MAC Enhancements to Support Quality of Service in Wireless Networks

1
MAC Enhancements to Support Quality of Service in
Wireless Networks

• Masters Thesis Presentation
• S.Rajesh
• AU-KBC Research Centre
• http//www.au-kbc.org
• http//www.annauniv.edu
• Department of Electronics Engineering,
• Faculty of Information and Communication
  Engineering, MIT Campus, Anna University,
  Chromepet, Chennai, TN 600044 INDIA.

2
Outline

• Introduction
• MAC for wireless networks
• Ad hoc networks
• Wireless LAN
• Problem Definition
• QoS support
• Differentiated
• Integrated
• MAC Enhancements
• In ad hoc networks with directional antennas
• System model
• Results
• In WLAN with QoS scheduler
• System model
• Results
• Conclusion

3
INTRODUCTION

• Introduction
• Problem Definition Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc / WLAN)
• Conclusion

4
Introduction

• WLAN Standard with QoS Enhancement
• Basics IEEE 802.11
• QoS enhancements in IEEE 802.11e
• HCF
• HCF Contention Free Channel Access Mechanism
• HCF Contention Based Channel Access Mechanism
  (EDCF)
• (or Enhanced Distributed Coordination Function)
• Scheduling Techniques
• Prioritized Scheduling for Differentiated and
  Integrated Traffic
• Rate Adaptive Scheduling

5
WLAN

• Typical Scenarios
• Independent BSS
• IBSS (Ad hoc mode)
• Distributed coordination
• Infrastructure BSS
• NOT called an IBSS
• Central coordination

6
ESS
LAN
Internet
Backbone
ESS
WLAN
BSS
BSS
7
Basic IEEE 802.11 MAC

• CSMA/CA
• Binary Exponential Back-off
• RTS/CTS/Data/ACK handshake
• Modes
• DCF
• Ad Hoc or infrastructureless
• PCF
• Infrastructure based
• Access Point polls the associated stations

8
DCF

D S
S
S D
N ST S
S
R
DATA
T
S
1
NAV (RTS from 3)
C
A
C
T
C
T
S
K
S
2
R
DATA
T
S
3
NAV (RTS from 1)

D S
S
S D
BkOff S
S time
à
802.11a parameters
m
S
à
SIFS (Short Inter Frame Space) 16
s
m
D
DIFS (DCF Inter Frame Space) 34
s
à
m
ST
Slot Time for each Back-off counter 9
s
à
9
DCF

• Beacon
• generated by any of the nodes in the IBSS
• MPDU transmission
• If channel is free for a DIFS transmit (RTS,,
  Data,...)
• else
• wait till it becomes free for a DIFS
• generate random backoff slot-times in
  (0-Cwmin)
• if channel is free count down one slot time
• else freeze and resume countdown
• the channel becomes free for a DIFS
• on reaching zero transmit
• if failed retry from first step (for max retries
  (7) times)

10
PCF
11
PCF

• Beacon
• always generated by the AP
• Transmission
• AP transmits Multicast/Broadcast data first
• AP transmits data to associated stations one by
  one and along with that it polls these stations
  to send data if any in contention free mode
• If the station doesnt respond within PIFS, the
  AP gets the channel with better chance as
  PIFSltDIFS
• After CFPmaxduration channel is left for
  contention based access
• Contention Free Period and Contention Period
  alternate

12
HCF
Contention Free Repetition Interval
Contention period
Contention Free Period
SIFS SIFS
SIFS PIFS
SIFS SIFS
AIFSBK SIFS SIFS SIFS

Beacon
Data CF Poll to 1
CF Poll to 2 Ack to 1
CF Poll to 3
CF End
PC NAV M1 NAV M2 Dead NAV M3 NAV M4,5 IBSS
NAV
Set by Beacon
Reserved by TXOP
Reserved by TXOP
Cleared by CF End
Reserved by TXOP
Data from 1 CF Ack
Set by Beacon
Reserved by TXOP
Reserved by TXOP
Cleared by CF End
CF Ack
Set by Beacon
Reserved by TXOP
Reserved by TXOP
Cleared by CF End
R T S
C T S
Data
A c k
Set by RTS
Time --gt
13
EDCF Queuing
14
AIFS
15
Problem Definition and Contribution

• Introduction
• Problem Definition Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc / WLAN)
• Conclusion

16
Problem Definition

• Link level QoS support in
• Ad hoc networks
• WLANs
• Differentiated
• Access Category based
• Integrated
• Guaranteed QoS

17
Contribution

• MAC enhancements to support QoS
• In ad hoc networks
• Using directionality of the directional antenna
• Using intermittent immobile nodes
• Using direction aware scheduler
• In WLAN
• Using estimation based
• Scheduler linked with
• Traffic shaping and policing
• Admission Control

18
MAC Enhancements in Ad hoc Networks

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (Ad hoc Networks)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc / WLAN)
• Conclusion

19
Solution - Structuring

• Ad hoc networks
• Structural aspects
• topology free, infrastructure independent
• Functional aspects
• multi-hop, common frequency band for all nodes,
  no central coordination
• More overhead / expense on
• routing, MAC, power consumption
• due to
• highly dynamic state and random state
  transitions, distributed coordination

20
Ad hoc Network - Challenges
21
Enhancements

• Use
• interspersed stationary nodes
• to reduce probability of any region getting void
  of even a single node to connect with
• directional antenna in these nodes
• to improve range (without power-back-off)
• to improve frequency reuse (with power-back-off)
• smart directionality scheduler
• to help high priority node(s) or traffic to gain
  access
• to prevent starvation of lower priority node(s)
  or traffic

22
contd
23
System Model Ad hoc Networks

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc Networks)
• Results (Ad hoc / WLAN)
• Conclusion

24
Antenna Patterns

• Omni-directional
• Directional Antenna
• Beam
• single
• multi
• Power
• same as omni-directional
• backed-off / increased

25
MAC Based on Antenna

• Omni-directional (e.g. RTS/CTS/Data/Ack, CSMA/CA
  as in DCF of IEEE 802.11)
• Directional
• Static directionality
• based on
• node distribution where node density is more
• need for bridging or relaying
• Dynamic directionality
• based on
• (source, destination) pairs
• Traffic
• Traffic intensity for uniform traffic
• Traffic category

26
State diagram of enhanced MAC
27
Results in

• Improved connectivity
• Improved QoS

28
Connectivity Improvement with Stationary Nodes

• Probability
• that two mobile nodes contact at single hop
• that a mobile and a immobile node contact at
  single hop
• Improvement factor in contact probability

29
Connectivity Improvement with Directional Nodes

• Coverage radius
• Romni 100m
• Rdirectional 500m
• Rnetwork 1000m
• Single hop probability with
• Omni-directional nodes
• ?1002/?10002 1/100
• Directional nodes
• ?5002/?10002 1/4
• Improvement factor 52 or (Romni/Rdirectional)2

30
Traffic Intensity Calculation
31
Traffic Based Direction Scheduling for Better QoS
?p is the priority weight of the corresponding
traffic class
32
Scenario

• All mobile case
• With intermittent mobile nodes
• without directional antennas
• with directional antennas
• with smart traffic-intensity based scheduling
• with smart traffic-category based scheduling

33
Simulation Parameters

• Network Diameter
• Antenna Pattern
• Omni-directional radius
• Directional
• beam width (lower limit)
• range (upper limit)
• Access Method
• Routing Technique

• 2000m
• 100m
• 2?/6.25 for reaching
• 250m with same power
• 802.11 MAC
• CSMA/CA
• RTS/CTS/Data/ACK
• Shortest Path

34
Contd...Simulation Parameters
More Delay sensitive
Less Delay sensitive
Non Preemptive scheduling
Preemptive scheduling
FCFS
Scheduler
35
Results

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc Networks)
• Conclusion

36
Throughput Performance
37
Delay Performance
38
MAC Enhancements in WLANs

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (WLANS)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc / WLAN)
• Conclusion

39
Solution Structuring

• To design a common scheduler
• that can handle both
• (a) Contention free traffic and
• (b) Contention based traffic
• or
• (1) Traffic with resource reservation and
• (2) Traffic without resource reservation
• Though not necessary, (1) is handled using (a)
  and (2) using (b).
• Exceptionally some bursts are allowed for (1) in
  (b) also called CFB or Contention Free Bursts

40
System - Block Diagram
41
Traffic Flow

• Traffic Classification
• Traffic corresponding to declared Traffic Streams
  (TSs)
• Shaped and Policed using Twin Token Bucket
• Sent as per TS scheduler in HCF
• Traffic not associated with Traffic Streams (TSs)
• RED queue mechanism used
• Sent as per EDCF budget declared by HC in HCF

42
Twin Token Bucket

• Traffic
• Shaping
• Policing

Bucket 1 Bucket 2 Token filling rate
(Constant) r1 Peak Data Rate r2 Mean Data
Rate Bucket Size s1 1 token (mimic leaky
bucket) s2 Maximum Burst Size tokens Token
extraction rate - At most Peak Data
Rate Major purpose Rate limiting Burst size
limiting
43
Scheduling Based on Packet Error Information

• Scheduler schedules and admits Traffic Streams
  based on effective bandwidth
• Effective Mean Data Rate (EMDR)
• control factor ?n is varied based on observed
  packet errors

44

• EMDR estimate 54/(1e-?n)
• ?n ?n-1 x
• where,
• x 1 if successful, -1 if unsuccessful
• In implementation
• ?n ranging to infinity can not be realized,
• ?n should itself adapt based on channel condition
• So, ?n is
• upper limited to /- 127
• replaced by ? which is a function of deviation
  in ?n

45

• Aggregate the mean and peak data rate
  requirements mentioned through TSPEC for each
  admitted TS
• Set rate of token filling in second bucket in
  Twin Token bucket , r2 to
• max(Estimated EMDR, algebraic sum of mean data
  rates of admitted TSs)

46
Admission Policy

• Admit Traffic Streams until aggregate mean data
  rate of existing traffic streams does not exceed
  EMDR,
• (reject otherwise).
• Bandwidth not used for TS is allocated through
  EDCF budget for Contention based access

47
Scheduling Based on Rate Adaptation Information

• Typically multiple rates are allowed
• 54, 48, 36, 24, 18,12, 9, 6 Mbps
• Rate adaptation is done based RSSI or other
  techniques
• ?n ?n-1 x
• where, x x(54106)/r
• where,
• x 1 if successful, -1 if unsuccessful
• and
• r is the rate of transmission of previous packet

48
Contd...Simulation Parameters
More Delay sensitive
Less Delay sensitive
Non Preemptive scheduling
Preemptive scheduling
FCFS
Scheduler
49
Results

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc / WLAN)
• Results (WLAN)
• Conclusion

50
Goodput of EDCF
51
Traffic Streams Supported for Different Peak Data
Rate Deviation on Ideal Channel
52
Traffic Streams supported and EDCF Throughput for
a network when scheduler in HCF handles TXOPs of
both contention free and contention based
categories
53
CONCLUSION

• Introduction
• Problem Definition and Contribution
• MAC Enhancements (Ad hoc / WLAN)
• System Model (Ad hoc / WLAN)
• Results (Ad hoc / WLAN)
• Conclusion

54
Conclusion

• In ad hoc networks
• Directional MAC - simple robust technique to
  improve
• Connectivity / Capacity
• QoS performance
• In WLANs
• a scheduler with knowledge of
• packet error information performs good
• rate adaptation mechanism provided by the
  management layer could help in better performance
  
• particularly in poor channel conditions

55
Node Interactions with a Directional Node

• Directional node
• Omni-directional nodes
• reached by the directional node
• not reached by the directional nodes but
• exposed
• not exposed and
• forming independent local network(s)
• astray

56
Service Provider Network

• Building support infrastructure
• Combination of
• Stationary nodes
• omni-directional (support in dense areas with
  less or no mobility)
• directional
• static (relays on highways)
• dynamic (in areas with highly random mobility)

57
References

• 1 M. Grossglauser and D. Tse, Mobility
  increases the capacity of ad-hoc wireless
  networks, Proc. INFOCOM, pp. 1360-1369, April
  2001.
• 2 M. Sanchez, T. Giles and J. Zander, CSMA/CA
  with Beam Forming Antennas in Multi-hop Packet
  Radio, Proc. Swedish Workshop on Wireless Ad hoc
  Networks, March 2001.
• 3 O. Somarriba, Multihop Packet Radio Systems
  in Rough Terrain", Licentiate Thesis, Radio
  Communication Systems, Department of S3, Royal
  Institute of Technology, Sweden, Oct. 1995.
• 4 L. E. Miler, Multihop Connectivity of
  Arbitrary Networks", Multihop Connectivity, NIST,
  March 2001.
• 5 IEEE 802.11b, Part 11 Wireless LAN MAC
  and PHY Specification High-Speed Physical Layer
  Extension in the 2.4GHz Band", 1999.
• 6 IEEE 802.11e, Part 11 Wireless LAN MAC
  and PHY Specification MAC Enhancements for QoS,
  D3.3, Oct 2002.

58
References
59
Thank You

• S.Rajesh,
• AU-KBC Research Centre,
• Department of Electronics Engineering, Faculty of
  Information and Communication Engineering, MIT
  Campus, Anna University, Chromepet, Chennai, TN
  600044 INDIA.
• http//www.au-kbc.org
• http//www.annauniv.edu

60
?!