CENTAUR: Realizing the Full Potential of Centralized WLANs Through a Hybrid Data Path Vivek Shrivastava*, Shravan Rayanchu, Suman Banerjee University of Wisconsin-Madison 1 Nabeel Ahmed, Srinivasan Keshav University of Waterloo, Ontario Konstantina Papagiannaki Intel Labs, Pittsburgh Arunesh Mishra Google Inc. *

2 Centralization of Enterprise WLANs 2 Vivek Shrivastava Wireless controller Access Point Clients Internet

3 Centralization of Enterprise WLANs 3 Vivek Shrivastava Power control Channel assignment Common control plane functions

4 Centralization of Enterprise WLANs 4 Vivek Shrivastava What about data plane functions? Power control Channel assignment Data scheduling ?

55 Vivek Shrivastava Can centralized scheduling help?

66 Vivek Shrivastava Can centralized scheduling help? Hidden terminals

77 Vivek Shrivastava Hidden terminals Can centralized scheduling help?

88 Vivek Shrivastava Hidden terminals Can centralized scheduling help?

99 Vivek Shrivastava 1.Carrier sense Hidden terminals Can centralized scheduling help?

10 Vivek Shrivastava 2. Channel free, transmit Hidden terminals Can centralized scheduling help?

11 Vivek Shrivastava Collision! Backoffs Low throughputs 3. Collision ! Can centralized scheduling help?

12 How bad is it ? 12 Vivek Shrivastava Experiments on production building-wide WLANs W1: 5 floors 9 APs, 45 clients W2: 1 floor 21 APs, 51 clients 10% links suffer more than 70% throughput reduction

13 Vivek Shrivastava A lost opportunity ? Suppose infrastructure can gather conflict data

14 Vivek Shrivastava A lost opportunity ? And when packets arrive …

15 Vivek Shrivastava A lost opportunity ? … realize interference will happen …

16 Vivek Shrivastava A lost opportunity ? 1.Transmit first packet

17 Vivek Shrivastava A lost opportunity ? 1.Transmit first packet 2.Transmit second packet with delay

18 Vivek Shrivastava Use an in-band scheduler Simple FIFO schedule with interference avoidance Scheduling functionality

19 What about exposed terminals ? 19 Vivek Shrivastava In our experiments, about 41%of link pairs suffer from exposed terminal interference Disabling carrier sense to solve the problem can be dangerous for uplink, non-enterprise traffic We will show how centralization can help even here

20 Challenge Centralization has obvious overheads How to make this feasible and useful under Real applications Common large-scale wireless environments Presence of uplink and non-enterprise traffic Requirements of no client modifications 20 Vivek Shrivastava

21 Contributions 1. Design CENTAUR, a hybrid (partly centralized, partly distributed) scheduling approach Resolves hidden and exposed terminals Requires no client modifications Requires no carrier sense disabling 2. Evaluate CENTAUR on two WLAN testbeds with real-world traffic traces 1.48x greater throughput for bulk data traffic 1.38x reduction in web-transaction times 21 Vivek Shrivastava

22 Outline A naïve attempt at centralized scheduling (DET) Our hybrid centralized scheduler (CENTAUR) Evaluation Related Work Summary 22 Vivek Shrivastava

23 Outline A naïve attempt at centralized scheduling (DET) Our hybrid centralized scheduler (CENTAUR) Evaluation Related Work Summary 23 Vivek Shrivastava

24 Vivek Shrivastava DET: A Simple Deterministic Scheduler

25 Vivek Shrivastava DET: A Simple Deterministic Scheduler Input: Conflict graph

26 Vivek Shrivastava DET: A Simple Deterministic Scheduler Goal: Schedule each incoming downlink packet Input: Conflict graph

27 Vivek Shrivastava DET: A Simple Deterministic Scheduler 1234 Transmission slots Goal: Schedule each incoming downlink packet Input: Conflict graph

28 Vivek Shrivastava DET: A Simple Deterministic Scheduler 1234 Transmission slots 1.New packet arrives

29 Vivek Shrivastava DET: A Simple Deterministic Scheduler 1234 Transmission slots 1.New packet arrives 2.Find the earliest conflict free slot

30 Vivek Shrivastava DET: A Simple Deterministic Scheduler 1234 Transmission slots 1.New packet arrives 2.Find the earliest conflict free slot

31 Vivek Shrivastava DET: A Simple Deterministic Scheduler 1234 Transmission slots 1.New packet arrives 2.Find the earliest conflict free slot 3.Schedule the packet in that slot 5

Performance of DET 4x No gains for exposed terminals; Non- conflicting links perform worse under load 32 Vivek Shrivastava

33 Outline A naïve attempt at centralized scheduling (DET) Our hybrid centralized scheduler (CENTAUR) Evaluation Related Work Summary 33 Vivek Shrivastava

34 Overview of CENTAUR Incorporate basic DET scheduler Tackle DET’s shortcomings: Amortize scheduling overhead Improve performance for exposed links Avoid degrading normal links Coexist with non-enterprise and uplink traffic 34 Vivek Shrivastava

35 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 35 Vivek Shrivastava

36 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 36 Vivek Shrivastava

37 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 37 Vivek Shrivastava Wired ack

38 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 38 Vivek Shrivastava

39 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 39 Vivek Shrivastava

40 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 40 Vivek Shrivastava

41 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 41 Vivek Shrivastava

42 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 42 Vivek Shrivastava Wired ack

43 (1) Avoid Scheduling Overheads Problem: Per-packet scheduling performs poorly under high network loads Solution: Schedule packets in batches (or epochs) 43 Vivek Shrivastava

44 (2) Improve Exposed Terminals 44 Vivek Shrivastava Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

45 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

46 Vivek Shrivastava Variable wired delay (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

47 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

48 Vivek Shrivastava Carrier sense, deferral (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

49 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

50 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

51 Vivek Shrivastava Schedule packets in batches (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

52 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

53 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

54 Vivek Shrivastava First packets can be out of sync (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

55 Vivek Shrivastava Waiting packets synchronized by carrier sense ! (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

56 Vivek Shrivastava (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

57 Vivek Shrivastava After first packet, both APs transmit simultaneously. (2) Improve Exposed Terminals Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets, fix backoff periods

58 (3) Avoid Degrading Normal Links 58 Vivek Shrivastava AB CD Hidden terminals Non- hidden/Non- exposed

59 (3) Avoid Degrading Normal Links 59 Vivek Shrivastava AB CD Scheduler

60 (3) Avoid Degrading Normal Links 60 Vivek Shrivastava AB CD Scheduler

61 (3) Avoid Degrading Normal Links 61 Vivek Shrivastava AB CD Scheduler

62 (3) Avoid Degrading Normal Links 62 Vivek Shrivastava AB CD Scheduler Hybrid scheduling

63 Outline A naïve attempt at centralized scheduling (DET) Our hybrid centralized scheduler (CENTAUR) Evaluation Related Work Summary 63 Vivek Shrivastava

64 Large-Scale Experiments Platform: Two WLAN testbeds in separate buildings Topology Representative: 7 APs, 12 clients Traffic and metrics UDP, TCP, VoIP, HTTP (real traces) Throughput, delay, MOS, web transaction delay 64 Vivek Shrivastava

65 Result 1: UDP/TCP Performance 65 Vivek Shrivastava

66 Result 1: UDP/TCP Performance 66 Vivek Shrivastava DCF

67 Result 1: UDP/TCP Performance 67 Vivek Shrivastava DCF

68 Result 1: UDP/TCP Performance 68 Vivek Shrivastava DCF Per Packet

69 Result 1: UDP/TCP Performance 69 Vivek Shrivastava DCF Per Packet

70 Result 1: UDP/TCP Performance 70 Vivek Shrivastava Epoch Based DCF Per Packet

71 Result 1: UDP/TCP Performance 71 Vivek Shrivastava Epoch Based DCF Per Packet Hidden terminal starves some clients

72 Result 1: UDP/TCP Performance 72 Vivek Shrivastava Epoch Based DCF Per Packet Better fairness than DCF

73 Result 1: UDP/TCP Performance 73 Vivek Shrivastava Epoch Based DCF Per Packet Exploits exposed terminals, higher system throughput

74 Result 1: UDP/TCP Performance 74 Vivek Shrivastava Epoch Based DCF Per Packet Avg. delay is smallest for epoch based scheduling

75 Result 1: UDP/TCP Performance 75 Vivek Shrivastava Epoch Based DCF Per Packet 90 th percentile delays are higher for epoch based scheduling

76 Result 1: UDP/TCP Performance 76 Vivek Shrivastava Epoch Based DCF Per Packet CENTAUR yields up to 60% higher total throughput and 50% lower per-packet delay compared to DCF

77 More Results in Paper Centaur micro-benchmarks: performance for exposed and hidden terminals under centaur Uplink traffic: coexistence and persistent gains with different fractions of uplink traffic Data rate: robust to changes in data rate and ARF Realistic HTTP traces: significant reduction in web transaction delay VoIP traffic: better performance (MOS) for voice traffic with small epoch duration 77 Vivek Shrivastava

78 Related Work Commercial WLAN offerings (Aruba, Meru) Theoretical formulations (Vaidya ‘00, Kanodia ‘01) Epoch based scheduling (Kompella ‘05, n/e) Interference mitigation (CMAP, SIC, Shuffle) 78 Vivek Shrivastava

79 Summary Interference a growing problem in enterprises Careful design of a centralized data plane provides substantial performance gains CENTAUR implements a hybrid data path to improve aggregate performance without client modifications CENTAUR does not disable carrier sense and co-exists with non-enterprise and uplink traffic 79 Vivek Shrivastava

80 Future work Even more efficient conflict graph generation What if we were allowed client modifications ? Questions ?

Characterizing System Latencies System delays are high and variable, leading to inaccuracies for per-packet scheduling 81

82 Evaluation of Micro-Probing 82 Vivek Shrivastava Topologies20 node30 node Bandwidth Tests 16.2 mins1hr 11 mins Micro- Probing ~4 secs~11 secs Can be computed in stages with each instance taking ~ 2.5ms

83 Result 2: Impact of Uplink Traffic 83 Vivek Shrivastava Vary proportion of downlink/uplink traffic 6 different configurations 80/20 ➔ 40/60 (downlink/uplink) Results: Downlink: 1.6x ➔ 6.8x gain in throughput Uplink: 1x ➔ 1.18x gain in throughput CENTAUR provides persistent gains for different proportions of uplink and downlink traffic load

84 Three topologies Hidden Heavy topology ➔ 10 links Exposed Heavy topology ➔ 6 links Mixed Topology ➔ 12 links Results: Up to 50% gain in overall system throughput Up to 6x gain for HT; Up to 1.7x gain for ET Result 3: Impact of Topology 84 Vivek Shrivastava Improvements from using CENTAUR can be seen across many different network topologies

85 Result 2: Impact of Uplink Traffic 85 Vivek Shrivastava CENTAUR provides persistent gains for different proportions of uplink and downlink traffic load

86 Result 3: Impact of Topology 86 Vivek Shrivastava Improvements from using CENTAUR can be seen across many different network topologies

87 CENTAUR Micro-Benchmarks (1) 87 Vivek Shrivastava Exposed Terminals

88 CENTAUR Micro-Benchmarks (II) 88 Vivek Shrivastava Hidden Terminals

89 Result 1: UDP/TCP Performance 89 Vivek Shrivastava Client Index (1-12) CENTAUR yields up to 60% higher total throughput and 50% lower per-packet delay compared to DCF

Other approaches to Hidden/Exposed Terminals Mechanis m Target Problem Approach Client Changes Evaluation CMAP [NSDI ‘07] ZigZag [Sigcomm ’08] SIC [Mobicom ’08] Centau r Expose d Hidden Expose dHidden Disable CS Signal manipulation Centralized Scheduling Yes No GNU Radio Adaptive RTS/CTS [VTC ’03] Hidden Extra Signaling Yes

March, Prelim 2009 Other approaches to Hidden/Exposed Terminals Mechanis m Target Problem Approach Client Changes Evaluation CMAP [NSDI ‘07] ZigZag [Sigcomm ’08] SIC [Mobicom ’08] Centau r Expose d Hidden Expose dHidden Disable CS Signal manipulation Centralized Scheduling Yes No GNU Radio Adaptive RTS/CTS [VTC ’03] Hidden Extra Signaling Yes Solve both hidden/exposed

March, Prelim 2009 Other approaches to Hidden/Exposed Terminals Mechanis m Target Problem Approach Client Changes Evaluation CMAP [NSDI ‘07] ZigZag [Sigcomm ’08] SIC [Mobicom ’08] Centau r Expose d Hidden Expose dHidden Disable CS Signal manipulation Centralized Scheduling Yes No GNU Radio Adaptive RTS/CTS [VTC ’03] Hidden Extra Signaling Yes No client side changes for Centaur

CENTAUR: Realizing the Full Potential of Centralized WLANs Through a Hybrid Data Path Vivek Shrivastava*, Shravan Rayanchu, Suman Banerjee University of Wisconsin-Madison 93 Nabeel Ahmed, Srinivasan Keshav University of Waterloo, Ontario Konstantina Papagiannaki Intel Labs, Pittsburgh Arunesh Mishra Google Inc.

94 WLANsHP Labs Seoul National University Our Testbed Exposed Terminals 39%9%50% Hidden Terminals 43%70%39% Growth of Interference in Enterprise Wireless LANs 94 Interference an increasing problem according to leading enterprise WLAN vendor Vivek Shrivastava

95 Centralization of Enterprise WLANs 95 Centralized control for better network security and manageability Vivek Shrivastava

96 Can Centralized Data Plane Scheduling Help? 96 Vivek Shrivastava

97 (2) Improve Exposed Terminals 97 Vivek Shrivastava Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets to each exposed AP Fix back-off periods and use carrier-sensing to align transmissions -- double throughput!

98 Related Work Commercial WLAN offerings (Aruba, Meru) Research proposals (MiFi, DenseAP, Smarta) TXOP in e/802.11n packet aggregation Interference Mitigation (CMAPs, ZigZag, SIC) 98 Vivek Shrivastava

99 Types of Interference Hidden Terminals Collisions Carrier Sense Exposed Terminals Focus on downlink conflicts Vivek Shrivastava 11

100 Can Centralized Data Plane Scheduling Help? 100 Vivek Shrivastava X Y

101 Can Centralized Data Plane Scheduling Help? 101 Vivek Shrivastava X Y

102 Can Centralized Data Plane Scheduling Help? 102 Vivek Shrivastava X Y

103 Vivek Shrivastava Can Centralized Data Plane Scheduling Help? Hidden terminals

104 Quantifying Downlink Hidden Terminals 104 Vivek Shrivastava 10% links suffer severe hidden terminal interference

105 Quantifying Downlink Exposed Terminals % links can obtain double the throughput with CS disabled, indicating exposed terminal interference Vivek Shrivastava

106 Result 1: UDP/TCP Performance 106 Vivek Shrivastava

107 Result 1: UDP/TCP Performance 107 Vivek Shrivastava

108 Result 1: UDP/TCP Performance 108 Vivek Shrivastava

109 Centralization of Enterprise WLANs 109 Vivek Shrivastava

110 (2) Improve Exposed Terminals 110 Vivek Shrivastava Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets to each exposed AP Fix back-off periods and use carrier sensing to align transmissions -- double throughput!

111 (2) Improve Exposed Terminals 111 Vivek Shrivastava Problem: Exposed links can operate in parallel but don’t due to carrier-sensing Solution: Schedule batch of packets to each exposed AP Fix back-off periods and use carrier sensing to align transmissions -- double throughput!

112 (3) Avoid Degrading Normal Links 112 Vivek Shrivastava

Can it solve any interference problems prevalent in enterprise WLANs Hidden and exposed terminals 88% of links suffer some losses due to co- channel interference (Jigsaw, Sigcomm 2007) 2. If so, how can we implement it efficiently for practical WLAN deployments No client modifications, support legacy clients Coexistence with non-enterprise, uplink traffic 113 Vivek Shrivastava Can Centralized Data Plane Help?

114 Quantifying Downlink Interference Prior Work: Jigsaw [Sigcomm2006] analysis reveals 56% of all interference traffic is downlink in nature. Our Work: Two production WLANs W1: 5 floors, 9 APs, 45 clients W2: 1 floor, 21 APs, 51 clients Download ‘bulk’ traffic from the Internet 114 Vivek Shrivastava

115 WLANsUW-MadisonUW-Ontario Exposed Terminals 39%9% Hidden Terminals 43%70% Growth of Interference in Enterprise Wireless LANs 115 Interference an increasing problem according to leading enterprise WLAN vendor Vivek Shrivastava

116 Centralization of Enterprise WLANs 116 Vivek Shrivastava Can centralized data plane be useful for improving performance in WLANs ? About 70-80% of enterprise traffic is downlink in nature

117 (3) Avoid Degrading Normal Links 117 Vivek Shrivastava AB CD Scheduler

118 (3) Avoid Degrading Normal Links 118 Vivek Shrivastava AB CD Scheduler

119 (3) Avoid Degrading Normal Links 119 Vivek Shrivastava AB CD Scheduler

120 (3) Avoid Degrading Normal Links 120 Vivek Shrivastava AB CD Scheduler Hybrid scheduling

121 Outline DET - A deterministic scheduler CENTAUR - A hybrid centralized scheduler Evaluation Related Work Summary 121 Vivek Shrivastava

122 Outline DET - A deterministic scheduler CENTAUR - A hybrid centralized scheduler Evaluation Related Work Summary 122 Vivek Shrivastava

Performance of DET 4x No gains for exposed terminals; Non- conflicting links perform worse under load 123 Vivek Shrivastava

124 Outline DET - A deterministic scheduler CENTAUR - A hybrid centralized scheduler Evaluation Related Work Summary 124 Vivek Shrivastava

125 DET: A Simple Deterministic Scheduler Key Idea: Perform per-packet scheduling Given Conflict graph G = (L,E) Scheduled packets {P 1, P 2,..., P r }; Unscheduled P r+1 Objective Minimize t(P r+1 ) Constraint: No two packets on interfering links are scheduled together Schedule downlink packets only 125 Vivek Shrivastava

What are the problems it can solve ? Hidden and exposed terminals 2. Can we implement it efficiently ? No client modifications Coexistence with non-enterprise & uplink traffic Don’t disable carrier sensing 126 Vivek Shrivastava Can centralized scheduling help?

127 Vivek Shrivastava Can centralized scheduling help?

128 Vivek Shrivastava DET: A Simple Deterministic Scheduler Packets arriving Schedule one packet at a time Consider packets in order of arrival 1234 Transmission slots

129 What about exposed terminals ? % links can obtain double the throughput with CS disabled, indicating exposed terminal interference Vivek Shrivastava

130 Vivek Shrivastava Use an in-band scheduler Simple FIFO schedule with interference avoidance scheduler