A Study of WiMax QoS Mechanisms

A Study of WiMax QoS Mechanisms
A Study of WiMax QoS Mechanisms
RMOB Project
By:
Masood KHOSROSHAHY
Vivien NGUYEN
April 2006
Project supervisor:
Prof. Philippe Godlewski
Table of Contents
I. INTRODUCTION ……………………………………………………………………………………………...3
II. MEDIUM ACCESS CONTROL LAYER …..……………………………………………………………….3
III. PHYSICAL LAYER ………….……………………………………………………………………………...6
3.1 OFDM Physical Layer …………..………………………………………………………………………….6
3.2 OFDMA Physical Layer……………………………………………………………………………………. 7
3.3 Issues concerning the resource allocation methods………………………………………………………… 7
IV. QUALITY OF SERVICE ARCHITECTURES ……………………………………………………………8
4.1 MAC LAYER QoS ARCHITECTURES………………………………………………………………. 10
A. “A QoS Architecture for the MAC Protocol of IEEE 802.16 BWA System”………………………………… 10
B. “Quality of Service Support in IEEE 802.16 Networks”………………………………………………………... 11
C. “Providing integrated QoS control for IEEE 802.16 broadband wireless access systems”……………… 13
D. Brief introduction of five other studies……………………………………………………………………………... 16
D.1. “A Quality of Service Architecture for IEEE 802.16 Standards”………………………………………… 16
D.2. “Quality of service scheduling in cable and broadband wireless access systems”……………………… 16
D.3. “Exploiting MAC flexibility in WiMAX for media streaming”…………………………………………...... 16
D.4. “Algorithms for routing and centralized scheduling to provide QoS in IEEE 802.16 mesh networks”..17
D.5. “Modeling and performance analysis of the distributed scheduler in IEEE 802.16 mesh mode”……….17
4.2 PHYSICAL LAYER QoS ARCHITECTURES…………………………………………………………18
A.
Proposed solutions in “A Low Complexity Algorithm for Proportional Resource Allocation
in OFDMA Systems”………………………………………………………………………………………………18
B. Proposed solutions in “QoS Aware Adaptive Resource Allocation Techniques for Fair Scheduling in
OFDMA Based Broadband Wireless Access Systems”………………………………………………………… 19
Physical Layer QoS Architectures Summary…………………………………………………………………………. 21
REFERENCES…………………………………………………………………………………………………...22
A Study of WiMax QoS Mechanisms
Abstract: In this study, we first give a brief overview of the WiMax/IEEE 802.16 technology covering
almost all aspects which are discussed in the standard. Next, as the main focus of the study, we introduce
the QoS mechanisms that have been proposed in the past few years and are available in the open
literature. This study, being the first of its kind, tries to enlighten the reader regarding the mechanisms
available, so that they can compare the pros and cons of each and adopt the method most suitable to any
given case. The QoS mechanisms have been categorized based on the layer, MAC or PHY, in which they
propose their architectures.
Keywords: WiMax, IEEE 802.16, QoS, Scheduling, MAC layer, Physical layer
I.
INTRODUCTION
The standards for Broadband Wireless Access Systems have been developed by the Institute of
Electrical and Electronics Engineers (IEEE) which insures a global and open process that enjoys
worldwide participation. So far, IEEE 802.11 (short-range:
IEEE 802:
~100 m), which is a standard for Wireless Local Area The LAN/MAN Standards Committee
Networks, often called “Wi-Fi” for “Wi-Fi Alliance” and Wired:
IEEE 802.16 (long-range: ~10 km), which is a standard for – 802.3 (Ethernet)
Wireless Metropolitan Area Networks, often called “WiMAX” – 802.17 (Resilient Packet Ring)
Wireless:
for “WiMAX Forum”, have been developed.
WiMax is considered as the “Last Mile” solution which
provides fast local connection to the network. Compared to
high-capacity cable/fiber, it is less expensive to deploy. The
WiMax standardization process has had the milestones that are
mentioned in the following table.
802.16 Projects
IEEE 802.16-2001
•MAC
•10-66 GHz PHY
802.16c (Profiles)
802.16a
2-11 GHz PHY
802.16-2004
802.16e
•Mobile Amendment
– 802.11: Wireless LAN
• Local Area Networks
– 802.15: Wireless PAN
• Personal Area Networks
– 802.16: WirelessMAN
• Metropolitan Area Networks
– 802.20:
• Vehicular Mobility
The WiMax forum has the mission of
promoting deployment of Broadband Wireless Access (BWA) by using a global
standard and certifying interoperability of products and technologies. It supports
IEEE 802.16 standard and proposes and promotes access profiles for it. The forum
certifies interoperability levels and hence achieves global acceptance.
In the table below, properties of the IEEE 802.16 standard are provided, in order
to give a quick and short overview of the range of issues considered in the
standard.
Properties of IEEE Standard 802.16
II. MEDIUM ACCESS CONTROL LAYER
• Broad bandwidth
–Up to 134 Mbit/s in 28 MHz channel (in 10-66 GHz air interface)
• Supports multiple services simultaneously with full QoS
– Efficiently transport IPv4, IPv6, ATM, Ethernet, etc.
• Bandwidth on demand (frame by frame)
•MAC designed for efficient use of spectrum
• Comprehensive, modern, and extensible security
• Supports multiple frequency allocations from 2-66 GHz
–OFDM and OFDMA for non-line-of-sight applications
• TDD and FDD
• Link adaptation: Adaptive modulation and coding
– Subscriber by subscriber, burst by burst, uplink and downlink
• Point-to-multipoint topology, with mesh extensions
• Support for adaptive antennas and space-time coding
• Extensions to mobility
The IEEE 802.16 MAC is a scheduling
one where the subscriber station (SS) that
wants to attach to the network, has to
compete once when it initially enters the
network. A time slot allocation is made by
the base station (BS) which can be enlarged
and constricted. It remains assigned to the
subscriber station meaning that other
stations are not supposed to use the same
resources. This scheduling algorithm has its
advantages since it remains stable under
overload and oversubscription, has more
Adapted from [Mar-04]
bandwidth efficiency and also allows the
base station to control QoS, meaning that it is balancing the resources among the needs of the
subscriber stations.
3
802.16 MAC: Overview
The main goal of the MAC layer is to
manage the resources of the air interface • Point-to-Multipoint
efficiently. Indeed, access and bandwidth • Metropolitan Area Network
• Connection-oriented
allocation algorithms must serve hundreds of • Supports difficult user environments
terminals per channel. Those terminals can – High bandwidth, hundreds of users per channel
eventually be shared by multiple end users. To – Continuous and burst traffic
support a large variety of services such as voice, – Very efficient use of spectrum
data or internet connection, the 802.16 MAC • Protocol-Independent core (ATM, IP, Ethernet, …)
Balances between stability of contentionless and
must accommodate both continuous and bursty •efficiency
of contention-based operation
traffic. The issues concerning the transport • Flexible QoS offerings
efficiency are also addressed at the interface – CBR, rt-VBR, nrt-VBR, BE, with granularity within classes
between the MAC and the PHY layers. The • Supports multiple 802.16 PHYs
modulation and coding schemes are specified in
Adapted from [Mar-04]
a burst profile adjusted in function of each burst
sent to each SS. The MAC can make use of bandwidth-efficient burst profiles under favorable link
conditions and shift to more reliable and robust ones if the opposite is the case even though the
spectral efficiency will be lower. [EMSW-02]
The MAC includes service-specific convergence sublayers (ATM and Packet) that interface to
layers above. At the core, there is the MAC common part sublayer that carries out the key MAC
functions. Below the common part sublayer is the privacy sublayer. Extensive bandwidth allocation
and QoS mechanisms are provided, but the details of scheduling and reservation management have not
been specified in the standard. The functions of the common part sublayer will be discussed more in
the following paragraphs. [EMSW-02]
On the downlink, data to Subscriber Stations (SSs) are multiplexed in TDM fashion. The uplink is
shared between SSs in TDMA fashion. The 802.16 MAC is connection-oriented, according to which,
all services, including connectionless services, are mapped to a connection. This provides a
mechanism for requesting bandwidth, associating QoS and traffic parameters, transporting and routing
data to the appropriate convergence sublayer, and all other actions associated with the contractual
terms of the service. Connections are referenced with 16-bit connection identifiers (CIDs) and may
require continuously granted bandwidth or bandwidth on demand. Upon entering the network, the SS
is assigned three management connections in each direction. These three connections reflect the three
different QoS requirements used by different management levels. The first of these is the basic
connection, which is used for the transfer of short, time-critical MAC and radio link control (RLC)
messages. The primary management connection is used to transfer longer, more delay-tolerant
messages such as those used for authentication and connection setup. The secondary management
connection is used for the transfer of standards-based management messages such as Dynamic Host
Configuration Protocol (DHCP), Trivial File Transfer Protocol (TFTP), and Simple Network
Management Protocol (SNMP). In addition to these management connections, SSs are allocated
transport connections for the contracted services. Transport connections are unidirectional to facilitate
different uplink and downlink QoS and traffic parameters. [EMSW-02]
The MAC builds the downlink subframe starting with a frame control section containing the DLMAP and UL-MAP messages. These indicate PHY transitions on the downlink as well as bandwidth
allocations and burst profiles on the uplink. The advanced technology of the 802.16 PHY requires
equally advanced radio link control (RLC), particularly the capability of the PHY to transition from
one burst profile to another. The RLC must control this capability as well as the traditional RLC
functions of power control and ranging. [EMSW-02]
Burst profiles for the downlink are each tagged with a Downlink Interval Usage Code (DIUC).
Those for the uplink are each tagged with an Uplink Interval Usage Code (UIUC). Burst profile
determines the modulation and FEC and is dynamically assigned according to link conditions. It is
determined burst by burst and per subscriber station. There is always a trade-off between capacity and
robustness in real time. Their utilization has roughly doubled capacity for the same cell area. Burst
profile for downlink broadcast channel is well-known and robust, but other burst profiles can be
configured “on the fly”. SS capabilities are recognized at registration. [EMSW-02] [Mar-04]
4
Point-to-Multipoint mode
Mesh Mode (Subscriber-to-Subscriber communications)
Each connection in the uplink direction is mapped to a scheduling service. Each scheduling service
is associated with a set of rules imposed on the BS scheduler responsible for allocating the uplink
capacity and the request-grant protocol between the SS and the BS. The detailed specification of the
rules and the scheduling service used for a particular uplink connection is negotiated at connection
setup time. The uplink scheduling types include Unsolicited Grant Service (UGS), for real-time flows
or periodic fixed size packets (e.g. VoIPor ATM CBR), Real-Time Polling Service (rtPS), for realtime service flows or periodic variable size data packets (e.g. MPEG), Non-Real-Time Polling Service
(nrtPS), for non real-time service flows with regular variable size bursts (e.g. FTPor ATM GFR) and
Best Effort (BE), for best effort traffic(e.g. UDPor ATM UBR). [EMSW-02] [Mar-04]
The IEEE 802.16 MAC accommodates two classes of SS, differentiated by their ability to accept
bandwidth grants simply for a connection or for the SS as a whole. Both classes of SS request
bandwidth per connection to allow the BS uplink scheduling algorithm to properly consider QoS when
allocating bandwidth. In Bandwidth Grant per Subscriber Station (GPSS), base station grants
bandwidth to the subscriber station (SS), and SS in turn may re-distribute bandwidth among its
connections, maintaining QoS and service-level agreements. This method is suitable for many
connections per terminal by off-loading base station’s work. It also allows more sophisticated reaction
to QoS needs. It has low overhead but requires intelligent subscriber station. The GPSS is mandatory
for P802.16 10-66 GHz PHY. In contrast, in Bandwidth Grant per Connection (GPC) mode, base
station grants bandwidth to a connection. This method is mostly suitable for few users per subscriber
station. It has higher overhead, but allows simpler subscriber station. [EMSW-02] [MEK-01]
There are several methods for bandwidth request: Implicit requests (UGS), in which there is no
actual message and the bandwidth is negotiated at connection setup; BW request messages, which uses
the special BW request header and it can request up to 32 KB with a single message; Piggybacked
request (for non-UGS services only), which can be up to 32 KB per request for the CID and Poll-Me
bit (for UGS services only), which is used by the SS to request a bandwidth poll for non-UGS
services. [MEK-01]
Maintaining QoS in GPSS follows a semi-distributed approach. In which, the BS sees the requests
for each connection; based on this, grants bandwidth (BW) to the SSs (maintaining QoS and fairness).
On the other side, the SS scheduler maintains QoS among its connections and is responsible to share
the BW among the connections (maintaining QoS and fairness). It’s worth mentioning that algorithms
in BS and SS can be very different; SS may use BW in a way unforeseen by the BS. [MEK-01]
Subscriber Station (SS) initialization has several steps: At first, the SS scans for downlink channel
and establishes synchronization with the BS. Then, it obtains transmit parameters. As next step, it
performs ranging and negotiating basic capabilities. Then it is authorized by the BS and performs key
exchange. Afterwards, it performs the registration and IP connectivity establishment. Then it’s the turn
of time of day establishment and the transfer of operational parameters. At the end, it sets up the
connections. [MEK-01]
5
III.
PHYSICAL LAYER
802.16a PHY Alternatives:
• OFDM (WirelessMAN-OFDM Air Interface)
256-point FFT with TDMA (TDD/FDD)
• OFDMA (WirelessMAN-OFDMA Air Interface)
2048-point FFT with OFDMA (TDD/FDD)
• Single-Carrier (WirelessMAN-SCa Air Interface)
TDMA (TDD/FDD)
BPSK, QPSK, 4-QAM, 16-QAM, 64-QAM, 256QAM
[Mar-04]
Adapted from [Mar-04]
In this section, we introduce the techniques that are used in the physical layer: Orthogonal
Frequency Division Multiplexing (OFDM) and Orthogonal Frequency Division Multiple Access
(OFDMA).
Those techniques have been developed for the last few years to deliver broad band services that
can be compared to those of wired services in terms of data rates. The main issue addressed for the
PHY layer is to allocate the resources efficiently by assigning a set of subcarriers and by determining
the number of bits to be transmitted for each subcarrier in an OFDMA system. An optimal algorithm
has to be chosen to obtain a certain level of performance by considering some constraints such as
delays, the total number of connected SS and the total power. [ECV-03]
OFDM will be first studied in order to understand OFDMA used to share the physical resources
among each SS.
3.1
OFDM Physical Layer
Definition and advantages of OFDM
Orthogonal frequency-division multiplexing (OFDM) is a transmission technique that is based on
the same idea as frequency-division multiplexing (FDM). In FDM, multiple signals are sent out at the
same time, but on different frequencies. It actually divides a broadband channel into many narrowband
subchannels. In OFDM, a single transmitter transmits on several different orthogonal frequencies. This
technique, associated with the use of advanced modulation techniques on each component, give a
transmitted signal with high resistance to multi-path interference and a much higher spectral efficiency
is obtained.
As the chose of the manufacturers, the WiMAN OFDM PHY layer is the most commonly used
because of the reasons previously quoted. It was also selected, rather than other techniques such as
single-carrier (SC) or CDMA, due to its superior non line-of-sight (NLOS) performance. This
multiplexing technique allows important equalizer design simplification to support operations in
multipath propagation environments and overcome channel fading quite efficiently (Rayleigh channel
model). [Intel-04]
Subchannelization
The OFDM PHY layer supports UpLink (UL) subchannelization, with the number of subchannels
being 16. This feature is particularly useful when a power-limited platform such as a laptop is
considered in the subscriber station in an indoor environment. With a sub-channelization factor of
1/16, a 12-dB link budget enhancement can be achieved. Sixteen sets of 12 subcarriers each, are
defined, where one, two, four, eight or all of the sets can be assigned to a subscriber station in the
uplink. Eight pilot carriers are used when more than one set of sub-channels are allocated. [Intel-04]
Multiplexed channel
This multiplexing technique supports Time Division Duplexing (TDD), in which the uplink and
downlink share a channel but do not transmit simultaneously, and Frequency Division Duplexing
6
(FDD), in which the uplink and downlink operate on separate channels, possibly simultaneously, with
support for FDD and also Half-Duplex FDD (H-FDD). In both TDD and FDD modes, the length of
the frame can vary (under the control of the BS scheduler) per frame. In TDD mode, the division point
between uplink and downlink can also vary per frame, allowing asymmetric allocation of an air time
between uplink and downlink if required. The point-to-multipoint architecture forces the BS to
transmit TDM signals in which time slots are allocated serially for each individual SS’s. [Intel-04]
Error correcting codes
The specification defines as a requirement for the error correcting code, a combined variable-rate
Read-Solomon (RS) and Convolutional Coding (CC) scheme, supporting code rates of 1/2, 2/3, 3/4,
and 5/6, although variable-rate Block Turbo Code (BTC) and Convolutional Turbo Code (CTC) are
optional. Also, the standard supports multiple types of modulation, including Binary Phase Shift
Keying (BPSK), Quadrature Phase Shift Keying (QPSK), 16-Quadrature Amplitude Modulation
(QAM) and 64-QAM. [Intel-04]
Diversity
Finally, the PHY layer supports optionally the diversity transmission technique in which the
information-carrying signal is transmitted along different propagation paths. It uses multiple
transmitting and receiving antennas. In the Downlink (DL), it is using Space Time Coding (STC) and
Adaptive Antenna Systems (AAS) with Spatial Division Multiple Access (SDMA). [Intel-04]
3.2
OFDMA Physical Layer
Multiple access
OFDMA is referred as Multiuser-OFDM as it uses OFDM as multiple access method especially
for the 4G wireless networks.
Subchannelization
The OFDMA PHY layer is based on OFDM and supports subchannelization in both the UL and
(DownLink) DL. Five different sub-channelization schemes in total are supported. The OFDMA PHY
layer supports both TDD and FDD operations. CC is the required coding scheme by the specification
and the code rates are the same as the ones supported by the OFDM PHY layer. BTC and CTC coding
schemes are optionally supported. [Intel-04]
Multiplexing and diversity
The same signal modulations are supported as in OFDM. STC and AAS with SDMA are
supported, as well as Multiple Input, Multiple Output (MIMO). MIMO regroup a certain number of
techniques for using multiple antennas at both the BS and SS in order to increase the channel capacity
and to decrease the Bit Error Rate (BER). [Intel-04]
To transmit information at a high data rate, the BWA has to provide efficient and flexible resource
allocation. It has been shown that frequency hoping and adaptive modulation used in subcarrier
allocation permit to have higher performances. The frequency hoping technique allows to compensate
the negative effect of channel fading. [ECV-03]
But ODMA was chosen as it is a promising multiple access scheme having the same advantages as
OFDM, that is to say, inter symbol interference immunity as well as frequency selective fading
immunity while having a higher spectral efficiency.
3.3
Issues concerning the resource allocation methods
In OFDM, only a single user can transmit on the entire spectrum at a given time. Time division or
frequency division multiple access is therefore needed. This scheme is not optimum. That is why
OFDMA is preferred because of the fact that is allows multiple users to transmit simultaneously by
sharing the sub-channels among the users. [WSEA-04]
The two studied papers have different approaches on the resource allocation methods for OFDMA
systems:
7
IV.
•
“The problem of assigning subcarriers and power to the different users in an OFDMA system
has recently been an area of active research. In [WCLM-99], the margin-adaptive resource
allocation problem was tackled, wherein an iterative subcarrier and power allocation algorithm
was proposed to minimize the total transmit power given a set of fixed user data rates and bit
error rate (BER) requirements. In [JL-03], the rate-adaptive problem was investigated, wherein
the objective was to maximize the total data rate over all users subject to power and BER
constraints. It was shown in [JL-03] that in order to maximize the total capacity, each
subcarrier should be allocated to the user with the best gain on it, and the power should be
allocated using the waterfilling algorithm across the subcarriers. However, no fairness among
the users was considered in [JL-03]. This problem was partially addressed in [YL-00] by
ensuring that each user would be able to transmit at a minimum rate, and also in [RC-00] by
incorporating a notion of fairness in the resource allocation through maximizing the minimum
user’s data rate. In [SAE-03], the fairness was extended to incorporate varying priorities.
Instead of maximizing the minimum user’s capacity, the total capacity was maximized subject
to user rate proportionality constraints. This is very useful for service level differentiation,
which allows for flexible billing mechanisms for different classes of users. However, the
algorithm proposed in [SAE-03] involves solving non-linear equations, which requires
computationally expensive iterative operations and is thus not suitable for a cost-effective realtime implementation.” [WSEA-04]
•
“In multiuser environment, a good resource allocation scheme leverages multiuser diversity
and channel fading [VTL-02]. It was shown in [KH-95] that the optimal solution is to schedule
the user with the best channel at each time. Although in this case, the entire bandwidth is used
by the scheduled user, this idea can also be applied to OFDMA system, where the channel is
shared by the users, each owing a mutually disjoint set of subcarriers, by scheduling the
subcarrier to a user with the best channel among others. Of course, the procedure is not simple
since the best subcarrier of the user may also be the best subcarrier of another user who may
not have any other good subcarriers. The overall strategy is to use the peaks of the channel
resulting from channel fading. Unlike in the traditional view where the channel fading is
considered to be an impairment, here it acts as a channel randomizer and increases multiuser
diversity [VTL-02]. The resource allocation problem has been recently considered in many
studies. Almost all of them define the problem as a real time resource allocation problem in
which Quality of Service (QoS) requirements are fixed by the application. QoS requirement is
defined as achieving a specified data transmission rate and bit error rate (BER) of each user in
each transmission. In this regard, the problem differs from the water-pouring schemes wherein
the aim is to achieve Shannon capacity under the power constraint [WCLM-99].” [ECV-03]
QUALITY OF SERVICE ARCHITECTURES
Although the IEEE 802.16 standard is already a very mature and a sophisticated one, nevertheless,
researchers around the world have not stopped proposing amendments to the standard. For example, in
[GWAC-05], the authors have suggested the utilization of spatial multiplexing and multi-user OFDM to
maximize the achieved throughput. They have also suggested the use of interference cancellation of
dominant interferers and hybrid ARQ in order to increase the range and robustness of the system.
The various protocol mechanisms described in the standards, [Std-04] and [Std-05], may be used to
support QoS for both uplink and downlink through the SS and the BS. The requirements for QoS
include the following:
a) A configuration and registration function for pre-configuring SS-based QoS service flows and
traffic parameters.
b) A signaling function for dynamically establishing QoS-enabled service flows and traffic
parameters.
c) Utilization of MAC scheduling and QoS traffic parameters for uplink service flows.
d) Utilization of QoS traffic parameters for downlink service flows.
e) Grouping of service flow properties into named Service Classes, so upper-layer entities and external
8
applications (at both the SS and BS) may request service flows with desired QoS parameters in a
globally consistent way.
The principal mechanism for providing QoS is to associate packets traversing the MAC interface
into a service flow as identified by the transport CID. A service flow is a unidirectional flow of
packets that is provided a particular QoS. The SS and BS provide this QoS according to the QoS
Parameter Set defined for the service flow. The primary purpose of the QoS features defined in the
standards is to define transmission ordering and scheduling on the air interface. However, these
features often need to work in conjunction with mechanisms beyond the air interface in order to
provide end-to-end QoS or to police the behavior of SSs. Service flows exist in both the uplink and
downlink direction and may exist without actually being activated to carry traffic. All service flows
have a 32-bit SFID; admitted and active service flows also have a 16-bit CID.
A service flow is a MAC transport service that provides unidirectional transport of packets either
to uplink packets transmitted by the SS or to downlink packets transmitted by the BS. A service flow
is characterized by a set of QoS Parameters such as latency, jitter, and throughput assurances. In order
to standardize operation between the SS and BS, these attributes include details of how the SS requests
uplink bandwidth allocations and the expected behavior of the BS uplink scheduler. It is useful to
think of three types of service flows:
1) Provisioned: This type of service flow is known via provisioning by, for example, the network
management system. Its AdmittedQoSParamSet and ActiveQoSParamSet are both null.
2) Admitted: This type of service flow has resources reserved by the BS for its
AdmittedQoSParamSet, but these parameters are not active (i.e., its ActiveQoSParamSet is null).
Admitted Service Flows may have been provisioned or may have been signalled by some other
mechanism.
3) Active: This type of service flow has resources committed by the BS for its ActiveQoSParamSet,
(e.g., is actively sending maps containing unsolicited grants for a UGSbased service flow). Its
ActiveQoSParamSet is non-null.
The service flow is the central concept of the MAC protocol. It is uniquely identified by a 32-bit
(SFID). Service flows may be in either the uplink or downlink direction. There is a one-to-one
mapping between admitted and active service flows (32-bit SFID) and transport connections (16-bit
CID). Outgoing user data is submitted to the MAC SAP by a CS (Convergence sublayer) process for
transmission on the MAC interface. The information delivered to the MAC SAP includes the CID
identifying the transport connection across which the information is delivered. The service flow for the
connection is mapped to MAC transport connection identified by the CID. A Classifier Rule uniquely
maps a packet to its transport connection.
The Service Class serves the following purposes: It allows operators to move the burden of
configuring service flows from the provisioning server to the BS. Operators provision the SSs with the
Service Class Name; the implementation of the name is configured at the BS. This allows operators to
modify the implementation of a given service to local circumstances without changing SS
provisioning. Also, it allows higher-layer protocols to create a service flow by its Service Class Name.
The Service Class Name is “expanded” to its defined set of parameters at the time the BS successfully
admits the service flow. The Service Class expansion can be contained in the following BS-originated
messages: DSA-REQ (Dynamic Service Addition), DSC-REQ (Dynamic Service Change), DSA-RSP,
and DSC-RSP.
Mobile networks require common definitions of service class names and associated
AuthorizedQoSParamSets in order to facilitate operation across a distributed topology. Global service
class names shall be supported to enable operation in this context. In operation, global service class
names are employed as a baseline convention for communicating AuthorizedQoSParamSet or
AdmittedQoSParamSet. Global service class name is similar in function to service class name except
that 1) Global service class name use may not be modified by a BS, 2) Global service class names
remain consistent among all BS, and 3) Global service class names are a rules-based naming system
whereby the global service class name itself contains referential QoS Parameter codes. In practice,
global service class names are intended to be accompanied by extending or modifying QoS Param Set
defining parameters, as needed, to provide a complete and expedited method for transferring
9
Authorized- or AdmittedQoSParamSet information. Global service flow class name parameters:
Uplink/Downlink indicator, Maximum sustained traffic rate, Traffic Indication Preference, Maximum
traffic burst, Minimum reserved traffic rate, Maximum latency, SDU indicator and Paging Preference.
In the above paragraphs, a summary of the information provided in the QoS section (6.3.14) of the
standard was given. But, as mentioned on several occasions, the details of actual QoS provisioning are
not given in the standard. In what follows, we introduce, in a short format, the QoS architectures that
have been proposed so far, which are available in the open literature. Specifically, in the MAC layer,
three QoS architectures, which seem to offer more promising results, have been selected and
described. Also, another five studies in this layer have been introduced briefly.
4.1
MAC LAYER QoS ARCHITECTURES
A . “A QoS Architecture for the MAC Protocol of IEEE 802.16 BWA System” [CWM-02]
In this research, the authors have proposed a QoS architecture based on priority scheduling (A
scheduling strategy for the schedulers) and dynamic bandwidth allocation.
The have mentioned the following points as the responsibilities of the implementers: 1) A method
for efficiently combining bandwidth request strategy and bandwidth allocation (Scheduling) strategy
to maintain QoS and fairness for different traffic. 2) A method for choosing different algorithms
considering implementation and computation complexity. 3) A method for recognizing high priority
requests during first access.
The proposed QoS architecture, the following figure, has the following modules as its building
blocks: Traffic Classifier, SS’s Upstream Scheduler, BS’s Upstream and Downstream Schedulers.
Adapted from [CWM-02]
The Contention Slot Allocator (CSA) is used by the BS to dynamically adjust the ratio of the
bandwidth allocated to the contention slots and reservation slots: too few contention slots increase the
chances of bandwidth request collision, on the other hand, too many of them reduces the bandwidth
left for data transmission. So there is a trade-off in the design of the CSA, the output of which defines
the bandwidth resources allocated to the CRA (Collision Resolution Algorithm)[GSS-99] and BS’s
upstream grants scheduler. The CRA defines the utilization of the contentions slots and the rules used
to resolve contention.
In the Upstream Scheduler implementation, GPSS (Grant Per SS) mode is preferred. Since only
the information about the overall required bandwidth is needed to inform the BS’s upstream scheduler,
10
a small amount of bandwidth is needed to update this information. Furthermore, in this way, the
resulting problems due to the time lag in receiving updated information from SSs can be resolved. For
example, even if the BS does not know that an urgent packet arrives a the SS, this packet can still
overtake other packets and thus provide tight guarantees as long as a good scheduling algorithm is
used at the SS. In this way, the BS’s upstream scheduler does not require detailed information on
every connection at the SS.
At the SS, when traffic generates, the traffic classifier guides them into different priority queues
based on traffic priority. Then the CRC (Contention Ratio Calculator) dynamically assigns different
competitive ratio parameters Ru, Rua, Rrp, Rnp and Rbe to UGS (Unsolicited Grant Service), UGSAD (UGS with Activity Detection), rtPS, nrtPS and BE queues, respectively. The traffic scheduler
allocates the granted upstream slots to different connections with different QoS requirement.
According to the information of the request of the SS, the BS’s upstream scheduler schedules
grants to the SS. Then, the SS’s upstream scheduler is responsible for the selection of appropriate
packets from the respective UGS-AD, rtPS, nrtPS and BE queues and sends them through the
upstream data slots granted by the BS’s upstream scheduler.
When implementing the SS’s upstream scheduler, one can use MPFQ(Multiclass Priority Fair
Queuing)[MLK-00] scheduler with some modification. For each service category, there is a priority
class in MPFQ. Each of the priority classes has its own packet scheduler that determines the packet
order it its class. One can use a Wireless Fair Queuing (WFQ) [LBS-97] Policy in the higher priorities
in order to provide a lower delay bound for this service class, Weighted Round Robin (WRR) [PGa-93]
scheduling for the middle priorities since the delay requirements are not that tight for these classes and
WRR is simple, and FIFO scheduling at the lower priorities.
When implementing the BS’s upstream scheduler, one finds that WRR is more suitable because
the exact arrival time of a packet is not involved in the computing the virtual finishing time. This
actually is the real situation where BS has limited information on the traffic generated at SS, and hence
computing the time for transmission and bandwidth allocation just based on the bandwidth requests is
more appropriate.
The practical issues such as performance and stability associated with the QoS architecture and
also comparison of the different implementation approaches for each block in the architecture,
including different scheduling algorithms, are left for future studies.
B. “Quality of Service Support in IEEE 802.16 Networks” [CLME-06]
In this article, the authors have reviewed and analyzed the mechanisms for supporting QoS at the
IEEE 802.16 MAC layer. Then, they have analyzed by simulation the performance of IEEE 802.16 in
two application scenarios, which consist of providing last-mile Internet access for residential and SME
(Small and Medium-sized Enterprises) subscribers, respectively. Their analysis is aimed at showing
the effectiveness of the 802.16 MAC protocol in providing differentiated services to applications with
different QoS requirements, such as VoIP and Web.
11
Adapted from [CLME-06]
The previous figure shows the blueprint of the functional entities for QoS support, which logically
reside within the MAC layer of the BS and SSs. Each downlink connection has a packet queue at the
BS (represented with solid lines). In accordance with the set of QoS parameters and the status of the
queues, the BS downlink scheduler selects from the downlink queues, on a frame basis, the next
service data units (SDUs) to be transmitted to SSs. On the other hand, uplink connection queues
(represented in the figure by solid lines) reside at SSs. Since the BS controls the access to the medium
in the uplink direction, bandwidth is granted to SSs on demand. Bandwidth requests are used on the
BS for estimating the residual backlog of uplink connections. In fact, based on the amount of
bandwidth requested (and granted) so far, the BS uplink scheduler estimates the residual backlog at
each uplink connection (represented in the figure as a virtual queue by dashed lines), and allocates
future uplink grants according to the respective set of QoS parameters and the virtual status of the
queues. However, although bandwidth requests are per connection, the BS nevertheless grants uplink
capacity to each SS as a whole. Thus, when an SS receives an uplink grant, it cannot deduce from the
grant which of its connections it was intended for by the BS. Consequently, an SS scheduler must also
be implemented within each SS MAC, in order to redistribute the granted capacity to all of its own
connections.
Here, the performance of 802.16 in two of the most promising application scenarios envisaged by
the WiMAX forum is assessed. They consist in providing last-mile Internet access for residential and
SME subscribers. Since a minimum reserved rate is the basic QoS parameter negotiated by a
connection within a scheduling service, the class of latency-rate [SVa-98] scheduling algorithms is
particularly suited for implementing the schedulers in the 802.16 MAC. Specifically, within this class,
they selected deficit round robin (DRR) [ShV-96] as the downlink scheduler to be implemented at the
BS, since it combines the ability of providing fair queuing in the presence of variable length packets
with the simplicity of implementation. In particular, DRR requires a minimum rate to be reserved for
each packet flow being scheduled. Therefore, although not required by the 802.16 standard, BE
connections should also be guaranteed a minimum rate. This fact can be exploited in order to both
avoid BE traffic starvation in overloaded scenarios, and let BE traffic take advantage of the excess
bandwidth which is not reserved for the other scheduling services. On the other hand, DRR assumes
that the size of the head-of-line packet is known at each packet queue; thus, it cannot be used by the
BS to schedule transmissions in the uplink direction. In fact, with regard to the uplink direction, the
BS is only able to estimate the overall amount of backlog of each connection, but not the size of each
backlogged packet. Therefore, they selected weighted round robin (WRR) [KSC-91] as the uplink
scheduler in their 802.16 simulator. Like DRR, WRR belongs to the class of rate-latency scheduling
algorithms. Finally, they decided to implement DRR as the SS scheduler, because the SS knows the
sizes of the head-of-line packets of its queues.
The metrics used for assessing the performance of 802.16 are the average packet-transfer delay
and the delay variation. The simulations were carried out by means of a prototypical simulator of the
IEEE 802.16 protocol. The simulator was event-driven and was developed using the C++
programming language. Specifically, the MAC layers of the SSs and the BS were implemented,
including all functions for uplink/downlink data transmission.
Residential Scenario
The Residential scenario consists of a BS providing Internet access to its subscribers, by means of
a variable number of BE connections evenly distributed among the SSs. Internet traffic is modeled as a
Web traffic source. Since the BS knows the current status of downlink queues, as soon as a downlink
packet is enqueued at the BS, it is immediately eligible for transmission to its intended SS. As long as
the system is underloaded, the connection queues are almost always empty. Thus, the average delay of
downlink packets is almost constant. However, the average delay increases sharply as soon as the
system starts to get overloaded, because the BS is not able to fully serve the backlog of downlink
connections before new packets are enqueued. The average delay of uplink traffic is higher than that
of downlink traffic. In fact, any SS has to request bandwidth to the BS, in order to receive an uplink
grant to transmit its backlog. The average time needed by an SS to request bandwidth by using
contention increases with the offered load. The maximum achievable throughput decreases when the
number of SSs increases, for both downlink and uplink curves.
12
SME Scenario
The SME scenario involves a BS providing several enterprise customer premises with three
different types of services: VoIP, videoconference, and data. It is assumed that each SS has four VoIP
sources multiplexed into an rtPS connection, two videoconference sources multiplexed into an rtPS
connection, and a BE connection loaded with data traffic. Also, it is assumed that the BS grants a
unicast poll to each VoIP and videoconference connection every 20 ms. Finally, for data traffic, the
same Web traffic source model is used as in the previous scenario.
When the system is underloaded, there is no service differentiation between connections with data
and multimedia traffic. However, when the system becomes overloaded, the average delay of data
traffic increases much more sharply than that of multimedia traffic. This is due to the way in which
capacity has been provisioned to the different connections. Specifically, scheduling algorithms have
been configured so that rtPS connections have a reserved rate equal to the mean rate of VoIP and
videoconference applications, respectively. On the other hand, the reserved rate for videoconference
connections accounts for the two videoconference sources multiplexed into the same connection.
Finally, BE connections are reserved a rate of 10 B/s. Note that the rate guaranteed to BE connections
is negligible with respect to the rate guaranteed to rtPS connections, and this justifies the different
performance of the BE and rtPS connections, respectively. With regard to uplink connections, the
average delay of rtPS connections is almost constant when the number of SSs increases. This is
because the BS grants a unicast poll to each rtPS connection every 20 ms. On the other hand, SSs have
to request bandwidth for BE connections on a contention basis. Thus, as soon as the system gets
overloaded, the average delay of BE connections increases remarkably, whereas the average delay of
rtPS connections remains low.
Since VoIP and videoconference are interactive multimedia applications, a relevant performance
index is also the 99th percentile of the delay variation. Considering the performance of downlink
traffic, when the number of SSs increases, the delay variation increases smoothly from 10 to 20 ms.
Under these conditions the system is underloaded. When the number of SSs increases further, the BS
downlink scheduler is not able, on average, to schedule each VoIP packet before the next one is
generated from the same application. Hence, the delay variation of multimedia traffic increases
sharply. In the study, the performance of uplink multimedia traffic and the bandwidth-request
mechanisms used by rtPS and BE connections have also been discussed.
In summary, they have assessed the performance of an IEEE 802.16 system under two traffic
scenarios. The first one (residential scenario) dealt with data (non-QoS) traffic only, and was thus
managed by the BE scheduling service. The results have shown that the average delay of the uplink
traffic is higher than that of the downlink traffic. Furthermore, the former increases more sharply than
the latter with the offered load. This behavior can be explained by means of both the bandwidthrequest mechanism and the overhead introduced by physical preambles. In the second scenario (SME
scenario), on the other hand, they have shown the service differentiation, in terms of delay, between
data (served via BE) and multimedia traffic (served via rtPS). This is achieved because scheduling in
802.16 is controlled by the BS in both the downlink and uplink directions. Therefore, it is possible to
employ scheduling algorithms that have been proposed for wired environments, which are able to
provide QoS guarantees. In the simulations, they have evaluated the DRR and WRR scheduling
algorithms as possible candidates for algorithms to be implemented in a production system.
C. “Providing integrated QoS control for IEEE 802.16 broadband wireless access systems” [CJG-05]
In this study, the authors propose a new integrated QoS architecture for IEEE 802.16 Broadband
Wireless MAN in TDD mode. A mapping rule for providing DiffServ between IP layer and MAC
layer is given and a fast signaling mechanism is designed to provide cross layer integrated QoS for
Point to Multi-Point (PMP) mode. Since IP network service is based on a connectionless and besteffort model, this service model is not adequate for many applications that normally require assurances
on QoS performance metrics. So, a number of enhancements have been proposed to enable offering
different levels of QoS in IP networks including the integrated services (IntServ) architecture, the
differentiated service (DiffServ) architecture.
IntServ is implemented by four components: the signaling protocol (e.g. RSVP), the admission
control, the classifier and the packet scheduler. Applications requiring guaranteed service or
13
controlled-load service must set up the paths and reserve resources before transmitting their data. The
admission control routines will decide whether a request for resources can be granted. After
classification of packets in a specific queue, the packet scheduler will then schedule the packet to meet
its QoS requirement. In [Std-04] and [Std-05], some rules to classify DiffServ IP packets into different
priority queues are also proposed based on IP QoS indication bits in IP header. So, in general, the QoS
architecture of IEEE 802.16 under PMP mode can support both IntServ and DiffServ.
Two ways for providing cross layer QoS control via WirelessMAN technology may be candidates:
For the first one, the traditional RSVP is used to provide cross layer QoS control. RSVP signaling
message will be regarded as the traffic data by convergence sub-layer in the MAC layer. RSVP
signaling message can be classified into a special high priority queue by a protocol-specific packet
matching criteria. This RSVP queue will be transmitted in the second management connection.
In summary, the QoS provision procedure will consist of the following two part: on one hand, the
secondary management connection will be used for RSVP to provide the layer 3 QoS; on the other
hand, the primary management connection will be used for DSA/DSC/DSD to provide the layer 2
QoS. Since the second management connection is defined for delay tolerant traffic and there are many
other IP protocol related message (DHCP, SNMP, TFTP, etc) sharing the same queue, the whole QoS
provision will be rather slow. Furthermore, considering the RSVP signaling has a periodical refreshing
procedures, which consume a lot of bandwidth, it is not efficient to use the secondary management
connection for the RSVP. The second one is the proposed way. Since there are so many similarities
between providing Internet IntServ using RSVP and MAC layer QoS using DSA/DSC/DSD, naturally,
the mechanism will be superior to the first one in high efficiency and fastness by mapping between the
cross layer QoS control and MAC QoS in IEEE802.16 network.
The message exchange for DSA and DSC can be deployed to achieve QoS guarantees through
end-to-end resource reservation for packet flows and to perform per-flow scheduling which IntServ
services require. For DiffServ services, on the other hand, a number of per-hop behaviors (PHBs) for
different classes of aggregated traffic can be mapped into different connections directly. They propose
an integrated QoS control architecture as shown in the following figure, which implements a cross
layer traffic-based prioritization mechanism in a comprehensive way.
Adapted from [CJG-05]
Step 1 and 2 in the figure show when a new service flow arrives in IP layer, it will be firstly parsed
according to the definition in PATH message (for InteServ) or Differentiated Services Code Point
(DSCP for DiffServ); then classified and mapped into one of four types of services (UGS, rtPS, nrtPS
or BE). In step 3, the dynamic service model in SS will send request message to the BS, then the
admission control in BS will determine whether this request will be approved or not. If not, the service
module will inform upper layer to deny this service in step 4; if yes, admission control will notify
scheduling module to make a provision in its basis scheduling parameter according to the value shown
in the request message and the accepted service will transfer into traffic grooming module in step 5.
14
According to the traffic grooming result, SS will send Bandwidth Request message to BS in step 6.
The scheduling module in BS will retrieve the requests (step 7) and generate UL-MAP and DL-MAP
message (step 8) following the bandwidth allocation results. Finally, the SS will package SDUs from
IP layer into PDUs and upload them in its uplink slot to BS (step 9-10).
Traffic Classification and Mapping Strategies for IntServ Services
The sender will send a PATH message including traffic specification (TSpec) information. The
parameters such as up/bottom bound of bandwidth, delay and jitter can be easily mapped into
parameters in DSA message such as Maximum Sustained Traffic Rate, Minimum Reserved Traffic
Rate, Tolerated Jitter and Maximum Latency. According to the response of DSA message, the
provisioned bandwidth can be also freely mapped into reserved specification (RSpec) into RESV
message. Four rules are defined to map IP layer service into MAC layer services.
Traffic Classification and Mapping Strategies for DiffServ Services
For DiffServ services, DSCP code is deployed for classification. There are three definitions of perhop behavior (PHB) to specify the forwarding treatment for the packet. Expedited forwarding (EF) is
intended to provide a building block for low delay, low jitter and low loss services by ensuring that the
EF aggregate is served at a certain configured rate. Assured Forwarding (AF) PHB group is to
provider different levels of forwarding assurances for IP packets. Four AF classes are defined, where
each AF class is allocated a certain amount of forwarding resources (buffer space and bandwidth).
Four rules are defined to map IP layer service into MAC layer services.
Admission Control and Scheduling in BS
It will collect all the DSA/DSC/DSD requests and update the estimated available bandwidth based
on bandwidth change. The hierarchical structure of the bandwidth allocation in BS is shown in the
following figure. In this architecture, twolayer scheduling is deployed. Six queues are
defined according to their direction (uplink or
downlink) and service classes (rtPS, nrtPS and
BE). Since service of UGS will be allocated
fixed bandwidth (or fixed time duration) in
transmission, these bandwidths will be cut
directly before each scheduling.
The algorithm of the first layer scheduling
is called Deficit Fair Priority Queue (DFPQ)
proposed in [CJW-05], which is basically based
on priority queue. Two policies of initial
priority are defined as following:
Service class based priority:
rtPS > nrtPS> BE
Transmission direction based priority:
Downlink >Uplink.
In the second layer scheduling, three
Adapted from [CJG-05]
different algorithms are assigned to three classes
of service to match its requirements. They have
applied earliest deadline first (EDF) for rtPS [GGP-94], which means packets with earliest deadline
will be scheduled first. The information module determines the packets’ deadline and the deadline is
calculated by its arrival time and maximum latency. Weight fair queue (WFQ) [DKS-89] is deployed
for nrtPS services. This type of packets is scheduled based on the weight (ratio between a connection’s
nrtPS Minimum Reserved Traffic Rate and the total sum of the Minimum Reserved Traffic Rate of all
nrtPS connections). The remaining bandwidth is allocated to each BE connection by round robin (RR).
Comparison between two ways of RSVP
The negotiation of QoS parameters for one traffic will be processed two times. For the first time,
the parameters are carried in RSVP messages and transmitted through the Secondary Management
connection. For the second time, the same parameters are mapped in MAC message and transmitted
through the Primary Management Connection. With the utilization of the new mapping rule, the RSVP
15
signaling messages are mapped directly into the MAC messages, and then transmitted through the
Primary Management Connection. In this way, the messages are transmitted only once, reducing the
delay.
They have also developed a simulation platform for the proposed integrated QoS architecture and
the behavior of the architecture has been verified by simulation.
D. Brief introduction of five other studies
D.1. “A Quality of Service Architecture for IEEE 802.16 Standards” [AMY-05]
In this paper, the authors have introduced an architecture to support Quality of Service in IEEE
802.16 standards. They have also proposed a design approach to implement such architecture.
Simulation result shows the performance of their architecture for all types of traffic classes defined by
the standard. They have developed some compatible methods for specific modules such as Scheduler,
Traffic Shaper, and Request and Grant Manager to optimize Delay, Throughput and Bandwidth
Utilization metrics. The simulation results show that the proposal meets these objectives.
According to them, considering the fact that IEEE 802.16 standards are connection-oriented, each
user first sends a Connection Establishment Request to BS. The request is then analyzed in Call
Admission Control and if accepted, attributes of QoS and also two identifiers for each direction of this
connection are registered in Service Flow Data Bases. In order to perform QoS process, packets are
classified according to their mentioned identifiers in MAC entrance point by Classifiers. For further
details, please refer to their article.
D.2. “Quality of service scheduling in cable and broadband wireless access systems” [HPE-02]
In this article, the authors have presented a new and efficient scheduling architecture to support
bandwidth and delay QoS guarantees for both DOCSIS and IEEE 802.16. Their design goals are
simplicity and optimum network performance. The developed architecture supports various types of
traffic including constant bit rate, variable bit rate (real-time and non-real-time) and best effort.
The architecture supports tight delay guarantees for UGS traffic and minimum bandwidth
reservations for rtPS, nrtPS and BE flows. They remind that vendor-specific QoS parameters can also
be used in DOCSIS and IEEE 802.16 and this means that users can also request QoS delay bounds for
their rtPS and nrtPS service flows. They mention that since they are using a fair queuing algorithm in
their scheduler, providing such guarantees is feasible and can be implemented easily given that the
service flows feeding the scheduler are properly policed.
They also introduced a dynamic minislot allocation scheme that should improve the performance
of the scheduling algorithm under varying load conditions. It speeds up the contention phase by
providing extra bandwidth for contending packets. They argue that the loss of throughput due to this
operation should not be a severe one. This is mainly due to the fact that request packets are much
smaller than actual data packets, and because their algorithm allocates fewer contention minislots as
the load on the system increases. For further details, please refer to their article.
D.3. “Exploiting MAC flexibility in WiMAX for media streaming” [SCGI-05]
In this article , the authors have studied the media access control (MAC) layer of WiMAX and
have exploited its flexible features to dynamically construct the MAC packet data units (MPDU). The
sizes of the MPDUs are constantly modified based on the channel state information. The desired
payload is obtained either by aggregation or fragmentation of the upper layer data units. The
robustness of MPDUs is also made tunable by means of cyclic redundancy code bits. They have
considered the both scenarios- with and without feedback. They have adhered to the 802.16
specifications and proposed adapting the MPDU length for streaming media for better performance.
Three metrics are defined: restore probability, goodput and dropping probability. Simulation
experiments are conducted which show the performance enhancements of the proposed ARQ-enabled
adaptive algorithm in terms of these three metrics.
They have focused on the MAC common part sublayer to explore its rich set of features. This
sublayer controls the on-air timing based on consecutive frames that are divided into time slots. The
size of these frames and the size of the individual slots within these frames can be varied on a frame-
16
by-frame basis. This allows effective allocation of on-air resources and they have applied this
mechanism on the MPDUs that are to be transmitted. Depending on the feedback received from the
receiver and on-air physical layer slots, they have exploited the feature of the common part sublayer
that modifies the size of the MPDUs by changing the size of the payload.
The optimal size of the MPDU must be matched to the channel conditions so as to obtain a desired
level of performance. Since packets often get lost being corrupted during transmission in error prone
wireless channels, ARQ mechanism is usually used to identify and possibly recover the missing
frames. In their case, ARQ plays a crucial role in estimating the channel condition and the fate of the
MPDUs that have been transmitted. As a result, the round trip time (RTT) becomes crucial in
determining the size of the MPDUs.
In summary, they have studied the problem of streaming media over WiMAX and exploited the
flexible features present in the MAC layer of 802.16a. They proposed that the size of MAC packet
data units be made adaptive to the instantaneous wireless channel state condition. Based on the type of
feedback received, variable size MPDUs were constructed either by aggregation or fragmentation of
MAC service data units. They conducted simulation experiments to verify the validity of their
proposed scheme. Packet restore probability, goodput, and dropping probability of MPDUs were
defined as the performance metrics. Simulation results demonstrate the effectiveness and performance
improvement of the proposed scheme. For further details, please refer to their article.
D.4 “Algorithms for routing and centralized scheduling to provide QoS in IEEE 802.16 mesh networks” [SSh05]
In this article, the authors have considered the problem of routing and centralized scheduling for
IEEE 802.16 mesh networks. They first have fixed the routing, which reduces the network to a tree.
Then, they have presented scheduling algorithms which provide per flow QoS guarantees to real and
interactive data applications while utilizing the network resources efficiently. The algorithms are also
scalable: they do not require per flow processing and queuing and the computational requirements are
minimal. They have also discussed admission control policies which ensure that sufficient resources
are available. They have handled UDP and TCP traffic separately at first and then jointly. They have
verified their algorithms via extensive simulations. For further details, please refer to their article.
D.5. “Modeling and performance analysis of the distributed scheduler in IEEE 802.16 mesh mode” [CMZ-05]
In this paper, the authors have presented an analytical model for the distributed scheduling
algorithm in the IEEE 802.16 mesh mode. The medium access control (MAC) layer of the IEEE
802.16 has point-to-multipoint (PMP) mode and mesh mode. In the mesh mode, all nodes are
organized in an ad hoc fashion and use a pseudo-random function to calculate their transmission time
based on the scheduling information of the two-hop neighbors. In this paper, they have developed a
stochastic model for the distributed scheduler of the mesh mode. With this model, they have analyzed
the scheduler performance under various conditions, and the analytical results match with the ns-2
simulation results. The analytical model developed in this paper is instrumental in optimizing the
IEEE 802.16 mesh mode system performance.
In the mesh mode, every node competes for the channel access and tries to broadcast its scheduling
information periodically. The channel contention result is correlated with the total node number,
exponent value and network topology. Their model assumes that the transmit time sequences of all the
nodes in the control subframe form statistically independent renewal processes. Based on this
assumption, they have developed methods for estimating the distributions of the node transmission
interval and connection setup delay, which are instrumental for evaluating performance, like
throughput and delay. Comparisons with ns-2 simulation results show that the model is quite accurate
in typical scenarios. Since the detail reservation scheme for the data subframe of the IEEE 802.16
mesh mode is left unstandardized, their model also sheds some light on the data subframe reservation
scheme. For example, based on their analysis, the nodes with real time traffic shall have smaller
holdoff exponents because they can have more chance to obtain data channel. However, too many
nodes with small exponent value generate intensive competition that wastes system resource. Then the
nodes can adjust their exponent values adaptively according to the competition node number variation
to meet the connection QoS requirements. A good reservation scheme should guarantee the bandwidth
17
allocation fairness and improve the channel utilization at the same time. Such a reservation scheme
needs the information like the connection setup time and success probability provided by their model.
For further details, please refer to their article.
4.2
PHYSICAL LAYER QoS ARCHITECTURES
Two different techniques proposed by two different papers will be presented. The concepts will be
explained but the different parts will be exempted of the equations and all the algorithms details as it is
not the goal of this paper.
A.
Proposed solutions in “A Low Complexity Algorithm for Proportional Resource Allocation in OFDMA
Systems” [WSEA-04]
In this paper, a subcarrier allocation scheme is developed. It aims to linearize the power allocation
problem while attaining approximate rate proportionality.
The goal is to reduce significantly the complexity and maintaining a reasonable performance. [WSEA04]
Orthogonal Frequency Division Multiple Access System Model
In the above picture is represented, a block diagram for an OFDMA downlink system from the BS
to the SSes. The BS transmitter emits some bit streams for each users k on different subcarriers with
an index n ranging from 1 to N (N being the total number of subcarriers), allocated to a user k with a
certain power p(k, n). The subcarriers are assumed not to be shared by several users. The transmitted
signal (user’s bits) is modulated into QAM symbols with a Gray bit mapping before being combined
into an OFDMA symbol using an IFFT block.
The transmission channel is modeled by a slowly time varying, selective Rayleigh channel of
bandwidth B.
At the reception, each SS decodes the bits on their assigned subcarriers only. This information is
provides thru a control channel.
Also a channel estimation operation is performed. The estimation results are sent back from the
receiver to the emitter (feedback), and give some information for the resource allocation algorithms.
The slowly time varying nature of the channel is essential in this case otherwise resource allocation,
realized thanks to the feedback, would not be efficient. [WSEA-04]
High Complexity Algorithm
“In [SAE-03], the approach was to first determine the subcarrier allocation, followed by the power
allocation. The subcarrier allocation was determined by allowing each user to take turns choosing the
best subcarrier for him. In each turn, the user with the least proportional capacity gets the priority to
choose his best subcarrier. After the subcarrier allocation, the power allocation is then simplified into a
maximization over continuous variables p(k,n).” They refer to this method of subcarrier and power
allocation as root-finding. But the complexity of the calculations, used in these algorithms, makes
them impractical for real-time systems. That is the reason why an approach called linear was proposed
for reducing the complexity as well as maintaining good performances. [WSEA-04]
18
Reduced Complexity Algorithm
The different steps of this proposed solution involve the determination of the number of
subcarriers N(k) assigned for each user, the assignment of the number of subcarriers distributed for
each user k to guaranty a certain proportionality, the assignment of total power P(k) allocated to a
given user k to maximize the channel capacity under the constrain of the proportionality, and finally
the assignment of power p(k, n) for each subcarrier assigned to one user. [WSEA-04]
The steps are roughly described without any equation or algorithm. Please refer to [WSEA-04] for
more details. The different steps involve:
• Step1 Number of subcarriers per use
“This initial step is based on the reasonable assumption also made in [YL-00] that the
proportion of subcarriers assigned to each user is approximately the same as their eventual
rates after power allocation, and thus would roughly satisfy the proportionality constraints.”
The total number of subcarriers N is distributed equally among the users and N∗ unallocated
subcarriers are left.
• Step2 Subcarrier assignment
This step allocates the per user set of subcarriers N(k) and then the remaining N∗ unallocated
subcarriers in a way that maximizes the overall capacity while maintaining rough
proportionality.
• Step3 Power allocation among users
The output of the previous two steps is a subcarrier allocation for each user that reduces the
resource allocation problem to the finding of an optimal power allocation.
• Step4 Power allocation across subcarriers per user
The previous step gives the total power P(k) for each user k, which are then used in this final
step to perform waterfilling across the subcarriers for each user. The waterfilling process aims
to attribute power to each subcarrier to increase capacity having a constant total power.
B.
Proposed solutions in “QoS Aware Adaptive Resource Allocation Techniques for Fair Scheduling in
OFDMA Based Broadband Wireless Access Systems” [ECV-03]
In this paper, an iterative multi-user bit and power allocation scheme are introduce to fulfill QoS
requirements to maintain reasonable performance for each user. The objective is to minimize the total
transmitted power while allocating the subcarriers to the users. It is also to determine the bit rate
transmitted on each subcarrier. An adaptive modulation was previously considered in [WCLM-99],
[KLKL-01]. The scheme is simple and sufficiently fair to meet real time applications criteria in which
a quick scheme is needed to allocate subcarriers before the channel changes and a fair scheme is
needed to treat each user. [ECV-03]
A continuous allocation scheme is also proposed where the allocator uses the previous channel
information per user for the current allocation. An extension of the point-to-point version of
proportional fair scheduling (as in [KLKL-01]) to a point-to-multipoint version is proposed. In this
scheme there is no fixed requirements per symbol, the aim is to maximize capacity. [ECV-03]
Orthogonal Frequency Division Multiple Access System
The difference between an OFDM and an OFDMA system is that OFDMA involve several users
that must share several subcarriers and therefore it involves the need of an FFT block. Otherwise, the
rest of the system is similar with OFDM. Each user is allocated a set of non overlapping subcarriers.
A guard insertion is needed in order to prevent Inter Symbol Interference (ISI) at the emission, at the
reception, after the sampler, the bits corresponding to the guard time are discarded. Each set of
subcarriers contain a given user’s bit. The modulation used to code the different bit stream for each
user is adaptive. [ECV-03]
19
[ECV-03]
“In a perfectly synchronized system, the allocation module of the transmitter assigns subcarriers
to each user according to some QoS criteria. QoS metrics in the system are rate and bit error rate
(BER). Each user’s bit stream is transmitted using the assigned subcarriers and adaptively modulated
for the number of bits assigned to the subcarrier. The power level of the modulation is adjusted to
overcome the fading of the channel. The transmission power for AWGN channel can be predicted. In
addition the channel gain of subcarrier to the corresponding user should be known.” [ECV-03]
In this study, the channel is assumed to be known at both transmitter and receiver. The transmitter
and receiver will be able to estimate the channel as long as the channel variation over time is slow.
The resource allocation should be done within the coherence time prior to this statement. Also this
channel property is required to apply a continuous allocation, where the allocator uses the previous
channel information per user for the current allocation. “With the channel information, the objective of
resource allocation problem can be defined as maximizing the throughput subject to a given total
power constraint regarding the user’s QoS requirements”. [ECV-03]
The resource allocation problem is formulated with a total transmission power constraint. This
transmission power is function of the required received power with unity channel gain for a reliable
reception of the modulated symbols. The maximum BER(max) is predefined and the required BER(k)
for each user should below BER(max). The data rate for each user should be equal to the required data
rate R(k). Therefore the modulation type and the BER get involved in the resource allocation decision
process. [ECV-03]
Optimal Solution
Several solutions have been described. Only the optimal one in this paper gives the exact solution
of the problems mentioned above. As the previous paper from an implementation point of view it is
not realistic because the allocation algorithm would not be executed fast enough as the channel is
varying in time and also because the Integer Programming that would be used increase the complexity
exponentially with the number of constraints. [ECV-03]
Suboptimal Solution
In most attempts to simplify the resource allocation problem, the problem is decomposed into two
procedures: A subcarrier allocation with fixed modulation, and bit loading. Subcarrier allocation with
fixed modulation deals with one matrix with fixed and then by using bit loading scheme, the number
of bits is incremented. [ECV-03]
20
The subcarrier allocation problem can be solved with Linear Programming (LP) or Hungarian
algorithms. Although the Hungarian algorithm is proposed as an optimal solution for resource
allocation with a fixed modulation in [PJ-02], [WTCL-99], it is considered as a suboptimal solution
for adaptive modulation. Linear programming is investigated in [KLKL-01].
The bit loading algorithm (BLA) appears after the subcarriers are assigned to users that have at
least a certain number of bits assigned. Bit loading procedure is as simple as incrementing bits of the
assigned subcarriers of the users until the total power is less or equal to the upper limit of the total
transmission power. It allows to convert the fixed modulation scheme into adaptive modulation
scheme for each subcarrier. [ECV-03]
Iterative Solution
“The GreedyLP and GreedyHungarian methods both first determine the subcarriers and then
increment the number of bits on them according to the rate requirements of users. This may not be a
good schedule in some certain cases: For instance, consider a user with only one good subcarrier and
low rate requirement. The best solution for that user is allocating its good carrier with high number of
bits. But if GreedyLP or Greedy-Hungarian is used, user may have allocated more than one subcarrier
with lower number of bits and in some cases, its good subcarrier is never selected. Consider another
scenario where a user does not have any good subcarrier (i.e. it may have a bad channel or be at the
edge of the cell). In this case, rather than pushing more bits and allocating less subcarriers as in
GreedyLP and GreedyHungarian, the opposite strategy is preferred since fewer bits in higher number
of subcarriers give better result. Another difficulty arises in providing fairness. Since GreedyLP and
GreedyHungarian are based on greedy approach, the user in the worst condition usually suffers. In any
event, these are complex schemes and simpler schemes are needed to finish the allocation 366 IEEE
TRANSACTIONS ON BROADCASTING, VOL. 49, NO. 4, December 2003 within the coherence
time. To cope with these challenges, we introduce a simple, efficient and fair subcarrier allocation
scheme with iterative improvement.” [ECV-03]
The proposed scheme is composed of two modules called scheduling and improvement modules.
For scheduling, bits and subcarriers are distributed to the users and passed to the improvement module.
The allocation is then improved iteratively by bit swapping and subcarrier swapping algorithms.
[ECV-03]
Physical Layer QoS Architectures Summary
“The paper [WSEA-04] presents a new method to solve the rate-adaptive resource allocation
problem with proportional rate constraints for OFDMA systems. It improves on the previous work in
this area [SAE-03] by developing a subcarrier allocation scheme that achieves approximate rate
proportionality while maximizing the total capacity. This scheme was also able to exploit the special
linear case in [SAE-03], thus allowing the optimal power allocation to be performed using a direct
algorithm with a much lower complexity versus an iterative algorithm. It is shown through simulation
that the proposed method performs better than the previous work in terms of significantly decreasing
the computational complexity, and yet achieving higher total capacities, while being applicable to a
more general class of systems.” [WSEA-04]
“In [ECV-03] is considered the problem of resource allocation for adaptive modulation in
OFDMA systems. Two different approaches are introduced. One maximizes the capacity and the other
one satisfies fixed QoS criteria (i.e the rate and bit error rate requirements) in each symbol. Recent
work has focused on developing algorithms to meet the QoS criteria [KT-01], [WCLM-99]–[RC-00].
In an OFDMA system, subcarriers are distributed among users and number of bits transmitted in each
subcarrier is adjusted according to the rate requirements of users to minimize total transmit power. It
has been shown that resource allocation can be optimized by Integer Programming [KLKL-01].
However, the optimal solution can not be implemented in real time. A simple suboptimal solution that
fairly allocates and efficiently converges close to optimal meeting the QoS criteria per symbol was
proposed. The algorithm showed good performance in terms of tight power control, iterative
betterment and fair scheduling among users when compared with the optimal solution and previously
proposed suboptimal schemes. The proposed solution can also be applied to the uplink when there is
21
perfect synchronization. We also considered a possible resource allocation scheme when the objective
is to maximize capacity, based on proportional fair scheduling algorithm for point-to-point
communication introduced in [VTL-02].” [ECV-03]
Consequently, both papers proposed efficient ways to solve the rate-adaptive resource allocation
problem to meet the required Quality of Service criteria for OFDMA systems.
REFERENCES
[AMY-05]
[CJG-05]
[CJW-05]*
[CLME-06]
[CMZ-05]
[CWM-02]
[DKS-89]*
[ECV-03]
[EMSW-02]
[GGP-94]*
[GSS-99]*
[GWAC-05]
[HPE-02]
[Intel-04]
[JL-03]*
[KH-95]*
[KLKL-01]*
[KSC-91]*
[KT-01]*
[LBS-97]*
H.S.Alavi, M.Mojdeh, N.Yazdani, “A Quality of Service Architecture for IEEE 802.16
Standards”, 2005 Asia-Pacific Conference on Communications, 03-05 Oct. 2005
Jianfeng Chen, Wenhua Jiao, Qian Guo, “Providing integrated QoS control for IEEE 802.16
broadband wireless access systems”, IEEE 62nd Vehicular Technology Conference, 25-28 Sept.,
2005
Jianfeng Chen, Wenhua Jiao, Hongxi Wang, “A Fair Scheduling for IEEE 802.16 Broadband
Wireless Access Systems”, ICC2005, May 16-20, Souel, Kerea
C.Cicconetti, L.Lenzini, E.Mingozzi, C.Eklund, “Quality of service support in IEEE 802.16
networks”, IEEE Network, Vol. 20, Issue 2, Mar./April 2006
M.Cao, W.Ma, Q.Zhang, X.Wang and W.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, Urbana-Champaign, IL, USA, May 2527, 2005
G.Chu, D.Wang, S.Mei, “A QoS Architecture for the MAC Protocol of IEEE 802.16 BWA
System”, IEEE 2002 International Conference on Communications, Circuits and Systems and
West Sino Expositions, 29 June-1 July 2002
Demers A, Keshav S, Shenker S. “Analysis and Simulation of a Fair Queuing Algorithm”.
SIGCOMM CCR 19 1989;
Mustafa Ergen, Sinem Coleri, and Pravin Varaiya “QoS Aware Adaptive Resource Allocation
Techniques for Fair Scheduling in OFDMA Based Broadband Wireless Access Systems”, IEEE
Trans. Broadcast, vol. 49, no. 4, Dec. 2003.
C.Eklund, R.B.Marks, K.L.Stanwood and S.Wang, “IEEE Standard 802.16: A Technical
Overview of the WirelessMAN™ Air Interface for Broadband Wireless Access”, IEEE
Communications Magazine, June 2002.
Georgiadis L, Guerin R, Parekh A. “Optimal Multiplexing on a Single Link: Delay and Buffer
Requirements”. Proceedings of IEEE INFOCOM 94; vol. 2, 1994; 524–532.
N.Golmie, Y.Saintillan, and D.H.Su, “A Review of Contention Resolution Algorithms for IEEE
802.14 Networks”, IEEE Commun. Surverys, 1st Quarter 1999
A.Ghosh, D.R.Wolter, J.G.Andrews and R.Chen “Broadband Wireless Access with
WiMax/802.16: Current Performance Benchmarks and Future Potential”, IEEE Communications
Magazine, Feb. 2005.
M. Hawa, D.W. Petr, “Quality of service scheduling in cable and broadband wireless access
systems”, Tenth IEEE International Workshop on Quality of Service, 15-17 May 2002
G. Nair et al. “IEEE 802.16 Medium Access Control and Service Provisioning”, Intel Technology
Journal, vol.08, issue 03, Aug. 20 2004
J. Jang and K. B. Lee, “Transmit Power Adaptation for Multiuser OFDM Systems,” IEEE J.
Select. Areas Commun., vol. 21, pp. 171–178, February 2003.
R. Knopp and P. Humblet, “Information capacity and power control in single cell multiuser
communications,” in ICC ’95, vol. 1, 1995, pp. 331–335.
I. Kim, H. L. Lee, B. Kim, and Y. H. Lee, “On the use of linear programming for dynamic
subchannel and bit allocation in multinser OFDM”, in IEEE GLOBECOM ’01, vol. 6, 2001, pp.
3648–3652.
M. Katevenis, S. Sidiropoulos, and C. Courcoubetis, “Weighted Round-Robin Cell Multiplexing
in a General-Purpose ATM Switch Chip,” IEEE JSAC, vol. 9, no. 8, Oct. 1991, pp. 1265–79.
I. Koutsopoulos and L. Tassiulas, “Channel state-adaptive techniques for throughput enhancement
in wireless broadband networks,” in INFOCOM 2001, vol. 2, 2001, pp. 757–766.
Songwu Lu, Vaduvur Bharghavan, and R.Srikant, “Fair scheduling in wireless packet networks”,
Proceedings of ACM SIGCOMM 1997
22
[Mar-04]
Roger B. Marks, (US) National Institute of Standards and Technology, IEEE 802.16 Working
Group Chair, IEEE Computer Society Distinguished Visitors Program Presentation, IEEE Oregon
Section, Beaverton, OR, USA, 14 July 2004
[MEK-01] Roger B. Marks, Carl Eklund et al. Presentation “The 802.16 WirelessMAN™ MAC: It’s Done,
but What Is It?” November 2001
[MLK-00]* Jay R.Moorman, John Lockwood, and Sung-Mo Kang, “Wireless quality of service using
multiclass priority fair queuing”, IEEE JSAC, Aug.2000
[PGa-93]* Abhay K.Parekh and Robert G.Gallage, “A generalized processor sharing approach to flow control
in integrated services networks: The single-node case”, IEEE/ACM Transactions on Networking
1(3) Jun. 1993.
[PJ-02]*
S. Pietrzyk and G. J. M. Janssen, “Multiuser subcarrier allocation for QoS provision in the
OFDMA systems,” in Proc. VTC 2002, vol. 2, 2002, pp. 1077–1081.
[RC-00]*
W. Rhee and J. M. Cioffi, “Increase in Capacity of Multiuser OFDM System Using Dynamic
Subchannel Allocation,” in Proc. IEEE Vehic. Tech. Conf., Tokyo, Japan, May 2000, pp. 1085–
1089.
[SAE-03]* Z. Shen, J. G. Andrews, and B. L. Evans, “Optimal Power Allocation in Multiuser OFDM
Systems,” in Proc. IEEE Global Communications Conference, San Francisco, CA, Dec. 2003, pp.
337–341.
[SCGI-05] S. Sengupta, M. Chatterjee, S. Ganguly, R. Izmailov, “Exploiting MAC flexibility in WiMAX for
media streaming”, Sixth IEEE International Symposium on a World of Wireless Mobile and
Multimedia Networks, 13-16 June 2005
[ShV-96]* M. Shreedhar and G. Varghese, “Efficient Fair Queuing using Deficit Round Robin,” IEEE Trans.
Net., vol. 4, no. 3, June 1996, pp. 375–85.
[SSh-05]
H.Shetiya and V.Sharma, “Algorithms for routing and centralized scheduling to provide QoS in
IEEE 802.16 mesh networks”, Proceedings of the 1st ACM workshop on Wireless multimedia
networking and performance modeling, Montreal, Quebec, Canada, Oct.13, 2005
[Std-04]
IEEE 802.16-2004, “IEEE Standard for Local and Metropolitan Area Networks — Part 16: Air
Interface for Fixed Broadband Wireless Access Systems,” Oct. 1, 2004.
[Std-05]
IEEE 802.16e-2005, “IEEE Standard for Local and Metropolitan Area Networks — Part 16: Air
Interface for Fixed and Mobile Broadband Wireless Access Systems,” Feb. 28, 2006.
[SVa-98]* D. Stiliadis and A. Varma, “Latency-Rate Servers: A General Model for Analysis of Traffic
Scheduling Algorithms,” IEEE/ACM Trans. Net., vol. 6, Oct. 1998, pp. 675–89.
[VTL-02]* P. Viswanath, D. N. C. Tse, and R. Laroia, “Opportunistic beamforming using dumb antennas,”
IEEE Trans. Information Theory, , vol. 48, no. 6, pp. 1277–1294, June 2002.
[WCLM-99]* C. Y. Wong, R. S. Cheng, K. B. Lataief, and R. D. Murch, “Multiuser OFDM System with
Adaptive Subcarrier, Bit, and Power Allocation”, IEEE J. Select. Areas Commun., vol. 17, pp.
1747–1758, Oct 1999.
[WSEA-04] Ian C. Wong, Zukang Shen, Brian L. Evans, and Jeffrey G. Andrews , “A Low Complexity
Algorithm for Proportional Resource Allocation in OFDMA Systems”, 2004
[WTCL-99]* C. Y.Wong, C. Y. Tsui, R. S. Cheng, and K. B. Letaief, “A real-time subcarrier allocation scheme
for multiple access downlink OFDM transmission”, in Proc. VTC 1999, vol. 2, 1999, pp. 1124–
1128.
[YL-00]*
H. Yin and H. Liu, “An Efficient Multiuser Loading Algorithm for OFDM-based Broadband
Wireless Systems,” in Proc. IEEE Global Telecommunications Conference, vol. 1, 2000, pp. 103–
107.
* This reference has not been consulted by the authors of this article; rather, it had been referenced by the
authors of one of the articles that were studied in this research.
23
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising