ELG 4183 QoS Routing Over Wireless Networks Wissam Nour 1180466 Atawo Mussa 1834251

1
ELG 4183 QoS Routing Over Wireless
Networks Wissam Nour 1180466Atawo Mussa
1834251
2
Introduction

• Research into this critical area of scheduling
  algorithms that support QoS has led to the
  implementation of Differentiated Services (DS).
• DS provides simple, easy to implement,
  low-overhead services differentiated on grounds
  of performance.
• Service level agreement is an agreement between
  customer and service provider stating traffic
  classifications.
• In DS QoS guarantees are static. Priority bit
  can be used.
• Within DS, the specification of how a packet
  should be handled (forwarding treatment) is
  called per-hop behavior (PHB).
• Packet scheduling is a major technical issue that
  needs to be resolved. Bandwidth allocation,
  multiplexing, admission control and congestion
  control can be dealt with packet scheduling
  algorithms.

3

• In the wireless network environment we encounter
  problems that are not seen in wireline networks.
  These include
• a) high error rate,
• b) bursty errors,
• c) location depended,
• d) time-varying wireless link capacity,
• e) scarce bandwidth,
• f) user mobility,
• g) and power restrictions of the mobile
  hosts.
• These problems still exist due to the
  inefficiencies of reliable scheduling algorithms.

4
Issues in Wireless Scheduling

• We must keep some facts in mind in dealing with
  wireless networks
• a) channel capacity in wireless networks are
  dynamically varying,
• b) channel errors are time and
  location-dependent and bursty in nature,
• c) mobil hosts have to compete for channel
  capacity in cases of multiple hosts,
• d) scheduling must take care of uplink (from
  mobile host to base station) and downlink flows
  (From base station to mobile hosts),
• e) mobile hosts are often constrained in
  terms of processing power and battery power.

5
Wireless Link Variability

• For not having a reliable transmission media,
  wireless channels are more error prone and suffer
  from interference, fading, and shadowing
  therefore resulting the capacity of the wireless
  link to be variable. Sometimes due to high brusty
  error state no packet can be transmitted in the
  wireless link.
• In wireless link capacity, we come across time
  and location dependent problems.
• The wireless link could be in any two states
  good (or error-free) state or bad (or error)
  state. In bad state, packet transmission on the
  wireless link would be corrupted with high
  probabilities causing no communication between
  base station and mobile host leading to
  location-dependent errors.
• Due to the mobility of the host there arises a
  need for Dynamic scheduling algorithms that could
  solve the issues of time and location problems.

6
Fairness

• Fairness issues in wireless networks is more
  complicated then in wireline networks. In
  wireline networks its the scheduling algorithm
  that prevents congestion development and
  guarantees a certain service rate to a flow.
• To assure that various flows suffer from
  congestion equally, we would like to have some
  kind of compensation technique in use. This
  compensation model should be able to resume
  transmission flow from the same state when it was
  lost.
• One way is to statically reserve bandwidth for
  compensation by creating a compensation flow and
  scheduling it in error-free service along with
  other flows.
• This could also potentially reduce the hand-off
  delay to.

7
Quality of Service (QoS)

• Broad-band wireless networks are expected to
  provide services and handle various classes of
  traffic with different QoS requirements.
• In the case of differential services, prioritized
  scheduling service for aggregated traffic with
  QoS differentiation should be implemented in the
  scheduling algorithm.
• QoS needs to be specified statistically or
  deterministically to insure physical channel
  degradation does not exceed certain thresholds.

8
Scheduling Algorithms

• a CSDPS
• Location-dependent, time-dependent and bursty
  errors can be death with channel state dependent
  packet scheduling (CSDPS).
• In CSDPS, a separate queue is maintained for each
  mobile packet that is processed on the basis of
  FIFO order. Therefore, when a channel experiences
  brusty errors, the scheduling algorithm defers
  transmission of packets on this link.
• The link status monitor (LSM) monitors the link
  state. LSM marks this affected queues, so that
  the scheduler does not serve this marked queue.
• CSDPS does not guarantee bandwidth or any
  guarantees on packet delay to the mobile users.
  No lower or upper limit is imposed on the amount
  of service received by a mobile user. A mobile
  user may receive less service then its fair share
  when a link is wrongly assumed to being in an
  error state.

9
b CSDPS with CBQ

• The unfair bandwidth sharing in CSDPS can be
  solved by using class-based queueing (CBQ) with
  CSDPS.
• CBQ consist of hierarchical channel-sharing
  guidelines for ensuring that classes receive
  their allocated share of bandwidth over a certain
  period of time.
• Users or traffic flows are grouped into classes
  each been given a certain amount of bandwidth.
• CBQ guarantees a percentage of the total
  effective throughput.
• Effective throughout is defined as the total
  successful data transmissions per unit time.
• CBQ with CSDPS allows for fair sharing of the
  channel while trying to maintain a high
  throughput.

10
(No Transcript)
11
c SBFA

• In the service-based fair approach (SBFA),
  certain amount of bandwidth is allocated to some
  compensation server(s), called long-term fairness
  server (LTFS).
• LTFS helps in compensating flows whose packet
  transmissions are deferred because of link
  errors, and shares the wireless channel with
  other regular data flows.
• The structure of SBFA is simple and provides
  throughput guarantees.
• SBFA cannot deal with channels that are getting
  excessive services than its promised share, and
  no restriction as imposed on them.
• SBFA is only designed based on the assumption
  that all flows whose wireless link is in good
  state should always be served at its promised
  service rate and not less then that amount.
• SBFA assumes fixed packet size for all flows,
  even thought they may have different packet
  sizes.

12
Working of SBFA
PQ
P12
P11
Flow 1
S1
S1
1
SQ
PQ
P22
P21
2
Flow 2
SQ
S2
S2
Scheduler
3
LTFS
PQ
S1
13
d I-CSDPS

• Improved channel state dependent packet
  scheduling (I-CSDPS) is a modified version of
  deficit round robin (DRR) scheduler working with
  an explicit compensation counter.
• With round robin (RR), each station in turn is
  given the opportunity to transmit. The station
  may decline to transmit or may transmit subjected
  to certain upper bound expressed as a maximum
  amount of data transmitted or time for this
  opportunity.
• In DRR, each flow has its own queue, each of
  these queues being served in a RR scheme.
• In I-CSDPS, flows are compensated equally unlike
  in SBFA.
• I-CSDPS also has the ability of handling varying
  packet sizes, which cannot be accomplished in
  SBFA.
• I-CSDPS cannot impose any restriction on flows
  receiving excessive service as in SBFA.

14
Wireless multiple access protocols (MAC)

• There is a growing need for the support of
  network-level quality of service and service
  differentiation in the wireless environment,
  which translates to MAC layer weighted fairness.
• Present multiple access protocols used often do
  not meet the requirements for fairness in shared
  channel wireless networks.
• Present research being done in the wireline and
  cellular networks cannot be of much assistance in
  the wireless environment due the unique
  location-dependent nature of contention in such
  networks.
• Wireless multiple access protocols typically have
  two components that work together collision
  avoidance and contention resolution.
• Research in the collision avoidance has in some
  form been able to handle this problem
  efficiently.

15

• But contention resolution in now been looked into
  by researchers, and so we will discuss it briefly
  below.
• A link layer flow is between a pair of
  neighboring nodes, and has a location-dependent
  contention for channel allocation.
• A network layer flow is between a pair of end
  hosts, and has a path-dependent contention for
  network bandwidth.
• In multiple access protocols, two types of
  mechanisms have resolved contention issue
  back-off and persistence.
• In the back-off mechanism, stations that are
  contending for channel access maintain a backoff
  counter.
• The backoff counter helps these stations to defer
  their transmission for a certain period of time.
• In the persistence mechanism, stations contend
  for clear channel with their individual
  persistence probability for transmitting or
  receiving information.

16
Algorithms for Wireless ATM MAC

• All the previously discussed algorithms are not
  integrated with the multiple-access control (MAC)
  protocol.
• To make the algorithms work with the MAC,
  modifications need to be implemented for them to
  work with wireless asynchronous transfer mode
  (ATM) networks.
• Wireline ATM services can be extended into
  wireless domain based on cell structure.
• Wireless ATM must support
• a) constant bit rate (CBR),
• b) variable bit rate (VBR),
• c) available bit rate (ABR), and
• d) unspecified bit rate (UBR) traffic
  classes.

17

• Prioritized regulated allocation delay oriented
  scheduling (PRADOS) is a more complex algorithm
  used for traffic scheduling in wireless ATM
• PRADOS is intended to work specifically with the
  MAC protocol MASCARA.
• PRADOS works on the concept of backward earliest
  due data first (B-EDF) with priority and leaky
  bucket traffic regulator.
• All connections are associated with a certain
  priority number according to which traffic class
  it belongs.
• PRADOS helps in reducing average cell delay and
  cell loss rate by keeping in mind the cell delay
  tolerance requirements.

18
Unified Wireless Fair Queueing Framework

• Fair queueing algorithms allow for fair queueing
  by swapping channel allocation between flows that
  are error prone and which are error free.
• The unified wireless fair queueing architecture
  has the ability to compare and having more than
  one algorithm working in conjunction.
• It is intended to serve as a framework to develop
  new wireless fair queuing algorithms.
• The algorithms consist of the following
• The error-free service (flows that the channel
  would have received at the same time instant if
  all channels were error free), which could be
  thought of as an ideal fair service model with no
  errors in channel.
• The lead and lag model in wireless services work
  by knowing which flows lead (flow which has
  received channel allocation in excess of its
  error free service) or lag (flow which has
  received channel allocation less then its error
  free service), and by what amount.

19

• The compensation model, in which the error free
  leading flows compensate for the lagging flows
  that are in error state, thereby addressing the
  wireless channel access issues of brusty and
  location-dependent channel error.
• Slot queues and pocket queues helps in the
  support of error sensitive and delay sensitive
  flows in a single framework and decouples
  connection-level packet management policies from
  link-level packet scheduling policies. Slots
  refer to the number of units of channel
  allocation, and packets refer to the number of
  unit of data transmission. The scheduler answers
  whom to send the next packet. Which packet to
  send next is for the flow specific decision.
  Number of slots at any instant of time equals the
  number of packets in flow queue.
• Channel monitoring and prediction provides
  reliable and correct measurement and estimation
  of channel state at any time for each backlogged
  flow.

20
Conclusion

• Therefore a good wireless packet scheduling
  algorithm should try to achieve minimize
  unproductive transmission on wireless channels by
  incorporating
• 1) High wireless channel utilization,
• 2) long-term fairness and throughput
  guarantees for flows with error free channels,
• 3) maintain few packet losses,
• 4) congestion control with error free links,
  and
• 5) able to support multiple classes of
  traffic with QoS differentiation.
•  

21
References

• Y.Cao and V. Li, Scheduling Algorithms in
  Broad-Band Wireless Networks, Proceedings of the
  IEEE, Vol. 89, No. 1, January 2001.
• S. Lu, V. Bharghavan, and R. Srikant, Fair
  scheduling in wireless packet networks, IEEE/ACM
  Trans. Networking, vol. 7, no. 4, pp. 473-489,
  1999.
• T. Nandagopal, T. Kim, X. Gao and V. Bharghavan,
  Achieving MAC Layer Fairness in Wireless Packet
  Networks, MOBICOM00, 2000.
• P. Sinha, N. Venkitaraman, R. Sivakumar and V.
  Bharghavan, WTCP A Reliable Transport Protocol
  for Wireless Wide-Area Networks, Proceedings of
  ACM MOBILCOM99, Seattle, Washington, USA, August
  1999.
• T. Nandagopal, S. Lu, V. Bharghavan, A unified
  Architecture for the Design and Evaluation of
  Wireless Fair Queueing Algorithms , 1999.
• J. Chen, A. McAuley and al. QoS Architecture
  Based on Differentiated Services for Next
  Generation Wireless IP Networks, IEFT Draft,
  2000.

22
Questions

• Q1) Name three problems seen in wireless network
  that are not seen in wireline networks
• A1) a) high error rate,
• c) location depended,
• d) time-varying wireless link capacity,
• Q2) State the two components that work together
  in wireless multiple access protocols (MAC)
• A2) collision avoidance and contention resolution

23

• Q3) In multiple access protocols, two types of
  mechanisms have resolved contention issue what
  are they?
• A3) back-off and persistence.
• Q4) Name some services that wireless ATM must
  support.
• A4) a) constant bit rate (CBR),
• b) variable bit rate (VBR),
• c) available bit rate (ABR), and
• d) unspecified bit rate (UBR) traffic
  classes.
• Q5) In the service-based fair approach (SBFA),
  certain amount of bandwidth is allocated to some
  compensation server(s), called long-term fairness
  server (LTFS). What is the purpose of LTFS?
• A5) LTFS helps in compensating flows whose
  packet transmissions are deferred because of link
  errors, and shares the wireless channel with
  other regular data flows.