Time-based%20Fairness%20Improves%20Performance%20in%20Multi-rate%20WLANs

1
Time-based Fairness Improves Performance in
Multi-rate WLANs

• Godfrey Tan and John Guttag
• MIT Computer Science Artificial Intelligence
  Laboratory

2
Outline

• Multi-rate WLANs support variable rates
• Problems with throughput-based fairness
• Alternate notion time-based fairness
• What it is
• Why it s good
• How to achieve it (Time-based Regulator)
• Evaluation

3
WLANs Facilitate Varying Speeds
Sending at 5.5 Mbps is better
TCP Throughput with RTS/CTS (Mbps)

• Tradeoff between data rate and loss rate
• Multiple standards compete in same channel
• e.g. 802.11b vs. 802.11g

4
AP sees Multiple Rates
Percentage of Bytes Transmitted

• Card manufactures implement auto-rate protocols
• Varying channel conditions at clients lead to
  rate diversity

5
Aggregate Throughput Reduced
6
Aggregate Throughput Reduced
7
Aggregate Throughput Reduced

• Total throughput lower than expected
• Faster node suffers
• Slower node benefits
• Less incentive to upgrade to 802.11g

8
Root Cause DCFs Fairness Notion

• Carrier sense multiple access protocol
• Distributed randomized access
• Goal equal number of frame transmissions
• Aim seems to be throughput-based fairness
• Assuming equal frame size and loss rate
• Irrespective of frame transmission time

• Consequence Aggregate thruput closer to slower
  nodes

9
Throughput-based Fairness (RF)

• Nodes achieve equal throughputs
• Suitable for
• Wired networks
• Single-rate wireless LANs

Ri is achieved throughput ?j js maximum
achievable throughput I the set of competing
nodes
10
Not Efficient Maybe Not Fair

• Throughput of node i should depend upon
• Number of competing nodes
• Transmission strategy used by node i
• Should not depend upon
• Transmission strategies used by other nodes
• Channel time is the shared resource
• Transmission opportunities are not

11
Time-based Fairness (TF)

• Nodes achieve equal channel time shares

Ri is achieved throughput ?i is maximum
achievable throughput I the set of competing
nodes

• Desirable in multi-rate WLANs
• Nodes throughput depends only upon
• Its transmission strategy
• Number of competing nodes

12
Throughputs Unchanged in Single-rate WLANs
11vs11
1vs11
1vs1
5.5vs5.5
2vs2
13
TF Improves Throughput in Multi-rate WLANs

• Total throughput improves by 115
• Faster node achieves 273 more
• Slower node achieves 42 less

11vs11
1vs11
1vs1
14
TF does not favor slower nodes

• Under RF, n1 achieves 84 of channel time
• Under TF, each node achieves 50

11vs11
1vs11
1vs1
15
Outline

• Multi-rate WLANs support variable rates
• Problems with throughput-based fairness
• Alternate notion time-based fairness
• What it is
• Why it s good
• How to achieve it (Time-based Regulator)
• Results

16
How to Achieve Time-based Fairness?

• Is tweaking DCF enough?
• Each node still achieves equal chance to transmit
• Number of transmissions depends on data rate
• Faster node can transmit more in each opportunity

• No! Not enough for AP-based WLANs!
• Downlink frames are transmitted at varying data
  rates
• Existing queuing schemes lead to thruput-based
  fairness

• AP's queuing scheme needs modifications

17
How to Achieve Time-based Fairness?

• Is having N queues at the AP enough?
• One queue for each data rate
• Faster queue gets dequeued more in each round
• Dequeue 6 packets from 11-Mbps-queue 1 from
  1-Mbps-queue

• No!
• Non-uniform client distribution at queues
  problematic
• E.g. 6 users at 11 Mbps and 1 user at 1 Mbps
  leads to RF

• Per-client queuing, monitoring and policing
  necessary

18
Our Time-based Regulator (TBR)

• AP shapes traffic to clients, i.e. downlink only
• Monitors channel time usage of each client
• Account both downlink and uplink traffic
• Deal with differing loss rates and varying
  demands
• Transmit frames to node i
• Only if it has not utilized its share of channel
  time

19
Is Shaping Downlink Traffic Enough?

• Yes for feedback-based congestion controlled apps
• Limiting rate of downlink traffic slows sending
  rate
• Regardless of clients traffic directions
• E.g. Applications using TCP, RTCP, etc.
• No for non-congestion controlled apps
• Modify clients so that AP can ask them to slow
  down
• Drop packets if clients do not react
  appropriately
• E.g. Applications using raw UDP
• TCP makes up 90 of WLAN traffic Tang02,Kotz02

20
A TBR Implementation

• Only runs at AP No modifications to clients
• Uses leaky buckets to shape downlink traffic
• Sets up a queue for each client
• Works with DCF
• Implemented in Linux HostAP Driver

21
TBR Impelementation Cont.

• tokensi available channel time (seconds not
  bits)
• ratei channel time share (e.g. 1/n)
• bucketi maximum amount of tokens
• Policing
• Packet to node i is transmitted if tokensi gt 0
• Tokens are periodically filled at ratei
• Accounting
• For each packet P transmitted, tokensi -
  chantime(P)

22
Example TBR Operations
At t 0, rate1 0.5 rate2
0.5 tokens1 0.025 tokens2 0.025
AP
11
1
11
1 Mbps
11
TCP Data
At t 0.074, tokens1 0.025 0.037 0.062 0
tokens2 0.025 0.037 0.012 0.05
1 Mbps
1
11
11
TCP Ack
TCP Ack
1 Mbps
1
11
TCP Data
11
11
1
From this time onwards, n1 can only use 50 of
channel time.
n1
n2
23
Computing Channel Occupancy Time

• Total time used to transfer each layer-2 frame

layer-2 ack
layer-2 frame
Idle
Per-frame Channel Occupancy Time

• Take into account retransmissions
• AP knows lost frames in downlink direction
• For uplink direction
• Client marks each header with retry info., or
• AP estimates based on heuristics

24
Dealing with Varying Traffic Conditions

• Not all nodes need 1/n of capacity
• Achieves time-based max-min allocation
• Smallest ratei must be as large as possible
• Second smallest ratej must be as large as
  possible, etc.
• Periodically adjusts ratei to fully utilize
  channel
• If under-utilized, ratei is reduced
• Excess capacity redistributed among other nodes

25
TBR Achieves Higher Downlink Throughputs
TBR achieves higher throughputs as analytically
predicted
5.5vs11
2vs11
1vs11
26
TBR Achieves Higher Uplink Throughputs
5.5vs11
2vs11
1vs11
27
Related Work

• Performance anomaly of 802.11b
• Heusse et al., Infocom03
• Opportunistic MAC protocol
• Sadeghi et al., Mobicom02
• 802.11e Qos Support (being drafted)

28
Conclusions

• Time-based fairness is desirable
• Better overall system performance
• In terms of throughput and completion time
• Faster nodes see significant improvement
• More incentive to upgrade to 11g
• Slower nodes not penalized severely
• APs shape downlink traffic to achieve TF
• Uplink downlink must both be considered
• No modifications to clients or DCF necessary

29
TF Improves Wait Time in Multi-rate WLANs

• n1 transfers X bytes _at_1 Mbps at t0
• n2 transfers X bytes _at_11 Mbps at t0

• Under time-based fairness,
• n2 completes earlier at t1
• n1 completes later at t2
• Under throughput-based fairness
• Both n1 and n2 complete at t2

t0
t1
t2
n2
n1
n2
n1
of channel time used by n1
30
TBR Keeps Channel Utilization High
TCP Throughput (Mbps) Varying sending rates TBR with DCF Max-min Allocation
n1 at 11 Mbps As fast as possible 2.95 2.97
n2 at 11 Mbps 2.1 2.11 2.1
Total 5.06 5.07

• Achieves max-min allocation
• Adapts to varying demands
• Redistributes excess capacity of underutilized
  nodes

31
TDMA

• TDMA provides equal time slots to clients each
  round
• Converges to time-based fairness
• If every node utilizes the entire slot each round
• Not very suitable for bursty traffic
• Time-based fairness notion
• Provides predictable long-term channel time
  shares
• Under bursty traffic, varying demands and loss
  rates
• Compatible with any MAC protocol
• CSMA (e.g. DCF)
• TDMA (e.g. HiperLAN)

32
TF does not favor slower nodes
11vs11
1vs11
1vs1
33
Traffic Models

• Fluid Model
• Finite number of flows transfer infinite streams
• Efficiency measured by aggregate throughput
• Corresponds to very busy networks
• Task Model
• Finite number of flows transfer finite number of
  bits
• Efficiency measured by average final completion
  time
• Corresponds to sometimes congested networks

34
Comparison
Criteria Measure Throughput-based Time-based
Fairness Throughput Channel Time Better Worse Worse Better
Efficiency (fluid) AggrThruput Worse Better
Efficiency (task) FinalTaskTime AvgTaskTime Same Worse Same Better
35
Long-term Time-share Guarantees Necessary
36
?(d, s, I) baseline throughput

• Depends on data rate, frame size, contention and
  channel conditions
• Maximum total achieved throughput