Creditbased slot allocation for multimedia mobile ad hoc networks

1
Credit-based slot allocation for multimedia
mobile ad hoc networks

• Authors Hsi-Lu Chao and Wanjiun Liao
• Source IEEE Journal on Selected Areas in
  Communications, vol. 21, Issue 10, pp.
  1642-1651, Dec. 2003
• SCI 8.9 (5/56, TELECOMMUNICATIONS)
• Reporter Chao-Yu Kuo (???)
• Date 2006/03/31

2
Abstract

• The goal is
• to satisfy the minimum bandwidth requirements of
  guaranteed flows
• to provide a fair share of residual bandwidth to
  all flows

3
Abstract

• Using two-tier slot allocation called
  credit-based slot allocation protocol (CSAP)
• The first tier slot allocation
• The scheduler assigns the next time slot to the
  node
• The second tier slot allocation
• The node assigns the time slot to the flow

4
Outline

• Introduction
• Credit-based Slot Allocation Protocol (CSAP)
• Simulations
• Conclusions
• Comments

5
Introduction

• An ad hoc network is a self-organizing wireless
  network
• To comprise only of mobile nodes
• Each node plays both roles of a terminal and a
  router

6
Introduction

• Fair Scheduling
• To allow all participating flows share resource
  fairly
• To support quality-of-service (QoS) for
  multimedia traffic

7
Introduction

• Providing QoS in ad hoc networks is a challenge
• A fully distributed scheduling mechanism
• It may cause serious collisions and degrade
  network throughput

8
Introduction

• Existing work focuses mainly on QoS routing
• QoS routing alone cannot guarantee QoS
  requirements of multimedia traffic

9
Introduction

• Existing scheduling disciplines for best effort
  ad hoc networks can be classified into two
  categories
• Timestamp-based and credit-based

10
Introduction

• Timestamp-based

Sfj the start tag for the jth packet of flow
f Ffj the finish tag for the jth packet of flow
f pfj the jth packet of flow f lfj the length
of pfj A(pfj) the arrival time of pfj vA(pfj)
the virtual arrival time of pfj wf the flow
weight of flow f
11
Introduction

• The problem of timestamp-based
• Packets in the queue are required to sort based
  on their service tags
• Their virtual clock cannot be reset unless the
  queue has become empty

12
Introduction

• Credit-based
• Using accumulated credit values for fair
  scheduling
• It can achieve the same performance as
  timestamp-based schemes without suffering their
  problems

13
Introduction

• Credit-based Slot Allocation Protocol (CSAP)
• To support QoS requirements for guaranteed flows
• Ensures fair share of residual bandwidth for best
  effort flows
• A cluster-based mechanism to achieve spatial
  channel reuse

14
Introduction

• Credit-based Slot Allocation Protocol (CSAP)
• A tow-tier hierarchy to allocate time slots

15
Credit-based Slot Allocation Protocol (CSAP)

• Assumptions
• In this paper, we only consider the error caused
  by collisions
• A time-division multiple-access (TDMA)-based
  system
• TDMA is overlaid on top of a CDMA system

16
Credit-based Slot Allocation Protocol (CSAP)

• All mobile nodes are logically divided into
  several clusters

17
Credit-based Slot Allocation Protocol (CSAP)

• Each node on any flow path maintains a table
  called Flow Allocation Table (FAT) for flow
  scheduling
• The scheduler nodes need to maintain an extra
  table called Node Allocation Table (NAT) for node
  scheduling

Seven fields a node ID, a service type, a Resv,
a Num, a Credit, a Usage
and an Excess
Eight fields a node ID, a service type, a
scheduler ID, a Resv, a
Credit, a Usage and an Excess, a Q-size
18
Credit-based Slot Allocation Protocol (CSAP)
G Guaranteed request B Best effort request U
Usage C Credit

• Scheduling Parameters
• Initial value

19
Credit-based Slot Allocation Protocol (CSAP)
G Guaranteed request B Best effort request U
Usage C Credit

• Scheduling Parameters
• When a time slot is allocated

1 It allocate a time slot
20
Credit-based Slot Allocation Protocol (CSAP)
G Guaranteed request B Best effort request U
Usage C Credit

• Scheduling Parameters
• When Credit value counts down to zero

21
Credit-based Slot Allocation Protocol (CSAP)

• Scheduling Parameters
• Q-size

NP number of packets
S
R
Q-sizeNP
Init
Q-size0
Q-size0
Q-size0
Q-size2
Q-size0
Step 1
Q-sizeNP-2
Q-size0
. . .
22
Credit-based Slot Allocation Protocol (CSAP)

• In NAT
• The scheduler assigns the next time slot to the
  node with the smallest Excess value

23
Credit-based Slot Allocation Protocol (CSAP)

• In FAT
• A node cannot schedule any flow packets unless it
  has been assigned a time slot
• It assigns the slot to a flow with nonzero
  Q-size, the smallest Excess value and service type

24
Credit-based Slot Allocation Protocol (CSAP)

• Example

Flow information
Network topology
25
Credit-based Slot Allocation Protocol (CSAP)
(e) FAT updates at node 3
26
Simulations

• Environment
• There are 20 mobile nodes randomly distributed in
  a 670-m 670-m area
• The transmission range of each node is 250 m
• We randomly select nodes, some as flow sources
  and some as flow destinations

27
Simulations

• Environment
• Each flow may be a best effort or guaranteed flow
• The duration of each slot is set to 577 µs and
  each frame is of eight slots
• The node with the smallest node ID becomes the
  scheduler

28
Simulations Multihop Flows Without Mobility
Support
Network topology
29
Simulations Multihop Flows Without Mobility
Support
Flow information
30
Simulations Multihop Flows Without Mobility
Support
31
Simulations Multihop Flows Without Mobility
Support
32
Simulations Multihop Flows Without Mobility
Support
33
Simulations Multihop Flows With Mobility Support

• Environment
• The mobility pattern of each node follows the
  random waypoint model
• Each node randomly selects a target position and
  speed to move
• The speed is within (0 and Max) m/s
• After arriving target position, the mobile node
  stay in that position for a predefined period of
  time

34
Simulations Multihop Flows With Mobility Support
Flow information
35
Simulations Multihop Flows With Mobility Support
36
Simulations Multihop Flows With Mobility Support
37
Simulations Multihop Flows With Mobility Support
38
Simulations Multihop Flows With Mobility Support
39
Simulations Multihop Flows With Mobility Support
40
Conclusions

• The simulation results show CSAP improves system
  throughput
• It ensures global fairness among all flows while
  satisfying the requirements of guaranteed flows

41
Comments

• Fairness is not good
• A fine mechanism calculates Excess value

42
Question

• How to calculate Excess value in FAT?
• A) Usage Credit
• B) Credit Usage
• C) Excess
• D) none

43
Question

• Is not it a characteristic of CSAP?
• A) a cluster-based mechanism
• B) a credit-based scheduling disciplines
• C) a two-tier hierarchy architecture
• D) Using the service tag to schedule packets