QoS issues in Mobile Ad Hoc Networks

1
QoS issues in Mobile Ad Hoc Networks

• Navid NIKAEIN
• Keywords Manet, QoS, QoS Model, DiffServ,
  IntServ, FQMM, Routing, Metrics, Cross-Layer
  design.

_at_
2
What is QoS ?

• A set of services requirements to be provided by
  the network to the application while transporting
  (routing) a packet stream (flow) from source to
  destination
• Internet today is connection- and state-less
• Each packet is routed independently (per-packet
  processiong)
• QoS is meaningful for a flow
• Packets belonging to the same flow need to be
  identified in order to receive the same service
• Necessitate a logical association for the packets

3
QoS Components

• QoS Model
• specifies an architecture and services to
  provided by the model
• QoS Routing
• Search for routes with enough resources
  respecting network constraints but does not
  reserve it
• QoS Resource Reservation Signaling
• Reserve the required resources along the selected
  path to guarantee the service requirements ?
  logical association
• QoS MAC
• Admission control
• Scheduling

4
QoS Model- Integrated Services

• IntServ Per-flow end-to-end guarantee
• First characterize the traffic, and then setup
  paths and reserve resources (RSVP)
• Guaranteed service for delay-sensitive
  applications
• Controlled-load service reliable and enhanced
  best-effort service
• Requires four components
• Signalling protocol, Admission control,
  Classifier, Packet scheduler
• Amount of state information increases
  proportionally with number of flows ? Reduce
  Scalability

5
QoS Model- Differentiated Services

• DiffServ Per-class service differentiation
• Packet are marked at edge routers and receive
  differentiated services accordingly at each core
  routers (per-hop behaviour)
• Expatiated forwarding provide low delay, low
  loss rate, and an assured bandwidth
• Assured forwarding provide guarantee throughput
  (reliable service)
• RED with In and Out RIO
• Amount of state information is proportional to
  the number of classes rather than number of flows
  ? improve scalability in comparison with IntServ
• However, does not guarantee the services on
  end-to-end basis

6
QoS Model Multiprotocol Label Switching

• MPLS a forwarding mechanism, evolution of
  Ciscos Tag Switching
• Packets are assigned a label at the ingress of an
  MPLS-capable domain
• Subsequent classification, forwarding, and
  services for the packet are based on the labels
• Placed between L2 L3, and can be triggered
  either by control traffic or by data traffic
• Faster packet classification and forwarding than
  process of parsing routing table (search for
  longest match prefix)
• Efficient tunnelling mechanisms

7
QoS Model Traffic Engineering

• Indeed, IntServ and DiffServ show performance
  degradation as the traffic load increases
• Note, QoS is not relevant when traffic load is
  light!
• Heavy traffic load produces network congestion
• Lack of resources
• Unbalanced distribution of traffic ? traffic
  engineering
• Traffic Engineering is the process of arranging
  how traffic flows through the network so that the
  congestion caused by uneven network utilization
  is avoided
• Load balancing issues
• Constraint-based routing is used for traffic
  engineering

8
QoS Model Flexible QoS Model for Manet

• FQMM combines IntServ and DiffServ
• Three types of nodes
• Ingress node source node
• Interior nodes nodes that forward packet
• Egress node destination node
• Traffic provisioning
• IntServ is used for high priority classes
• DiffServ is used for low priority classes
• Traffic conditioning
• Placed at ingress nodes
• Marking, discarding, and shaping packets
  according to traffic profile (relative link
  capacity)

9
MANET and QOS Motivation

• Mobile ad hoc networks QoS
• Limited wireless resources ? low-capacity
• Mobility ? time-varying topology
• Lack of infrastructure ? fully-distributed
• Low-capacity time-varying resources with a fully
  mobile infrastructure
• However, the network topology should be stable
  enough to allow successful propagation of all
  control traffic
• Deal with imprecise state information
• QoS class is different form priority class which
  can either be pre-emptive or non-pre-emptive

10
DiffServ IntServ FQMM

• Require accurate link state and topology
  information
• The low-capacity time-varying resources make
  maintaining accurate link state and topology
  information very difficult
• Quality of service that an application requires
  depends strictly to the quality of network

11
Tentative Definition of QoS in Manet

• To provide a set of parameters in order to adapt
  application to the quality of network while
  routing through the network
• e.g. Adaptive application, and soft QoS

12
QoS in Manet

• Parameters should represent
• available resources ? low-capacity
• stability of resources ? time-varying topology
• Protocols should be
• adaptive ? fully mobile infrastructure
  low-capacity time-varying topology and resources

13
A Cross-Layer QoS Model

• A cross-layer model to deploy QoS
• ALMs-- Application Layer Metrics
• NLMs-- Network Layer Metrics
• MLMs-- MAC Layer Metrics
• MLMs NLMs determine the quality of links to
  generate paths with good quality
• ALMs select one path which is more likely to meet
  application requirements
• Adapt to the network quality if needed

14
Parameters

• class I, delay
• class II, throughput
• class III, B-E
• power level
• buffer level
• stability level
• link SINR

At application layer
At network layer
At MAC layer
15
Mapping
16
Architecture
App. Requirement QoS Class D T B-E
Application Layer Network Layer MAC Layer
Application Layer Metrics
Network Quality Stability/ Power/buffer
Network Layer Metrics
Link Quality Link SINRs
MAC Layer Metrics
QoS Extension
Protocol Stack
17
Outline

• Mobile ad hoc networks/QoS
• Classical QoS models DiffServ IntServ
• New definition of QoS in Manet
• A cross-layer QoS model that separates MLM, NLMs,
  and ALMs

18
Conclusion

• Classical QoS models are not appropriate for
  Manet
• QoS that an application requires depends strictly
  to the quality of network
• Cross-layer QoS models respond to both network
  application requirements