pick a random configuration ( f A , b A ) After random (exp. distributed) time intervals: Pick a random configuration ( f new , b new ) Measure e 1 := � B ∈N A ( I A ( B ) + I B ( A )) + cost A ( b A ) if A uses ( f A , b A ) Measure e 2 := � B ∈N A ( I A ( B ) + I B ( A )) + cost A ( b new ) if A uses ( f new , b new ) Compute � 1 if e 2 < e 1 β T = exp e 1 − e 2 else T Set ( f A , b A ) = ( f new , b new ) with probability β T 7 / 14

Convergence Metropolis sampling for center frequency and bandwidth Theorem Denote X n the global state of the network after the n -th iteration. Consider a network where all the BSSs run our algorithm using a given parameter T . Then X n is a Markov chain, and it converges in distribution to π ( X ) ∝ e −E ( X ) / T , where X is a global state. • State gets arbitrarily close to optimal for T small enough • T encodes a trade-off between likelihood of local optima and asymptotic efficiency 8 / 14

Implementation • 802.11g with 5, 10 and 20 MHz channel widths • Interference measured by spending ≤ 50 ms. out-of-band • Optional client collaboration for interference measurement • C++ implementation using Click in userspace • cost A ( b A ) = 1 / b A 65 m 40 m 9 / 14

Performance Evaluation UDP traffic, client-agnostic: UDP traffic, client-aware: total throughput [Mbps] total throughput [Mbps] 80 80 70 70 60 60 Bench Bench 50 50 40 40 30 30 20 20 0 1000 2000 3000 4000 0 1000 2000 3000 4000 time [s] time [s] TCP traffic, client-agnostic: TCP traffic, client-aware: 65 65 total throughput [Mbps] total throughput [Mbps] 60 60 55 55 50 50 Bench Bench 45 45 40 40 35 35 30 30 25 25 20 20 0 1000 2000 3000 4000 0 1000 2000 3000 4000 time [s] time [s] ”Bench” line: centralized graph-coloring for fixed-width channels 10 / 14

Simulation • Random distribution of BSSs on the plane • Capacity of link l = b l · log(1 + SINR ) • cost A ( b A ) = c / b A , optimization objective becomes: � � � minimize I A ( B ) + c · 1 / b A B ∈N A A A • c = 0: minimize interference • c → ∞ : use largest bandwidth, irrespective of interference 11 / 14

Simulation • Random distribution of BSSs on the plane • Capacity of link l = b l · log(1 + SINR ) • cost A ( b A ) = c / b A , optimization objective becomes: � � � minimize I A ( B ) + c · 1 / b A B ∈N A A A • c = 0: minimize interference • c → ∞ : use largest bandwidth, irrespective of interference 1 . 0 capacity normalized value 0 . 8 fairness 0 . 6 0 . 4 interference 0 . 2 0 . 0 0 2 4 6 8 10 c 11 / 14

Simulation • Random distribution of BSSs on the plane • Capacity of link l = b l · log(1 + SINR ) • cost A ( b A ) = c / b A , optimization objective becomes: � � � minimize I A ( B ) + c · 1 / b A B ∈N A A A • c = 0: minimize interference • c → ∞ : use largest bandwidth, irrespective of interference 1 . 0 8000 capacity 7000 c = 1 normalized value 0 . 8 total capacity 6000 fairness 0 . 6 5000 c = 0 4000 c = 100 0 . 4 3000 interference 0 . 2 2000 1000 0 . 0 0 50 100 150 200 250 300 350 400 0 2 4 6 8 10 c average spatial density [ BSS/km 2 ] 11 / 14

Simulation Improvement with respect to random allocations after 5 iterations: after 20 iterations: percentage improvement percentage improvement 140 300 120 250 % interference 100 % capacity increase 200 decrease 80 150 60 100 40 % capacity increase % interference 50 20 decrease 0 0 0 20 40 60 80 100 0 20 40 60 80 100 percentage of BSSs running SAW percentage of BSSs running SAW 12 / 14

Simulation total spectrum: 45 MHz total spectrum: 70 MHz Capacity, c+w Capacity, c+w 1 . 0 1 . 0 normalized value normalized value Capacity, c Interference, c 0 . 8 0 . 8 0 . 6 0 . 6 Capacity, c Interference, c 0 . 4 0 . 4 0 . 2 0 . 2 Interference, c+w Interference, c+w 0 . 0 0 . 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 average iterations per AP average iterations per AP Fairness, c+w 0 . 8 0 . 8 Fairness, c+w fairness index fairness index 0 . 7 0 . 7 0 . 6 0 . 6 0 . 5 0 . 5 Fairness, c Fairness, c 0 . 4 0 . 4 0 . 3 0 . 3 0 10 20 30 40 50 60 0 10 20 30 40 50 60 average iterations per AP average iterations per AP 13 / 14

Conclusion • Distributed, joint allocation of center frequencies and bandwidths • Bandwidth influences both capacity and interference; ideal spectrum consumption should depend on network density • Optimization of an explicit trade-off between interference mitigation and use of advantageous bandwidths • Simple optimization objectives yield best results irrespective of network density • Large capacity improvements, even when not all BSSs run the algorithm • Testbed implementation shows feasibility and improvements compared to fixed-width graph coloring 14 / 14

1 / 5

Some Related Work • Channel allocation / graph coloring, e.g., [Akella et al. 2005, Kauffmann et al. 2007, Duffy et al. 2011, Leith et al. 2012] ◮ Main goal: minimize interference (no variable bandwidth) • Variable bandwidth / white spaces, e.g., [Chandra et al. 2008, Bahl et al. 2009, Rayanchu et al. 2011] ◮ Heuristics, no focus on self-organization 2 / 5

Micro-sensing micro-sensing t m-s decision t switch t sensing t switch time time timer fires optionally: receive compute list of bands to scan init. switch init. switch back t m-s link stats from clients unblock traffic optionally: inform clients block traffic 16 Link 1 starts sensing throughput [Mbps] 14 12 10 8 6 TCP link 1 4 TCP link 2 2 Link 1 switches band 0 0 20 40 60 80 100 120 time [s] 3 / 5

Channel widths 5 MHz 10 MHz 20 MHz −30 −30 −30 −40 −40 −40 dBm −50 −50 −50 −60 −60 −60 −70 −70 −70 −20 0 20 −20 0 20 −20 0 20 MHz MHz MHz 4 / 5

Performance Evaluation (uplink) UDP traffic, client-agnostic: UDP traffic, client-aware: total throughput [Mbps] total throughput [Mbps] 80 80 70 70 Bench Bench 60 60 50 50 40 40 30 30 20 20 0 1000 2000 3000 4000 0 1000 2000 3000 4000 time [s] time [s] TCP traffic, client-agnostic: TCP traffic, client-aware: 65 65 total throughput [Mbps] total throughput [Mbps] 60 60 55 55 Bench Bench 50 50 45 45 40 40 35 35 30 30 25 25 20 20 0 1000 2000 3000 4000 0 1000 2000 3000 4000 time [s] time [s] ”Bench” line: centralized graph-coloring for fixed-width channels 5 / 5