Ensuring the QoS Requirements in 802.16 Scheduling

Alexander Sayenko, Olli Alanen, Juha Karhula, Timo Hämäläinen

Telecommunication laboratory, MIT DepartmentUniversity of Jyväskylä, Finland

9th ACM international symposium on Modeling analysis and simulation of wireless and mobile systems (MSWiM 2006)

Outline

•Introduction•Our Scheduling Proposal

▫Allocation of the minimum number of slots▫Allocation of unused slots▫Order of Slots

•Simulation▫Scenario 1▫Scenario 2

•Conclusion

Introduction

•WiMAX is an IEEE standard for the wireless broadband access networks

•The main advantages▫Longer Range▫More sophisticated QoS supports at MAC

layer

Introduction

•802.16d defines two basic operational modes▫Point-to-Multipoint(PMP) Mode

SSs are only allowed to communicate through the BS

▫Mesh Mode SSs can communicate to each other and to

the BS

Introduction

•WiMAX is connection-oriented▫SS must register to BS before starting to

send or receive data▫During registration, SS can negotiate the

initial QoS requirements with BS▫QoS requirements can be changed later,

and a new connection may be established on demand

Introduction

•To provide the QoS guarantees in WiMAX network▫BS has to translate the QoS requirements of

SSs into the appropriate number of slots▫BS makes a scheduling decision and informs

all SSs by UL-MAP and DL-MAP Define explicitly slots that are allocated to

each SS in both directions

Introduction

•However,▫The Scheduling Policy is not defined in the

WiMAX specifications but rather is open for alternative implementations

•Purpose▫Design a scheduling solution for the

WiMAX BS

Our scheduling proposal

•Scheduling should comprise three major stages▫Allocation of the minimum number of slots

(mandatory)▫Allocation of unused slots▫Order of Slots

Allocating the minimum number of slots•Ni: number of slots within each frame to

ensure the bandwidth requirement of connection i

FPSS

BN

i

ii

The bandwidth requirement of the ith connection (Byte/sec)

The number of bytes a connection can send in one slot (Byte/slot)

The number of frames the BS sends per one second (frame#/sec)

Allocating the minimum number of slots•Terms

▫Bimin: minimum bandwidth requirements

▫Bimax: maximum bandwidth requirements

▫Ci: ith connection class

▫Ri: request size

Allocating the minimum number of slots•UGS

▫Do not send bandwidth requests▫Cannot participate in the contention

FPSS

BNN

i

iiimaxmin

Allocating the minimum number of slots•rtPS

▫Based on the bandwidth requirements and the request size

Allocating the minimum number of slots•ertPS

▫To combine efficiency of the UGS and rtPS classes

Allocating the minimum number of slots•nrtPS

▫Can participate in the contention

Allocating the minimum number of slots•BE

▫Not have any requirements at all

Allocating free slots

•It makes sense to allocate unused slots to rtPS, nrtPS, and BE connection

•There is no need to allocate slots to UGS or ertPS connection▫It is not likely that the constant-rate

applications will increase transmission rate

Allocating free slots•Ffree: number of free slots that remains after

ensuring minimum bandwidth guarantees

•A: a set with connections for which free slots should be assigned

Order of slots•Interleaving the slots to decrease the

maximum jitter and delay values

Order of slots•Calculate the ideal distance between slots

for every connection•Take all the connections one by one and

assign slots to the proper positions•Start to place slots for UGS connections

and then for ertPS and rtPS connections•If the chosen slot is already assigned to

another connection, then the closest free slot is chosen

•The remaining slots can be filled with the slots for the nrtPS and BE connections

Order of slots

•By allocating slots in several bursts▫Size of the UL-MAP or DL-MAP messages

increases▫Fewer number of slots available for user

data

Simulationscenario 1•Purpose

▫ensure that scheduler at BS allocates resource fairly between BE connections regardless of their number

•Environment▫One BS and 15 SSs that use BE connection▫All SSs use the same modulation, 64-QAM

¾▫An SS hosts one FTP application that sends

data over TCP protocol to the wired node

Total Throughput

Simulationscenario 2

• Purpose▫ Test how scheduler at

BS allocates resources between connections that belong to different services classes

• Environment▫ One BS and 7 SSs▫ All SSs use 64QAM ¾

modulation

Simulationscenario 2

•UGS connection is always provided the sufficient number of slots

Simulationscenario 2

• ertPS connection’s throughput is determined by ON/OFF model▫ During active phase

80,000 bps▫ During silence

phase 0

• The scheduler allocates one slots so that this connection can send the bandwidth request as soon as possible

Simulationscenario 2

•rtPS connections’ throughput is determined by the variable-rate IPTV source data

•Not experience packet drops during a simulation run

Simulationscenario 2

•nrtPS connections try to send as much data as possible

•nrtPS connections’ throughput never exceeds the maximum bandwidth

Simulationscenario 2

•BE connection’s throughput is explained by the amount of free resource left▫Determined

predominantly by the variable rate of rtPS connections

Conclusion•A scheduling solution for 802.16 BS to

ensure QoS requirements of SSs in both directions▫Based conceptually on the round-robin

scheduling which is fast and simple▫Simplify the translation of the QoS

requirements into the number of slots▫Take into account parameters

minimum/maximum bandwidth requirements, class type, slot size, and the bandwidth request size

▫Account for WiMAX network parameters Frames-per-second and modulation

Conclusion•Algorithms on how to allocate free slots

and order the slots to decrease the jitter

•Further studies▫Make our scheduling more flexible in terms of

providing the delay and jitter guarantees Used as constraints at the interleaving stage Determine the maximum distance between two

consecutive slots within the frame