Routing and Admission Control in IEEE 802.16 Distributed Mesh Networks

transcript

Routing and Admission Control in IEEE 802.16 Distributed Mesh

NetworksTzu-Chieh Tsai

Dept. of Computer ScienceNational Chengchi University

Taipei, Taiwan

Outline

Introduction

Problems

Related Work

Our Routing and CAC Algorithm

Simulations

Conclusion

Outline

IntroductionWireless Mesh Networks

IEEE 802.16 Mesh Mode

Problems

Related Work

Simulations

Conclusion

Introduction—Wireless Mesh Networks

Ad-hoc network basisSelf organization

Fault tolerance

Scalability

Lower infrastructure cost

Wider coverage

Standard activities802.11s (in progress)

802.15.5 (in progress, mainly in PHY)

802.16Mesh mode is already included in standards.

IEEE 802.16 mesh mode

Internet

IEEE 802.16 mesh mode

TerminologyMesh Base Station (MBS or BS)Mesh Subscriber Station (MSS or SS)Extended neighborhood (2-hop neighbors)

Defers from PMP mode (Point to Multi-Point)Traffic can occur directly between SSsOnly TDD is supported in MeshFrame format

Not compatible

A frame is composed withControl subframe

Network control subframeSchedule control subframe

Data subframe

IEEE 802.16 Mesh Mode Frame Formats

Frame n-1 Frame n Frame n+1

NENT NCFG NCFG … NCFG

Network Control subframe

CSCH CSCF CSCF … DSCH

Schedule Control subframe

CSCH CSCF CSCF … DSCHCSCH CSCF CSCF … DSCH

Data subframe

slot slot slot … slot slot

Data subframe

Scheduling mechanism:

CentralizedUsing MSH-CSCH, MSH-CSCF msgs.

DistributedCoordinated

Uncoordinated– Both using MSH-DSCH msgs.– Mesh Distributed Scheduling messages

In distributed coordinated mesh mode, each node periodically broadcasts :

MSH-NCFGMesh-Network Config

Exchanges the basic parameters between SSs

– ID of BS, hop count to BS, number of neighbors, …

MSH-DSCH msgs.Mesh-Distributed Scheduling

Both using Mesh election algorithm to determined next transmission time.

The information elements (IEs) of MSH-DSCH msgs.

Scheduling IE:Determines the next transmission time of MSH-DSCH msgs.

To avoid collision of MSH-DSCH msgs.

Request IE:Convey the resources over a link

Availability IE:Carry the information of the available resources

Grant IE:Convey the confirm information of the resources

IEEE 802.16 Mesh Mode—3-way Handshake

MSH-DSCH: Request

Src. sends to dest. along with MSH-DSCH: Availibilities

Indicate the empty timeslots of src. Node

MSH-DSCH: Grant

Dest. Chooses a range of empty timeslots according to MSH-DSCH:request, and,

Dest. replies with this msg.

MSH-DSCH: Grant

Src. Copies the received grant and sends it back to destination node

Requester Granter

MSH-DSCH:RequestAnd

MSH-DSCH:availbility

MSH-DSCH:Grant

Requester Granter

MSH-DSCH:RequestAnd

MSH-DSCH:Grant

IEEE 802.16 QoS Classes

Four QoS classesUnsolicited Grant Service (UGS)

Real-time Polling Service (rtPS)

Non-real-time Polling Service (nrtPS)

Best Effort (BE)

Class name Traffic type Application

UGS Real-time Constant Bit Rate (CBR) Voice over IP (VoIP)

rtPS Real-time Various Bit Rate (VBR) Real-time video

nrtPS Non-real-time Bandwidth-sensitive FTP

BE Non-real-time HTTP, Telnet

IEEE 802.16 Mesh Mode –QoS

Mesh mode uses CID (Connection ID) to define the service parameters

ReliabilityTo re-transmit or not

PriorityThe priority of the connection

Drop PrecedenceWhen congestion occurs, the likelihood of dropping the packets

Outline

Introduction

Problems

Related Work

Simulations

Conclusion

Problems

QoS provisioning for each class, we need:A Routing Method suitable in 802.16 distributed mesh mode

A way to do admission control

The above 2 things are not specified in the standard

Our solutions:SWEB (Shortest-Widest Efficient Bandwidth) metrics for routing

TAC (Token-bucket based Admission Control)

Outline

Introduction

Problems

Related Work

Simulations

Conclusion

Related Work

In [3], a token-based call admission control and a math model is proposed under IEEE 802.16 PMP mode

The bandwidth of a flow is estimated as

2fd and f

dm where,

Related Work -Token Bucket Mechanism

Token bucket parametersToken rate r

Bucket size b

In time duration t, the output volume of data would be:

, at most.

Token rate r

Packet Queue

Bucket size b

Output

Token rate r

Packet Queue

Bucket size b

Output

Related Work

In [7], routing metrics “ETX” is proposedExpected Transmission Count

Under 802.11 ad-hoc networks

Forwarding delivery ratio: ,reverse delivery ratio:

Determined by sending probe packets

ETX is calculated as:

rf ddETX

Outline

Introduction

Problems

Related Work

Our Routing and CAC AlgorithmRouting

Modified 3-way handshake

TAC algorithm

Simulations

Conclusion

Routing and CAC algorithms

Static Routing is suitable in IEEE 802.16 mesh networks:

Stations do not move or have the minimum mobility

Topology and channel conditions do not change severely

IEEE 802.16d standard does not support mobility

Providing QoS of one flow over multiple routes can be in-efficient and difficult

Routing

To minimize delay and achieve good throughputPi,j: packet error rate of link (i,j)

Ci,j: Capacity of link (i,j)

A bandwidth of a link is

Node 1 Node 2 Node 3 Node 4

2,1p 3,2p 4,3p

2,1C 3,2C 4,3C

)1( ,, jiji pC

Routing

Capacity of all links = C

Effective bandwidthPath1

=C*min((1-0.1),(1-0.2),(1-0.1))/2

=0.4*C

Path2=0.25C

Path3=0.4C

Path 3 is preferable.divided by hopCount

Path1=0.4C/3

Path3=0.4C/2

0.5 0.5

0.2 0.2

SWEB Routing

Routing is done off-line.

Path metric=

h: hop countSWEB (Shortest-Widest Efficient Bandwidth) Metrics

Node 1 Node 2 Node 3 Node 4

2,1p 3,2p 4,3p

2,1C 3,2C 4,3C

pCpCpC hhhh 1

))1(),...,1(),1(min( 1,1,3,23,22,12,1

Outline

Introduction

Problems

Related Work

TAC algorithm

Simulations

Conclusion

To shorten the call setup time, we modified the 3-way handshake

Original 3-way handshake:

Requester Granter

MSH-DSCH:RequestAnd

MSH-DSCH:Grant

Requester Granter

MSH-DSCH:RequestAnd

MSH-DSCH:Grant

Original 3-way handshake in a multi-hop environment

Outline

Introduction

Problems

Related Work

TAC algorithm

Simulations

Conclusion

Bandwidth Estimation

Assume that each flow is controlled by the token bucket mechanism.Each flow reports the parameters when it is initiated:

r: token rateb: bucket sized :Delay requirement (for real-time traffics)

Using token bucket, the required bandwidth is:

Over-estimated.

t t+7f

ri*f ri*f ri*f

t+4ft t+7f

ri*f ri*f ri*f

ri*f ri*f ri*f ri*f

t+12ft+5f

ri*f ri*f ri*f ri*f ri*f

t+9ft+6f

ri*f ri*f ri*f ri*f

t+12ft+5f

ri*f ri*f ri*f ri*f

t+12ft+5f

mi-1bi/

ri*f ri*f ri*f ri*f

t+12ft+5f

We estimate the Max. volume transmitted by a real-time flow in a frame as:

,where

TAC (Token-bucket Based Admission Control) algorithm

Goal:Guarantee delay requirements for real-time flows

Avoid starvation

To guarantee delayUse the above-mentioned bandwidth estimation.

To avoid starvationsSet minimum usage of each class:

CBR_min, VBR_min, and BE_min.

TAC algorithm

Concept:

TAC algorithm

Fields in CID (connection ID) are used to identify QoS levels

Priority

(3 bits)

Reliability

(1 bit)

Drop Precedence

(2 bits)

CBR 7 0 0

CBR_DG 4 0 1

VBR 6 0 0

VBR_DG 3 0 2

BE 5 1 0

BE_DG 2 1 3

Outline

Introduction

Problems

Related Work

SimulationsRouting

Conclusion

SimulationsParameters

ParametersFrame length = 8 ms

Slot capacity = 144 bytes

Data timeslots =165

QPSK coding rate =3/4

676 OFDM symbols per frameFor CTRL subframe=16

For DATA subframe=660

4 OFDM symbols per slot

SimulationsRouting

Topology16 nodes

4 km * 4km

Radio range = 1.5 km

Node 16 is the MBS.

SimulationsRouting

98.316.1106.491.1198.382.004.2

25456.102501.481.1132.55

86.02559.082.02525.656.177.225.6432.6581.1152.58

55.1266.1132.1191.102598.3

56.60156.109.485.60156.104.7256.177.259.04

82418.6023.182.0456.5554.5904.211.1118.1304.211.1023.1

4421.1156.12504.22523.1

04.204.204.24054.57

56.156.125.682.023.101.1

77.289.397.2411.477.242522.1102.0

77.225.677.225.604.277.267.1104.2249.6

56.5254.5977.223.11.22525.6998.0

08.1267.1177.2425.652.5871.13

4425.62577.211.1104.2098.24

125.6125.682.0252525404.204.2

23.101.423.169.0

Packet error rate over links: ( in 1/100 )

SimulationsRouting

Shortest Path

SimulationsRouting

Proposed Routing metrics

SimulationsRouting

Since we primarily focus on VBR traffics, VBR traffics are compared across 3 routing trees.

Throughput

Jitter

The number of each class traffic flow is ranging from 5 to 25.

SimulationsRouting

Throughput

5 10 15 20 25

number of flows

Shortest

SimulationsRouting

Avg. Delay

5 10 15 20 25

number of flows

Shortest

SimulationsRouting

J itter

5 10 15 20 25

Number of flows

shortest

Outline

Introduction

Problems

Related Work

SimulationsRouting

Conclusion

Simulationsparameters

Minimum usage:CBR 10 timeslots

VBR 40 timeslots

BE 75 timeslots

Token rate

Bucket size

(bits)

Delay requirement

CBR 960 64 40 ms

VBR 2000k 4000 80 ms

BE 750k 2000 --

SimulationsCAC

(wimax) throughput

20003000

40005000

60007000

5 10 15 20 25

number of flows

(TAC) throughput

010002000300040005000600070008000

5 10 15 20 25

number of flows

SimulationsCAC

avg. delay

5 10 15 20 25

number of flows

avg. delay

5 10 15 20 25

number of flows

WiMax TAC is added

SimulationsCAC

exceeds delay requirement

101214

5 10 15 20 25

number of flows

exceeds delay requirement

01234567

5 10 15 20 25

number of flows

CBR_DG

VBR_DG

WiMax TAC is added

Outline

Introduction

Problems

Related Work

Simulations

Conclusion

We proposes a CAC (called TAC) mechanism thatguarantee the delay requirements of real-time traffic flows, and

Avoid starvations

A simple routing metric SWEB that is suitable for IEEE 802.16 mesh networks

A modified 3-way handshake thatReduces call setup time

Reference[1] IEEE, “IEEE Standard for Local and metropolitan area networks Pa

rt 16: Air Interface for Fixed Broadband Wireless Access Systems”, IEEE standard, October 2004.

[2] Harish Shetiya and Vinod Sharma, "Algorithms for routing andcentralized scheduling to provide QoS in IEEE 802.16 mesh networks",Proceedings of the 1st ACM workshop on Wireless multimedianetworking and performance modeling ,WMuNeP '05. Pages: 140-149.[3] Tzu-Chieh Tsai, Chi-Hong Jiang, and Chuang-Yin Wang, "CAC andPacket Scheduling Using Token Bucket for IEEE 802.16 Networks", inJournal of Communications (JCM, ISSN 1796-2021), Volume : 1 Issue :

2, 2006. Page(s):30-37. Academy Publisher.

Reference[4] Fuqiang LIU, Zhihui ZENG, Jian TAO, Qing LI, and Zhangxi LIN,"Achieving QoS for IEEE 802.16 in Mesh Mode",8th InternationalConference on Computer Science and Informatics, Salt Lake City, USA[5] Hung-Yu Wei, Samrat Ganguly, Rauf Izmailov, and Zygmunt J. Haa

s, "Interference-Aware IEEE 802.16 WiMax Mesh Networks", inProceedings of 61st IEEE Vehicular Technology Conference (VTC 200

5 Spring).[6] Min Cao, Qian Zhang, Xiaodong Wang, and Wenwu Zhu, "Modelling

and Performance Analysis of the Distributed Scheduler in IEEE 802.16 Mesh Mode", Proceedings of the 6th ACM international symposium on Mobile ad hoc networking and computing

Reference[7] Douglas S. J. De Couto, Daniel Aguayo, John Bicket , and Robert M

orris, “A High-Throughput Path Metric for Multi-Hop Wireless Routing”, ACM MobiCom ’03.

Routing and Admission Control in IEEE 802.16 Distributed Mesh Networks

Documents