Dept' of Computer Science

1
QoS Provisioning for Mobile Wireless Ad Hoc
Networks (MANET)

• Chao Gui
• Jian Li
• Instructor Prasant Mahapatra

2
Outline

• Adapted QoS Model for MANET
• QoS from a Layered Perspective
• Cross-Layer Schemes
• Conclusion and Future Work

3
Types of QoS Metrics

• Concave metrics (e.g. bandwidth)
• Additive metrics (e.g. delay or jitter)
• NP-complete problem if QoS Spec contains at least
  two additive metrics

4
Differences and Difficulties in MANET

• Link properties unpredictable
• Hidden terminal problem
• Routing is based on imprecise information
• Rout maintenancedealing with mobility
• Limited battery life
• Security challenges

5
Adapted QoS Model (1) Soft QoS
6
Adapted QoS Model (2) Dynamic QoS
7
Outline

• Adapted QoS Model for MANET
• QoS from a Layered Perspective
• Cross-Layer Schemes
• Conclusion and Future Work

8
Physical Layer Issues

• Channel estimation
• Current channel state
• Accurate estimation at receiver side
• Feedback of estimation to sender side
• Coding schemes
• Higher source coding rate
• More channel protection

9
MAC Layer Issues

• Requirements on MAC protocols for MANETs
• Distributed
• Asynchronous
• Solve hidden terminal problem

10
QoS-aware MAC protocols

• IEEE 802.11 Real-time extension
• Black burst contention scheme
• MACA/PR (Multihop Access Collision Avoidance with
  Piggyback Reservation)

11
Black Burst

• Provides differentiation among real-time flow and
  best-effort flow
• Provides fairness and priority scheduling among
  real-time flows
• Fully distributed

12
Black Burst
A
C
B
DIFS for real-time flow
Media
Busy
A
B
C
13
Black Burst
A
B
C
DIFS
Media
Busy
A
B
C
14
Black Burst
A
B
C
DIFS
Media
Busy
A
B
C
15
Black Burst
A
B
C
DIFS
Media
Busy
A
B
C
16
Black Burst
A
B
C
DIFS
Media
Busy
A
B
C
17
Black Burst
A
B
C
DIFS
RT A
Media
Busy
RT A
A
B
C
18
Black Burst
A
B
C
DIFS
RT A
Media
Busy
RT A
A
B
C
19
Black Burst
A
A
B
C
B
C
DIFS
RT A
Busy
RT A
A
C
20
Black Burst
A
A
B
C
B
C
DIFS
RT A
Media
Busy
RT A
A
B
C
21
Black Burst
A
A
B
C
B
C
DIFS
RT A
Media
Busy
RT A
A
B
C
22
Black Burst
A
A
B
C
B
C
DIFS
RT A
Media
Busy
RT A
A
B
C
23
Black Burst
A
A
B
C
B
C
DIFS
RT A
RT B
Media
Busy
RT A
A
RT B
B
C
24
Black Burst
A
A
B
C
B
C
DIFS
RT A
RT B
RT C
RT A
Media
Busy
RT A
RT A
A
RT B
B
RT C
C
25
Network layer issuesQoS-aware Routing
Source A DestinationB Bandwidth ? 4
26
QoS-aware Routing Protocols

• CEDAR (Core Extraction Distributed Ad hoc
  Routing)
• Ticket-Based Probing Algorithm
• QoS Support with Bandwidth Calculations

27
Ticket Based Probing
i
t
s
j
28
Ticket Based Probing

• Routing overhead is controlled by number of
  tickets
• Multipath search is tolerant to imprecise
  information
• No centralized control and management

29
Ticket Splitting

• Tickets are distributed according to the chances
  of leading to low-cost feasible paths
• A neighbor having smaller end-to-end delay is
  given more tickets (delay-constrained routing)
• A neighbor having stationary link to destination
  is given more tickets

30
Tickets Generation

• How many tickets to generate?
• Two types of tickets

31
Ticket Generation (Delay Constrained Routing)
Y
?
1
D
estimated end-to-end delay range
G
?
1
D
32
Ticket Generation (Bandwidth Constrained Routing)
Y
?
1
B
estimated end-to-end bandwidth range
G
?
1
B
33
Transport Layer Issues

• Adaptation to QoS fluctuations
• Improve TCP performance for MANET
• New packet loss model
• Explicit feedback machenisms
• Route Failure Notification
• Delay feedback congestion control

34
Delay Feedback Congestion Control
delay
throughput
delay knee
congestion control cliff
load
TCP congestion control
delay feedback congestion control
35
Application Layer Issues

• Adaptive real-time audio/video streaming
• Compression algorithm features
• Layered encoding
• Rate shaping
• Adaptive error control
• Bandwidth smoothing

36
Other Application Layer Techniques

• Joint source-channel coding
• Joint source-network coding

37
Outline

• Adapted QoS Model for MANET
• QoS from a Layered Perspective
• Cross-Layer Schemes
• Conclusion and Future Work

38
Cross Layer Schemes

• Integrated QoS framework for MANETs
• INSIGNIA
• iMAQ

39
INSIGNIA

• IP-based QoS framework
• In-band signaling
• Soft-state resource management
• Goal support adaptive services which can provide
  base QoS (i.e. min bw) assurances to real-time
  voice and video flows and data to be delivered
  when resources become available.

40
In-band signaling vs. Out-of-band signaling

• Out-of-band signaling
• Quick reaction
• High overhead
• In-band signaling
• Low overhead
• INSIGNIA can restore flow-state with this in
  response to topology changes with the interval of
  a few consecutive IP packets.

41
Soft State

• Models the transient nature of resource
  reservation in MANET
• Resource reservations attempt to establish
  soft-state.
• Subsequent reception of data packets (associated
  with the reservation) at that router are used to
  refresh the existing soft-state reservation.

42
INSIGNIA
43
Outline

• Adapted QoS Model for MANET
• QoS from a Layered Perspective
• Cross-Layer Schemes

• Conclusion and Future Work

44
Questions for Future Research

• Will every node in MANET be equal in capacity and
  functionality?
• What can misbehaving nodes do?
• How to support multiple levels of QoS?

45
Thanks!