Basic functionality of QoS in multimedia Networks

1

Basic Functionality of Quality of Service
in Multimedia Networks
Piet Van Mieghem
2
Outline

• Introduction
• Multimedia
• QoS market trends
• Packet networks statistical multiplexing
• QoS What?
• Basic Network functionality required
• Conclusions

3
Outline

• Introduction
• QoS What?
• Conclusions

4
Evolution
Advances in Micro-electronics
Fast computers
High speed networks
Multimedia
best of both worlds
Data World non-real time
Telecom World real-time
5
Multimedia Marketing Trends

• Even if a minority asks for QoS, the network
  provider is forced to invest in QoS to maintain
  (or enlarge) its market share
• QoS provisioning linked to pricing will be a
  competitive weapon in future telecommunications
  networks

6
Customer Centric Evolution
QoS Complexity
network centric
CO
impact of guaranteed QoS
Time
CL
QoS granularity
end system
prioritised BE
BE
CoS
Customer fine-tuned QoS
flat fee
detailed invoice
7
Stochastic Nature of Information
correlation depth k
S(t)
x
max(S)
PCR
ES
SCR
Probxlt Slt xdx
t
R(k) ES(tk)S(t)
k
MBS
burst size
8
Statistical Multiplexing
bandwidth (Mb/s)
bandwidth (Mb/s)
4
4
3
3
2
2
1
1
0
0
120
100
80
60
40
20
0
120
100
80
60
40
20
0
time (ms)
time (ms)
No statistical multiplexing ? PCR 4 Mb/s
Statistical multiplexing min. capacity 3.4 Mb/s
9
Outline

• Introduction
• QoS What?
• Conclusions

10
QOS definition

• Application layer Grade of Service (GOS)
• hard to define precisely (perceived quality,
  human appraisal)
• alternative definition GoS is related to
  blocking/failures
• Transport layer QoS routing (end-to-end)
• path through network subject to QoS requirements
• Network layer Quality of Service (QoS)
• a measure (a ratio, a maximum, an
  average/variance/ quantile of a probability
  distribution,...) acting on a property (a loss, a
  delay, a cost, ...) of a cell/packet flow

11
QoS Associations

• QoS is related to RESOURCES in the network
• states in the network, basically CO in practice
• end systems cannot control network
  resources driver to a network centric
  view current Internet opportunity for
  network operators and network manufacturers
• Control of QoS (i.e. control of resources) is key
• resources assets money/capital
• QoS control implies OPTIMALITY, because
  suboptimal control simply means loss of
  revenue importance of Admission Control

12
Basic Network Functionality for QoS provisioning
Functionality
ATM
IP
RTP QoSR (still empty) RSVP (partly) Integrated
Services (current efforts)
real-time notion multiparameter
routing signalling traffic management admission
control input control failure control queue
control priority control
AAL1 PNNI PNNI/BISUP ATCs vendor
specific GCRA vendor specific intelligent
scheduling POB,PBS,HOL,EPD,..
13
QoS Routing
C
K
A
delay 10 ms bandwidth 100 Mb/s cell loss
ratio 1.0e-6 administrative weight 4
G

• QoS Routing Multiple parameter routing subject
  to constraints
• Link metrics are vectors
• NP-complete good heuristics needed
• Policy Routing is related to QoS routing
• Routing with partial information

14
Signalling
setup
setup
setup
call proceeding
call proceeding
setup time
Complex protocols Limited number of
connections
connect
connect
connect
connect ack
connect ack
connect ack

• Functionality that transports control information
  for installing resources along the path of each
  individual flow (or connection)
• consequence of CO-mode

15
Traffic Management

• Final (nodal) decision on installing resources
• admission function (CAC)
• CAC is inherently difficult
• Need for input control
• never trust the user
• check the contracted traffic parameters
• Congestion functionality
• Various ATM bearer capabilities (DBR, SBR, UBR,
  ABR, GFR, ...) or Internet services (guaranteed
  service, controlled load, BE)

16
Outline

• Introduction
• QoS What?
• Conclusions

17
QoS Summary

• Multimedia will evolve to QoS-awareness and
  eventually to QoS guarantees
• No smooth evolution from BE to guaranteed QoS
  provisioning QoS complexity sky-rockets due to
  guarantees
• QoS is complex and requires additional network
  functionality
• QoS is linked to revenue (money) via network
  resources QoS is associated with optimality
• Optimal QoS Control needs stochastic
  analysis Still a long way to go