Samsung Mobile Wimax

WiMAX Worldwide Interoperability for Microwave Access
White Paper
WiMAX is a wireless access technology for building networks with large coverage areas and high data rates, so-called Metropolitan Area Networks (MANs). It focuses on various usage scenarios for serving fixed, nomadic and mobile subscribers and incorporates a broad range of transmission and access technologies, which can be dynamically applied for serving these different types of subscribers. In addition, it provides mechanisms for giving Quality-of-Service (QoS) guarantees, and thus it is predestined for enabling real-time services like Voice over IP (VoIP), video on demand or multiplayer gaming.

1 Introduction
Broadband access is the main prerequisite for delivering highly sophisticated IT services to the end user, for example, video on demand, video conferencing, Voice over IP (VoIP) or interactive gaming. After the Internet and mobile communications reached the mass markets in the mid-1990s, it turned out very soon that existing network technologies such as the analog Plain Old Telephone System (POTS), the Integrated Services Digital Network (ISDN) or the Global System for Mobile Communications (GSM) could not fulfil the requirements imposed by these applications. The main reason for this lack was simply the fact that these systems were initially designed for speech telephony, which traditionally is a circuit-switched service of comparatively low bandwidth. As a result, standardization and manufacturers created enhancements and auxiliary technologies for bridging the gap between the capabilities of existing networks and the requirements of emerging applications. Examples are the Digital Subscriber Line (DSL) for delivering packswitched data at high rates over the telephone wire to the end user or the General Packet Radio Service (GPRS) for introducing packet-switched services in GSM networks. In the recent years, DSL has become the standard solution for fixed broadband access in the consumer market. However, it requires complex modifications on the infrastructure of telephony networks and is therefore often not available in rural environments with a low population density.
WiMAX — Worldwide Interoperability for Microwave Access

In the mobile area, on the other hand, the breakthrough of data services is still missing. GPRS has been introduced by all GSM operators in the meantime, but it suffers from low data rates and high delays. Even the Universal Mobile Telecommunications System (UMTS), which was introduced a few years ago as the successor of GSM and which actually targets also at packet-switched data, could not initiate a turn-around towards a broad acceptance of mobile data services so far. A new technology that specifically focuses on broadband access is called WiMAX (Worldwide Interoperability for Microwave Access). It is a wireless technology that does not necessarily replace the systems mentioned before, but that at least acts as an extension, for example, in regions where other broadband technologies are not available or do not provide sufficient capacity or bandwidth. WiMAX has been designed for operation in a broad range of licensed and unlicensed frequency bands, thereby being much more flexible than cellular networks like GSM and UMTS, which are confined to operation in dedicated, licensed frequency bands being subject to regulation. WiMAX covers different usage scenarios, ranging from supporting mobile users to connecting LANs (Local Area Networks) to the Internet. To fulfil the heterogeneous requirements on data transmission imposed by these scenarios, WiMAX incorporates several physical layers with different modulation schemes, antenna designs and other features. Furthermore, WiMAX provides sophisticated functions for guaranteeing a certain quality1

of-service (QoS) during transmission, which is of particular importance for real-time or near real-time applications like VoIP or video streaming. WiMAX is standardized by the Institute of Electrical and Electronics Engineers (IEEE), the same institution that is also responsible for standardization of other wired and wireless access technologies, for example, Ethernet and WLAN (Wireless Local Area Network). The group within IEEE consigned with the specification of WiMAX is known under the identifier 802.16 and denoted as Broadband Wireless Access Working Group. Strictly speaking, the official term for WiMAX is actually Wireless Metropolitan Area Network (WirelessMAN™). The term WiMAX stems from the WiMAX forum, which is an organization of more than 400 operators and manufacturers being concerned with “promoting and certifying the compatibility and interoperability of broadband wireless access equipment that conforms to the IEEE 802.16 standards” [1]. However, in the recent years, the term WiMAX has prevailed against “WirelessMAN” or “802.16”, and that is why this term is also used throughout this paper. The following sections give an overview of WiMAX and introduce its usages scenarios, transmission technologies and basic services.

Base station Subscriber station LoS

((( (((
Indoor network installation

Figure 1. Fixed WiMAX mal conditions, it may achieve transmission ranges of up to 70 km and data rates of up to 134 Mbps. As radio signals above 10 GHz can hardly penetrate obstacles like buildings or hills, an important prerequisite for successful transmission is that a line-of-sight (LoS) path exists between subscriber and base station that is not obstructed by obstacles. Thus, Fixed WiMAX represents an interesting alternative to older or proprietary LoS radio systems of less bandwidth, for example, Wireless Local Loop (WLL).

2 WiMAX Usage Scenarios
The WiMAX usage scenarios are commonly referred to as fixed, nomadic and mobile access, and they are covered by different documents of the IEEE 802.16 standards family. The scenarios impose very different requirements on the used frequency bands, modulation schemes, medium access, and mobility mechanisms, and hence WiMAX today incorporates a number of variants of these technologies.

2.2 Nomadic WiMAX
The major drawback of Fixed WiMAX is the need for outdoor antennas at the subscriber, which requires a cumbersome wiring inside buildings and fixed antenna installations at roofs of considerable height for guaranteeing LoS conditions to the next base station. In order to address these issues, the IEEE has released another standard in April 2003, which is called IEEE 802.16a and which focuses on the nomadic WiMAX access. Radio channels of Nomadic WiMAX occupy frequency bands in the range between 2 and 11 GHz, which in contrast to higher frequencies allow for nonlight-of-sight (NLoS) transmissions between subscriber and base stations and vice versa. As a result, it becomes possible to built WiMAX transceivers with integrated antennas, which can be connected directly to a PC or included into handheld devices or laptops, for example, in form of PCMCIA cards. A fixed-installed outdoor antenna is not necessary any longer, and the WiMAX customer can enter into contact from everywhere within the coverage area of a base station, even from the inside of buildings. This is illustrated in Figure 2. A radio channel of nomadic WiMAX occupies a bandwidth between 1.75 and 20 MHz. The bandwidth has been kept variable, because frequency allocation and licensing are managed very irregular in different countries of the world and significantly vary in the size of frequency bands assigned to the operators. However, the flexibility of nomadic access must be paid by a significant decrease in the transmission range and data rates when compared to Fixed WiMAX. The coverage area of a base station is limited to a radius of 5 km. The maximum data rate, which only has been achieved in field tests so far, is about 70 Mbps, but is expected to be much lower for
WiMAX — Worldwide Interoperability for Microwave Access

2.1 Fixed WiMAX
Initially, WiMAX was designed only for fixed access. The first in a series of standards was released in December 2001 by IEEE and was called IEEE 802.16. It defines a system for the wireless transmission between stationary senders and receivers in outdoor environments. The main components of the system are base stations, which are located at the cell sites of the WiMAX operator, and subscriber stations, which are usually installed at the roofs of buildings at the WiMAX customers, see Figure 1. The subscriber stations have antennas with dimensions comparable to those of satellite dishes. They are connected typically to a local network of the subscriber, for example, a WLAN or Ethernet installation inside the building. The base stations, on the other hand, may be interconnected to public networks like the Internet or to private ones. In an alternative scenario, Fixed WiMAX might also be used by a cellular network operator for realizing connectivity between the cell sites and the core network. Fixed WiMAX has been designed for operation in a very broad frequency range between 10 and 66 GHz with bandwidths of 20, 25 or 28 MHz per radio channel. Under opti2

Base station
Base station

Handover

Subscriber station

NLoS

Subscriber station

Figure 2. Nomadic WiMAX networks operating under real conditions. Both Fixed and Nomadic WiMAX can be operated in two modes, which are referred to as point-to-point (PTP) and point-to-multipoint (PMP) modes. In the former, a base station serves only a single subscriber station, which can exclusively use the entire bandwidth of the radio channel. In PMP, on the other hand, a base station supplies several subscriber stations at once, and hence the available bandwidth must be shared among all subscribers residing in the particular cell. The PTP mode is primarily intended for Fixed WiMAX, while PMP is the preferred choice for nomadic access. In June 2004, standardization activities for Fixed and Nomadic WiMAX were merged. The resulting standard document is called IEEE 802.16-2004 [2] and replaces the former versions IEEE 802.16 and 802.16a. Frequency licensing and first commercial trials for WiMAX in many countries started in 2005, while related products and services for the mass market have been announced to become available in 2007. Starting from this time, it is expected that Fixed and Nomadic WiMAX will be requested especially by customers residing in rural areas, which often suffer from the unavailability of wired broadband technologies like DSL, cable modem or T1 access.

Figure 3. Mobile WiMAX a so-called hard handover. This type of handover is characterized by the fact that the connection to the serving base station is terminated before a new one to another target base station is initialized (“break-before-make”). As a result, the customer experiences a short degradation in the quality of service, that is, an interruption of the data transfer, until the handover is completed. The mobile access mode, on the other hand, has been designed for supporting customers travelling at velocities of up to 125 km/h. It implements a soft-handover, where the connection to the target base station is established before the old connection is released (“make-before-break”). A soft handover happens seamlessly from the point of view of the customer and has a much lower latency than a hard handover. However, this reduced latency must be paid by a much higher complexity in the hardware. Besides these handover mechanisms, Mobile WiMAX includes location management functions, which enable to determine from the set of all base stations a WiMAX network is made up of the base station the target subscriber is currently attached to and which are necessary whenever network-initiated data, for example, incoming Emails, needs to be pushed to a subscriber. Furthermore, Mobile WiMAX defines different power-saving modes to which the device changes if there is no data transmission in progress and which thus contribute to a significant reduction of battery consumption when compared to devices used for fixed or nomadic access. Finally, as data transmission in mobile networks is always exposed to varying radio propagation conditions, Mobile WiMAX comes up with improved modulation and error correction schemes. The specification for Mobile WiMAX has been released as an amendment to the 802.16-2004 standard, and is called IEEE 802.16e [3]. It emerged from the Korean WiBro (Wireless Broadband) technology, which is being developed since the beginning of the millennium by the Korean telecommunications industry under significant participation of Samsung Electronics. Since 2004, WiBro is being standardized by the Korean Telecommunications Technology Association (TTA), and first WiBro networks went into operation in 2005. In November 2004, it was decided to adopt the WiBro technologies for Mobile WiMAX and to keep both systems compatible to each other.

2.3 Mobile WiMAX
A drawback of Nomadic WiMAX is that a service session can only be maintained as long as the subscriber resides in the coverage area of the base station where this session has been initiated. If the subscriber moves from one coverage area to that of another base station, the session is terminated and must be re-initiated at the new base station. An automatic transfer of the session from the serving to another target base station, a process which is called handover, is not possible in Fixed or Nomadic WiMAX systems. The missing support of mobile subscribers has led to initiatives for creating Mobile WiMAX, which, besides various handover mechanisms, also incorporates other mobility functions (see also Figure 3). Mobile WiMAX envisages two access modes, which are called portable and mobile access. The portable access mode serves customers travelling at pedestrian speeds. When changing the cell, the service session is transferred to the target base station by
WiMAX — Worldwide Interoperability for Microwave Access

3

2.4 Mobile WiMAX - Difference to other Systems
The emergence of Mobile WiMAX networks expected for the next years imposes the question of how this system relates to classical wireless technologies like WLAN, GSM and UMTS. The answer to this question is not clear yet and has led to controversies among experts whether WiMAX is rather a competing or complementary technology. In order to get an idea about the role of Mobile WiMAX in the orchestra of wireless consumer technologies it might be helpful to consider these systems regarding their data rates and mobility support capabilities, see Figure 4. Similar to WiMAX, WLANs according to the IEEE standards family 802.11 offer network access, which, however, in contrast to WiMAX is limited to local environments, predominantly inside buildings. A WLAN access point has a typical range of a few hundreds of meters (rather only a few dozens of meters indoors), and a couple of them may be interconnected to a so-called extended service set for providing larger areas with seamless coverage. A cell change is supported by a handover function, which, however, causes noticeable interruptions during transmission and only works at very low speeds. The data rate supported by most WLAN installations today is about 54 Mbps and is thus beyond of what the typical Mobile WiMAX subscriber can expect. Due to its limited mobility support, WLAN is the preferred choice whenever an expensive wiring inside buildings should be avoided, for example, when a PC or notebook needs to be connected to a DSL modem, or for nomadic customers, which require high data rates on the spot, but do not move considerably. For mobile customers, however, WLAN is less suited, as it is very difficult and expensive to build a WLAN that seamlessly cover larger outdoor areas. Other than WLAN and WiMAX, which only provide network access capabilities, traditional cellular systems like GSM and UMTS realize several high-level services such as speech and video telephony, transfer of short messages, or browsing the Internet via the Wireless Application Protocol (WAP). A single network usually spans an entire country, and it consists of many locally operating access networks that are interconnected via a common core network. GSM and UMTS provide full mobility support, including handover, localization and roaming capabilities. Roaming enables customers to request and use services in foreign networks,
Data rates

and was one of the main driving forces behind the success of GSM. As GSM and UMTS in the meanwhile are offered by over 700 network operators in 214 countries and territories, customers on the move experience a nearly seamless world-wide mobility support that no other network technology can provide today. On the other hand, GSM and UMTS only support moderate data rates when compared to those that can be achieved with Mobile WiMAX. GSM was initially designed for circuit-switched speech telephony only and data rates of the packet-switched GPRS are limited to about 60 kbps (depending on the capabilities of the used terminal and the configuration of the serving network). Data rates in UMTS are considerably higher. In the first network expansion stage, these networks provide services with a maximum of 384 kbps, which may be extended to up to 14 Mbps if UMTS is combined with a new technology known as Highspeed Downlink Packet Access (HSDPA) and Highspeed Uplink Packet Access (HSUPA) respectively. To draw a conclusion, from a today's perspective, Mobile WiMAX may be classified as a technology that bridges the gap between traditional cellular networks (seamless mobility support and comparatively low data rates) on the one hand and local wireless technologies like WLAN (high data rates, but only rudimentary mobility functions) on the other.

3 WiMAX Protocol Stack
The WiMAX specifications do not define an entire network infrastructure or high-level services as known from telecommunications systems like GSM or UMTS. They only fix an access technology for connecting subscriber stations over the so-called last mile to a base station, comparable to DSL in the wired domain. This base station then provides interconnectivity with a fixed network, however, the related protocols and mechanisms used for this are out of scope of the IEEE specifications for WiMAX. In terms of the seven layers of the OSI protocol stack, WiMAX covers only the physical (PHY) and medium access (MAC) layers and is thus in close compliance to other IEEE specifications like WLAN 802.11 or Ethernet 802.3. The resulting protocol stack is depicted in Figure 5. The physical layer primarily deals with the representation of data bits by radio signals, for which different modulation schemes are envisaged, as well as with related aspects like antenna technologies and power control. Furthermore, it manages the separation of uplink and downlink transmission, which is called duplexing, and incorporates methods for error correction and detection. For the different variants of WiMAX several physical layers are envisaged, which are called WirelessMAN-SC, WirelessMAN-SCa, WirelessMANOFDM and WirelessMAN-OFDMA. Some characteristic features of them are highlighted in the following sections. As suggested by its name, the medium access layer provides mechanisms that define how a radio channel provided by the physical layer is shared between different subscriber stations. The primary goal of medium access is to avoid collisions that would occur when two or more subscriber
WiMAX — Worldwide Interoperability for Microwave Access

WLAN
(IEEE 802.11)

(IEEE 802.16e)

Mobile WiMAX

HSDPA/ HSUPA

UMTS

GSM/ GPRS
Mobility

Figure 4. Mobile WiMAX - Difference to other systems
4

WiMAX

{

Network layer (e.g., IP)
MAC convergence sub-layer MAC common part MAC privacy sub-layer

Medium access layer (MAC) Physical layer (PHY)

Figure 5. WiMAX Protocol Stack

stations use the same radio resources at the same time. Another important focus is on control mechanisms for guaranteeing a certain performance of data transmission, which is referred to as Quality of Service (QoS). This performance can be described by several parameters, among them data rate, delay, jitter (variation in delay) and error rates. Different applications, for example, multimedia streaming, VoIP and web browsing, impose different requirements on QoS, and WiMAX provides adequate mechanisms to fulfil them. As can be derived from Figure 5, the common part of the medium access layer is supplemented by two sub-layers, referred to as MAC privacy sub-layer and MAC convergence sub-layer. The former provides the usual security mechanisms needed for the authentication of subscribers, the exchange of key and the ciphering of messages. The convergence sub-layer acts as an interface between external non-WiMAX protocols and the WiMAX medium access layer. Its main task is the encapsulation and decapsulation of external Protocol Data Units (PDUs) into and from socalled Service Delivery Units (SDUs), which are exchanged between subscriber and base station. The convergence sub-layer is also responsible for bandwidth allocation and the adherence of negotiated QoS parameters. Two specific convergence sub-layers so far exist, one for carrying data of packet-switched networks like IPv4 or IPv6 and another one for connecting to networks being operated according to the Asynchronous Transfer Mode (ATM).

4 WiMAX Physical Layer
This section highlights the physical layer and gives an overview of modulation schemes, antennas, error correction schemes and frame formats used for WiMAX. 4.1 Modulation Schemes Modulation is a process to represent data by changing the parameters of a periodic sinusoidal electromagnetic wave, which is known as carrier. WiMAX incorporates a multitude of modulation schemes, which can be dynamically deployed under consideration of the error characteristics of the radio channel and the required data rates. Single Carrier Modulation In a single carrier modulation scheme, the transmitter genWiMAX — Worldwide Interoperability for Microwave Access

erates a single carrier of a certain amplitude, frequency and phase. For data transmission, one or several of these parameters are changed depending on the data to be transmitted, which, as mentioned before, is called modulation or, using an alternative term, shift keying. The resulting signal is then emitted by the antenna connected to the transmitter, propagates in the environment, and is finally caught by another antenna, which is connected to a receiver. This receiver then interprets the incoming signal and recovers the data bits originally sent, which is called demodulation. In each modulation scheme, data bits are represented in form of symbols, and each symbol is given by a certain constellation of the carrier’s amplitude, frequency, and phase, the so-called signal state. WiMAX envisages different variants of phase shift keying. The simplest variant is Binary Phase Shift Keying (BPSK) and modulates data by shifting the carrier phase between two signal states, one representing the binary “1” and the other the binary “0”. Thus, each symbol only carries a single bit. For transferring more bits per symbol, one needs a modulation scheme that defines more signal states. Quadrature Phase Shift Keying (QPSK) fixes four signal states and thus represents two bits by one symbol. The modulation of a carrier with QPSK is demonstrated in Figure 6. The four symbols “00”, “01”, “11”, and “10” are assigned to the carrier phases 45°, 135°, 225°, and 315°. The number of bits per symbol can be further increased by changing the signal’s amplitude in addition, which is called Quadrature Amplitude Modulation (QAM). WiMAX supports 16, 64 and 256-ary QAM (16-QAM, 64QAM, 256-QAM), which represent 4, 6 and 8 bits by one symbol. Figure 7 shows the signal states of 64-QAM. Of particular concern is the symbol rate, which denotes the number of symbols transmitted per second. The symbol rate is an important measure for the bandwidth the signal adopts in the frequency domain. The higher the symbol rate, the more bandwidth is required and vice versa. The data rate is the product of symbol rate and bits carried per symbol. For increasing the data rate, either the symbol rate must be increased or the number of bits per symbol must be increased by using another modulation scheme. The former spreads the bandwidth of the radio channel, while the latter makes the signal more susceptible to interferences. This is due to the fact that with an increasing number of symbols the signal states need to be spaced closer and closer together, and hence even small interferences during the propagation may result in misinterpretations of the incoming signal at the receiver.

Unmodulated Carrier
Symbol duration T

Q

10

00

11

01

I

00 (45° shift)

10 (135° shift)

11 (225° shift)

01 (315° shift)

Figure 6. Modulation of a carrier with QPSK
5

Q 100000 100010 101010 101000 001000 001010 000010 000000

100001 100011

101011 101001

001001 001011

000011 000001

100101 100111

101111 101101

001101 001111

000111 000101

100100 100110

101110 101100

001100 001110

000110 000100

110100 110110

111110 111100

011100 011110

010110 010100

I

110101 110111

111111 111101

011101 011111

010111 010101

110001 110011

111011 111101

011001 011011

010011 010001

110000 110010

111010 111000

011000 011010

010010 010000

Figure 7. 64-QAM signal states Single carrier modulation is used for Fixed and Nomadic WiMAX and is part of the physical layers WirelessMANSC and WirelessMAN-SCa. As mentioned before, Fixed WiMAX operates in the frequency ranges between 10 and 66 GHz, and thus it is only suitable for LoS transmission. A radio channel has a bandwidth of 20, 25 or 28 MHz, and the supported modulation schemes are QPSK, 16-QAM and 64-QAM, which can be deployed depending on the error characteristics of the radio link and the desired data rates. For Nomadic WiMAX, single carrier modulation is only optional. Nomadic WiMAX focuses on NLoS transmission and has therefore been developed for operation in frequency ranges between 2 and 11 GHz. The channel bandwidth is scalable and may vary between 1,75 and 20 MHz. Single carrier modulation works similar as in Fixed WiMAX, but has been extended by BPSK and 256-QAM. Multi Carrier Modulation There are multiple error sources a radio signal is exposed to during transmission, for example, multipath propagation, attenuation, noise, shadowing by buildings and, in the case of Mobile WiMAX, frequency deviations, which are called Doppler shifts and which are caused by movements of the mobile subscriber station during transmission. Of particular concern in WiMAX as well as in all other wireless networks with large data rates and long transmission ranges is multipath propagation. As depicted in Figure 8, this phenomenon arises if a signal is reflected, scattered and diffracted from and at obstacles like buildings, trees or hills. As a result, the signal is copied during transmission, and the receiver not only receives the primary impulse of a signal, but also several delayed secondary impulses of it as shown in Figure 9a. The travelling time of a signal impulse corresponds to the length of the path at which it propagates from the transmitter to the receiver. The delay between arrival of a signal’s primary impulse and the arrival of its last secondary impulse is called delay spread, and its size significantly depends on the range of the transmitter and the density of obstacles in the close surrounding. The longer the ranges and the higher the density of obstacles at which the signal is reflected and scattered, the larger is the delay
6

spread. Multipath propagation may cause heavy interferences if the symbol duration T used during transmission is smaller than the delay spread. The symbol duration denotes the length of time a single symbol is transmitted, and thus it corresponds to the reciprocal of the symbol rate. As depicted in Figure 9b, the delayed secondary impulses of a symbol may destruct the impulses of subsequent impulses if the symbol duration is much smaller than the delay spread. This phenomenon is called intersymbol interference and is one of the main sources for transmission errors. Intersymbol interference does not represent a serious problem for Fixed WiMAX, as these networks operate above 10 GHz, where effects of multipath propagation hardly appear and where radio signals are increasingly radiated in a directional fashion from the emitting antenna. As a consequence, the antennas of subscriber and base stations must be adjusted for LoS transmission, and a significant delay spread does not occur. However, one of the main motivations behind the development of Nomadic and Mobile WiMAX was to enable NLoS transmission. As consequence, radio signals in these systems are reflected and scattered for several times until they reach the receiver. In order to cope with the resulting intersymbol interferences, Nomadic and Mobile WiMAX apply a technique known as multi carrier modulation. As suggested by its name, in multi carrier modulation a single radio channel of a certain bandwidth is subdivided into N sub-carriers, and the data stream to be sent is distributed over these subcarriers. The total symbol rate of the radio channel remains the same, but because each sub-carrier transmits only the N-th part of the entire data, the symbol duration at each sub-carrier is N times larger compared to the symbol duration of a conventional single carrier modulation. Accordingly, each sub-carrier occupies the N-th part of bandwidth of the entire radio channel. Following this approach, intersymbol interferences are avoided, because the symbol duration on each sub-carrier is larger than the expected delay spread, assuming N is chosen sufficiently large However, multi carrier modulation may suffer from socalled side lobes in the frequency domain, which result from out-of-band radiation in the frequency bands below and

Figure 8. Multipath propagation
WiMAX — Worldwide Interoperability for Microwave Access

Symbol duration T Delay spread Power Power

Symbol duration T Delay spread

Secondary impulses Primary impulse of symbol n

Secondary impulses Time Primary impulse of symbol n+1 Primary impulse of Primary impulse of Primary impulse of symbol n symbol n+1 symbol n+2

Time

(a) Delay spread without intersymbol interferences

(b) Delay spread with intersymbol interference

Figure 9. Delay spread and intersymbol interference above each sub-carrier. These side lobes do not carry any useful information that is needed for interpreting the incoming signal at the receiver, but they can distort the transmission in neighbouring sub-carriers. An important concern when using multi carrier modulation is therefore to select an appropriate frequency space between the sub-carriers. For this purpose, the sub-carriers are placed orthogonal to each other in the frequency domain, a technique that is called Orthogonal Frequency Division Multiplexing (OFDM). A pair of sub-carriers is said to be orthogonal if the frequency space between them is given by 1/Ts Hz, where Ts represents the symbol duration on each sub-carrier. As depicted in Figure 10, the advantage of orthogonality is that the peak of a subcarrier’s main lobe corresponds to the zero crossings of the neighbouring sub-carriers. In this way, out-of-band radiation in the side lobes neutralize each other, and the transmission in a sub-carrier have no negative impacts on its neighbouring sub-carriers. Furthermore, OFDM allows the overlapping of the main lobes of neighbouring sub-carriers, and hence they can be arranged very close together, which is very bandwidth efficient when compared to a non-orthogonal multi carrier modulation. In WiMAX, OFDM has been extended with a feature called sub-channelization, see Figure 11. The OFDM radio channel is subdivided into several sub-channels, and each sub-channel, in turn, is composed of several sub-carriers. Instead of using all sub-carriers the radio channel consists of, a transmitter may send on only one or several selected sub-channels. In this way, multiple users can share the same OFDM channel simultaneously. Therefore, sub-channelization in OFDM is basically a multiple access scheme, and therefore this variant of OFDM is called Orthogonal Frequency Division Multiple Access (OFDMA). OFDMA
fn fn+1 fn+2 fn+3 fn+4
Main lobes Side lobes

has also advantages regarding power control and battery consumption. For example, base stations can increase the transmit power on sub-channels serving indoor subscriber stations, and decrease it for subscriber stations staying outdoors or in the close surrounding of the base station. Subscriber stations, on the other hand, may concentrate transmit power in a few sub-carriers by OFDMA, thereby saving valuable battery resources, which is especially of advantage for small, mobile devices with integrated subscriber station as intended for Mobile WiMAX. The physical layers envisaged for Nomadic and Mobile WiMAX incorporate different variants of OFDM and OFDMA respectively. In WirelessMAN-OFDM, the radio channel is subdivided into 256 sub-carriers, each of which can be modulated with QPSK, 16-QAM or 64-QAM. The channel can adopt different bandwidths between 1,75 and 20 MHz. From the 256 sub-carriers, only 192 carry user data. The remaining ones are needed for frequency synchronization (pilot sub-carriers) or as guard bands (NULL sub-carriers) for avoiding neighbour channel interferences that result from side lobes of adjacent radio channels. Sub-channelization is only applied on an optional basis for transmissions in the uplink. WirelessMAN-OFDMA, on the other hand, subdivides the radio channel into 2048 sub-carriers. Thus, the symbol duration on each sub-carrier is much longer here than in WirelessMAN-OFDM, and hence the signals are less susceptible to intersymbol interferences. In contrast to WirelessMAN-OFDM, sub-channelization is mandatory for both directions. It can be used in different configurations that differ from each other regarding the fragmentation of the OFDM radio channel into sub-channels. Mobile WiMAX adopts the WirelessMAN-OFDMA physical layer, but introduces a new feature that is called ScalOFDM channel SubSubSubSubchannel 1 channel 2 channel 3 channel 4

Guard bands

Guard bands

1/T

Frequency

Figure 10. OFDM
WiMAX — Worldwide Interoperability for Microwave Access

Figure 11. OFDMA
7

able-OFDMA (SOFDMA). While in Nomadic WiMAX, the number of sub-carriers remains constant irrespective of the channel bandwidth, which can vary between 1,25 and 20 MHz, in SOFDMA the number of sub-carriers is scaled in dependence on the channel bandwidth. As a result, the spacing between sub-carriers and the symbol durations remain constant for varying channel bandwidths, which reduces the system complexity needed for smaller channels and improves the performance of wider ones.

4.2 Antennas
Besides multi carrier modulation, another key factor for making the transmission more robust and for achieving high data rates is the choice of an appropriate antenna technology. Most wireless systems today follow a single-antenna approach, where each base station is connected to a single antenna, which either radiates power in all directions equally (omnidirectional antenna) or which concentrates power in a beam of a certain direction and width (directional antenna) for serving only the sector of a radio cell. For WiMAX, the usage of intelligent multiple-antenna architectures is envisaged, where base stations and subscriber stations are equipped with several highly directional antennas (arranged in a so-called multi-antenna array), each of it connected to a dedicated transmitter and receiver respectively. The antennas can be dynamically adjusted to radiate power in a certain direction under consideration of the subscribers’ positions within the coverage area and the current conditions of multipath propagation. Because the power is concentrated into a beam of small width, the coverage area can be increased and interferences eliminated. Furthermore, the different transmitters connected to a multi-antenna array can independently transmit different data streams on the same radio channel, assuming that their beams are sufficiently separated in space. This technique is known as Space Division Multiplexing (SDM) and increases the capacity within a radio cell linearly with the number of antennas deployed. WiMAX incorporates two different multiple-antenna technologies, which are called Adaptive Antenna System (AAP) and Multiple Input Multiple Output (MIMO). Both of them are compared in Figure 12. The former technique is based

on beamforming and generates a beam that is directed towards a subscriber or a group of subscribers staying close by. MIMO, on the other hand, utilizes the effects of multipath propagation and is the preferred choice in cluttered environments. Signals from the different antennas are radiated in a way that they travel at different paths from the sender to the receiver. The different paths may either carry the same, redundant copies of the data stream or they might be used to transfer different data streams to the receiver. The former approach makes the transmission more robust, because interferences on a certain path may be compensated by the transmissions received from another path, or transmissions from different paths may be combined at the receiver to get a useful signal. This option is the preferred choice for serving mobile subscribers, which suffer from rapidly changing radio conditions. The transfer of different data streams, on the other hand, increases the capacity, but is less robust. It is primarily intended for fixed and nomadic subscribers.

4.3 Channel Coding
The goal of channel coding is to prepare the data stream to be transmitted in a manner that errors that may occur during transmission can be reliably detected and corrected at the receiver. This is accomplished by calculating redundant data from the data stream. WiMAX applies different error coding schemes, and their deployment and parameters depend on the physical layer used. In general, it can be distinguished between block and convolutional codes. Block coding subdivides the data stream into blocks of n bits, and generates a parity word for each block that is attached to it, resulting in a block of size m bits (m>n) that is then further processed. The type of block code used in WiMAX is called Reed-Solomon code. Convolutional coding takes n bits from a continuous input data stream and maps them onto m bits of an output stream. The generation of output bits is realized by combining (convolving) the outputs of several linear feedback shift registers in a certain manner. The quality of error coding can be measured by the maximum number of errors that can be corrected in a data block or stream of fixed size and whether error bursts or only single-bit errors can be corrected. These capabilities mainly depend on the algorithms used for error correction as well as on the code rate r=n/m, which expresses the number of output bits per input bit. The lower the code rate is, the higher is the probability that errors can be corrected, but the lower is the net data rate that can be achieved at the radio channel. Therefore, the WiMAX standards envisage to dynamically fix an appropriate code rate under consideration of the expected degree of interferences. Error correction mechanisms reliably detect and correct errors. Unfortunately, each radio transmission suffers from the appearance of error bursts, which are characterized by a large number of errors occurring in consecutive bits. Because it is difficult or even impossible in many cases to correct such errors, the output bits generated by error coding can be mixed prior to transmission, a process that is known as interleaving. For this purpose, the data stream is subdiWiMAX — Worldwide Interoperability for Microwave Access

(a) Adaptive Antenna System

(b) Multiple Input Muliple Output

Figure 12. Comparison of AAS and MIMO
8

vided into code words of fixed length, and the consecutive bits of a code word are exchanged with the bits of previous and subsequent words according to a certain algorithm. At the receiver, the original bit sequence is then reassembled by a de-interleaving process. Thus, error bursts occurring during transmission are distributed over several code words, that is, they are subdivided into single bit errors that can be corrected in most cases. As stated before, WiMAX supports different options for error coding. Figure 13 depicts a two-step error coding process that applies both block and convolutional coding. In the first step, which is also referred to as outer coding, the transmitter encodes the data stream with a Reed-Solomon code. The resulting blocks together with their parity words are then interleaved. In order to improve robustness, a convolutional coding process is then applied in the last step, which is also called inner coding. The decoding steps at the receiver are then executed in reverse order.

(a) Frequency Divsion Duplexing (FDD)
Downlink frame n xx MHz BC SS#1 (VD) SS#2 (HD) SS#3 (HD) BC Downlink frame n+1 SS#1 (VD) SS#2 (HD) SS#3 (HD)

yy MHz

SS #1 (VD)

SS #3 (VD) Uplink frame n

SS #2 (VD)

SS #1 (VD)

SS #3 (VD) Uplink frame n+1

SS #2 (VD)

(b) Time Divsion Duplexing (TDD)
Frame n Downlink subframe n xx MHz BC SS#1 (VD) SS#2 (HD) SS#3 (HD) Uplink subframe n SS #1 (VD) SS #3 (VD) SS #2 (VD)

Figure 14. Channel coding in WiMAX Figure 14b demonstrates full and half duplex modes for different subscriber stations. SS #1 is a full duplex station, and can thus send and receive simultaneously. SS #2 and #3, on the other hand, are half duplex stations. Their uplink and downlink bursts must be arranged in a way that they do not overlap. It must also be considered that their uplink bursts are not in conflict with broadcast transmissions from the base station. Using TDD, downlink and uplink share a common radio channel and are separated in the time domain as demonstrated in Figure 14b. A transmission frame is subdivided into downlink and uplink subframes, each of it consisting again of a number of data bursts assigned to different subscriber stations. A challenge of TDD is to avoid an overlapping between downlink and uplink subframes. The overlapping may result from the fact that different subscriber stations are located at different distances to the base station, and hence do not receive the end of a downlink subframe simultaneously. Therefore, uplink and downlink subframes must be separated by guard periods during which no transmission is allowed. Both FDD and TDD are available for all physical layers of WiMAX. FDD is the preferred solution for regulated operation in licensed frequency bands, while TDD is primarily deployed in unlicensed bands, which require less regulatory and organizational constraints. The following section gives a more detailed overview of the structure of transmission frames used for FDD and TDD.

4.4 Duplexing
Another task of the physical layer is the separation of uplink and downlink transmissions, which is commonly referred to as duplexing. Two fundamental approaches exist, which are called Frequency Division Duplex (FDD) and Time Division Duplex (TDD) and which are both supported in all WiMAX variants. In FDD mode, uplink and downlink are separated in the frequency domain, that is, there exists a dedicated radio channel for each direction, which is demonstrated in Figure 14a. Both uplink and downlink channels are subdivided into frames of a certain duration, and each frame, in turn, consists of several data bursts. Each subscriber station is assigned two data bursts, one on the downlink channel for receiving data from the base station and another on the uplink channel for transferring data to the base station. In addition, there is a dedicated data burst for broadcast transmissions, which is used by the base station to supply all subscriber stations with control information. This will be explained in subsequent sections. FDD can be operated in full duplex (FD) or half duplex (HD) mode. In full duplex, the stations can send and receive simultaneously. However, a major problem of FDD is that the transmission power of an outgoing signal is much higher than the received power of an incoming signal, and therefore the side lobes of the outgoing signal may drown out the incoming signal. To cope with this problem, it is recommended to arrange uplink and downlink far away from each other in the frequency domain. Nevertheless, there often remain interferences, which can only be avoided by using frequency filters, which, however, makes mobile devices complex and expensive. Another solution is therefore to use half duplex, where subscriber stations do not receive and transmit at the same time.
Block coding (Outer coding) Interleaving Convolutional Coding (Inner coding)

4.5 Frame Format
The physical layers of WiMAX come along with different frame formats, which, however, only slightly differ from each other. Therefore, only their common elements are explained here. Figure 15 shows a simplified version of the frame structure as used for the TDD mode. A frame consists of a downlink and uplink subframe and lasts 5, 10 or 20 ms. Consecutive downlink and uplink subframes are separated by guard periods (as explained in the last section), which are called Transmit/Receive Transition Gap (TTG) and Receive/Transmit Transition Gap (RTG) and during which no data transmission is allowed. For FDD, basically the same format is used: the downlink and uplink subframes shown in Figure 15 are assigned to different radio channels for
9

Figure 13. Channel coding in WiMAX
WiMAX — Worldwide Interoperability for Microwave Access

Frame n-1 Downlink sub-frame DL DL Burst Burst #1 #2

Frame n

Frame n+1 Uplink sub-frame

Time

DL Burst #n

UL UL Burst Burst #1 #2

UL Burst #n

Broad- MAC cast PDUs

MAC PDU #1

MAC PDU #n

DL MAP

UL MAP

DCD

UCD

MAC MAC CRC Header Payload

Figure 15. TDD frame format parallel transmission and constitute an entire frame with a maximum length of 20 ms. Furthermore, there is no need in the FDD mode to separate consecutive frames by TTG and RTG respectively. The following descriptions refer to both TDD and FDD. Because the WiMAX physical layers provide several options, for example, regarding modulation schemes or error coding rates, it is necessary to inform all subscriber stations in a radio cell about the configuration of the radio channel. For this purpose, the serving base station broadcasts control information at the beginning of each frame, which is received by all subscriber stations connected to that base station. The broadcast is constituted by a preamble, a so-called frame control header (FCH) and the first data burst, see Figure 15. Modulation and coding of these fields are standardized in order to make them interpretable for all subscriber stations being in the process of network entry, which will be explained below. The preamble indicates the beginning of a frame and enables the synchronization of subscriber stations to the transmissions of the base station. It always has a length of two OFDM symbols of a fixed radio pattern and is modulated with QPSK. The preamble is followed by the FCH field, which carries the so-called burst profile of the first downlink burst. This burst profile indicates the modulation scheme and code rate used in the first burst. The FCH field consists of only one OFDM symbol and is modulated with BPSK. The first burst then carries a so-called broadcast control field, which is composed of further fields denoted as DLMAP, UL-MAP, Downlink Channel Descriptor (DCD) and Uplink Channel Descriptor (UCD). DL-MAP and UL-MAP indicate the positions of all downlink and uplink bursts within the corresponding subframes as well as their burst profiles. DCD and UCD are complex descriptions of the configuration of downlink and uplink, and contain information like the identifier of the serving base station, the frame length, the length of various fields within a frame, the frame number and information for power adjustment and initial ranging, to name only a few. Each subscriber station that wants to get access to a base station has to receive these descriptions and adjust to it. An uplink subframe starts with two fields denoted as initial ranging and bandwidth request. The former is accessed by subscriber stations in order to determine the range to the base station. This process is performed during the net10

work entry and will be described below. Using the bandwidth request field, a subscriber station can announce its bandwidth requirements to the base station, which will also be explained later. The broadcast fields (in the downlink) and the fields for initial ranging and bandwidth request (in the uplink) are followed by data bursts for individual transmissions to and from subscriber stations. A data burst is of variable length and carriers the protocol data units of the medium access layer (MAC PDU). The assignment of data bursts to subscriber stations is part of the medium access layer and is executed by the base station under consideration of QoS requirements. For each data burst, another configuration of modulation scheme and error coding rate can be used, which is specified in the DL-MAP and UL-MAP fields of the frame. The configuration can be dynamically selected under consideration of the capabilities of the subscriber station, the required data rates and the expected robustness of transmission. For example, subscriber stations located close-by to the base station may be served by 64-QAM, which provides high data rates but which is very susceptible to interferences, while for subscriber stations located farther away the more robust QPSK modulation may be preferred. This is demonstrated for the downlink in Figure 16. However, the usage of different modulation schemes imposes certain constraints regarding the ordering of bursts within a frame. A particular concern is that a subscriber station that wants to transmit in (or receive) a burst has to detect the end of the previous burst assigned to another subscriber station. This can only be guaranteed if for the previous burst either the same modulation scheme is used or another one that is more robust against interferences. In Figure 15, SS#1 is located farthest away from the base station and is served with a QPSK modulation in the first burst. SS#2, on the other hand, is closer by and receives in the second burst modulated with the less robust 16-QAM. It can easily detect the end of the first burst, because QPSK is more robust than 16-QAM. If, however, the order of the two bursts would be exchanged, SS#1 could hardly detect the end of the first burst, because it is located out of the range where 16-QAM modulated signals can be reliably received. Therefore, the bursts within a frame must always be arranged in decreasing order with
Decreasing robustness of modulation

TTG Initial ranging Bandwidth request

Preamble

FCH

RTG

FCH

Downlink subframe

Preamble

SS #1 SS#2 SS#3 SS#4 (QPSK) (16-QAM) (16-QAM) (64-QAM)

BS

SS#2 SS#4 SS#1 SS#3

Increasing interferences

Figure 16. Modulation of data bursts
WiMAX — Worldwide Interoperability for Microwave Access

DL sub-frame

UL sub-frame

DL Burst #2

UL Burst #1

DL Burst #3

UL Burst #2

DL Burst #5

DL Burst #1

DL Burst #4

BW IR UL Burst #1

Time domain - OFDMA symbols

TTG

RTG

Figure 17. Example of frame structure when using OFDMA regard to the robustness of the used modulation schemes. Finally, Figure 17 shows a possible appearance of downlink and uplink frames for the case that OFDMA is used. The different data bursts are not only separated in the time domain here. They can also be transmitted simultaneously assuming that they adopt different sub-channels, which are composed of the several sub-carriers built by the multi carrier modulation.

tection are activated as well as the length of the entire PDU. The bandwidth request header additionally contains the number of bytes the subscriber station intends to transmit in the uplink. The payload field carries the actual user data as well as control and management information and is of variable length. For example, the payload field may carry IP data packets, which are filled into the payload field by the convergence sub-layer. Finally, the Cyclic Redundancy Check (CRC) field contains a checksum that the transmitter calculates from the header and payload fields. The term CRC denotes a special mechanism of error detection, where the checksum is given by the remainder of a polynomial division. The checksum is analyzed by the receiver in order to detect those errors that could not be corrected by the channel decoding process of the physical layer (see also Section 4.3).

Frequency domain - OFDMA sub-carriers

DL-MAP

FCH

DL Burst #5 (Multicast/Broadcast burst)

Preamble

DL-MAP

UL-MAP

5.2 Service Flows and MAC Connections
The medium access layer of WiMAX organizes the exchange of data between subscriber and base station by the concept of service flows. A service flow is always unidirectional, that is, it is defined either for uplink or downlink direction. It is represented by a unique Service Flow Identifier (SFID) and characterized by a set of QoS parameters, for example, data rate, latency and jitter. The requirements of different applications on these parameters are very heterogeneous. For example, VoIP without silence suppression demands for a constant bit rate and a guaranteed maximum latency and jitter, while a simple file transfer only requires a minimum data rate, but no guarantees regarding other QoS parameters. Each service flow is realized by a MAC connection, which is referenced by a Connection Identifier (CID) and which is constituted by a series of data bursts allocated by the base station in the different transmission frames. This allocation has to be organized in a way that the QoS requirements of the service flow the connection carries are fulfilled. This process represents the core mechanism of medium access. It is called scheduling and is based on sophisticated algorithms. The allocation of data bursts has to be considered for downlink and uplink direction differently. For the downlink, the allocation is comparatively simple, because the base station is the only sender in this direction. The data of an external network, for example, the Internet, arrives at the base station and is there assigned to the service flow that is maintained between the base station and the subscriber station the data is intended for. The scheduling algorithm of the base station then identifies one or several bursts within one or several frames for data transmission. In the uplink, the medium access is much more complicated, because it has to be coordinated among all subscriber stations within a cell. In classical mobile networks, for example, GSM, the problem of assigning transmission capacity to mobile stations is often solved by reserving a burst of fixed length in each frame and for each active station. In other wireless system, for example, WLAN, access to the radio channel is not centrally coordinated. Instead, the stations enter the channel whenever they have data to
11

5 WiMAX Medium Access Layer
If a base station operates in the point-to-multipoint mode (see Section 2.2), subscriber stations located within its coverage area compete against each other for access to the radio channel. This access is coordinated by the base station and belongs to the main tasks of the medium access layer. The primary focus of medium access on the one hand is to avoid collisions, which would occur if two or more subscriber stations would enter the same radio channel (or some of its sub-carriers if OFDMA is applied) simultaneously and, on the other, to guarantee the access in a way that QoS requirements are met. Besides this, the medium access layer also provides related functions, for example, authentication and ciphering as well as error correction and radio link control. The following sub-sections provide a short overview of the most important procedures of medium access.

5.1 MAC Protocol Data Units
Data is transferred via protocol data units (PDUs) of the MAC layer, which, in turn, are included into the data bursts provided by the physical layer. There may be several MAC PDUs per data burst. The PDUs carry user data, control and management information as well as bandwidth requests issued by the subscriber stations to announce their bandwidth requirements for uplink transmission. Apart from the bandwidth request, which only consists of a single header, a PDU contains a header field, a payload field and another field for error detection, see also Figure 15. The header is of fixed length and carries control information, for example, the identifier of the connection (see description below), whether or not encryption and error deWiMAX — Worldwide Interoperability for Microwave Access

Polling
[UL-MAP]

Bandwidth request
[Bandwidth request field]

Bandwidth grant
[UL-MAP]

UL transmission
[assigned data burst]

Figure 18. Polling send. Collisions between the transmissions of different stations are avoided in that the channel has to be sensed free prior to transmission. The former approach is suitable for the adherence of QoS guarantees, but it suffers from an inefficient utilization of the channel if the stations do not use the full capacity of a burst, for example, during periods of silence in a VoIP session. The latter approach, on the other hand, performs much better with regard to channel utilization, but it is not suitable for providing a negotiated QoS, for example, when too many stations contend for channel access. Therefore, in order to cope with the antagonism of efficiency and quality, WiMAX provides different access mechanisms for the uplink that can be dynamically deployed under consideration of QoS requirements. One of these mechanisms is polling, which is depicted in Figure 18. The base station here explicitly invites a subscriber station to announce its uplink bandwidth demand for a particular connection. The polling request specifies the CID of the connection the polling refers to, and it is encoded as a special element of the uplink map. Upon arrival of a poll, the subscriber station determines the number of bytes it wants to transmit in the uplink and returns this number to the base station by sending a bandwidth request header (see Section 5.1). Besides the byte number, this header also specifies the connection the request refers to and whether the bandwidth request is incremental or aggregated. Using an incremental request, the subscriber station indicates a change of bandwidth demand with regard to previous requests, while an aggregated request specifies the total amount of bytes that needs to be sent. The bandwidth request header is included in the bandwidth request field of the uplink frame, see Figure 15. After the base station has received the bandwidth request header, it reserves a burst of appropriate size for the next uplink frame. The parameters for uplink transmission, for example, the burst number and length, are then indicated to the subscriber station in the next UL-MAP sent on the downlink. The polling of a subscriber station may be performed regularly or irregularly, which depends on the base station’s scheduling algorithm and the QoS parameters that have been negotiated for the respective service flow. Furthermore, it is distinguished between unicast and multicast/ broadcast polling. In the former category, the polling refers to only a single subscriber station, while multicast/broadcast polling addresses several or all subscriber stations located in a cell.
12

Another mechanism for requesting bandwidth is piggybacking. The name is derived from the fact that bandwidth requests are piggybacked (or attached) to the regular uplink transmissions of a subscriber station, instead of sending a dedicated bandwidth requests header. Piggybacking is performed independently from polling, that is, a subscriber station does not have to wait until it is polled, but can immediately inform the base station about changing bandwidth demands if necessary. The bandwidth request is included into the header of a conventional MAC PDU and always refers to the connection this PDU is part of. Finally, a subscriber station can get assigned uplink bursts of fixed length at regular intervals without the need to explicitly request them. This mechanism is called unsolicited scheduling and is the preferred choice for applications that require a constant bit rate during the entire session. The reservation may hold for the entire duration of the service session, but it may be temporarily cancelled in the case of inactive time periods. Furthermore, a subscriber station can indicate additional bandwidth demand for an unsolicited connection if it turns out that the amount of unsent data exceeds a pre-defined value. In this case, the subscriber station sets a so-called slip indicator bit in the MAC PDU header, whereupon the base station allocates more bandwidth for the respective connection.

5.3 Service Classes
The different mechanisms of bandwidth request presented previously are used for realizing different service classes, which differ from each other in the QoS they provide. These service classes are basically descriptions of service flows with a pre-configured set of QoS parameters and are supported by associated scheduling algorithms in the base station. The following services classes have been defined for WiMAX: • Unsolicited Grant Service (UGS). This service class has the strongest requirements on QoS mechanisms and has been designed for supporting real-time applications of constant bit rate, that is, for applications that periodically create a certain amount of data for realtime transfer over the network. A typical example is VoIP without silence suppression, where both periods of conversation and silence are encoded and transferred with a constant bit rate. The bandwidth request mechanism in the uplink used for this service class is unsolicited scheduling. • Real-time Polling Service (rtPS). Another real-time service class is the Real-time Polling Service, which in contrast to UGS supports the periodic transfer of data packets of variable size. A typical example is MPEGcompressed video, where the single frames of a video stream are encoded depending on the data of previous and following frames and therefore differ in size. This service class is based on the polling and piggybacking mechanisms for bandwidth request.

WiMAX — Worldwide Interoperability for Microwave Access

Subscriber station

Base station

rameters for a service flow, which, however, increases the complexity of configuration.

5.4 Procedures of the MAC Layer
Network entry
Regular DL transmissions
1 Downlink channel

synchronization
2 [Initial ranging field]

This section gives an overview of the procedures taking place between subscriber station and base station in order to register with the network and to get assigned resources. The different steps are depicted in Figure 19. Network Entry Before a subscriber station can use any services, it must first introduce to the base station, a process that is known as network entry. An important goal when designing WiMAX was to avoid complex and cumbersome manual configurations to be made by the subscriber, as often required, for example, in order to get access to a WLAN system. Instead, the subscriber should enter into contact with a WiMAX network in a plug-and-play fashion, that is, in a similar manner as mobile phones register with a GSM network, for example. Therefore, the different steps of network entry are hidden from the subscriber as far as possible, and may only require a pre-configuration of subscriber stations by the respective operator (to be made before they are delivered to the subscriber). The first step a subscriber station has to perform for network entry is called downlink channel synchronization (see Step (1) in Figure 19). The subscriber station scans the frequency range for detecting the downlink channel of a base station and then listens to the preamble periodically broadcast in each downlink frame. If the subscriber station is synchronized, it derives information about the organization of uplink and downlink, that is, about the type of physical layer and the used modulation and error correction schemes, from the broadcast control field of the first burst. In the next step, which is denoted as initial ranging, the range between subscriber and base station is determined in order to fix a suitable transmission power and timing corrections. For this purpose, the subscriber station enters the initial ranging field of an uplink frame and sends a ranging request message (2) with the minimum transmission power. If no response is received from the base station within a certain timeout period, this message is resent with an increased transmission power. This process is repeated until the subscriber station receives a ranging response message (3), which either contains corrections for transmission power and timing or which indicates success. The last step of network entry is capability negotiation, and it is used to inform the base station about the modulation schemes, error correction schemes and rates as well as duplexing methods supported by the subscriber station. Upon arrival and checking of the capability request message from the subscriber station (4), the base station can accept or deny network entry in a capability response message (5). Authentication and Key Exchange After network entry is completed, the subscriber station must authenticate towards the network, which is necessary in or13

Ranging request

Ranging response 3
4 Capability request

Capability response 5

Authentication and key exchange
6 Authentication request

Authentication response 7

Registration
8 Registration request

Registration response 9
10

DHCP/Internet Time Protocol/TFTP

Connection Setup
11 Dynamic service addition request

DS received 12
13

Authorization

Dynamic service addition response 14
15 Dynamic service addition ack.

Figure 19. MAC procedures

• Non-real-time Polling Service (nrtPS). This service class supports typical non-real-time applications such as file transfer or Internet browsing. This service class does not necessarily need periodic transmission opportunities or a guaranteed end-to-end latency. However, in order to provide an acceptable QoS, the base station issues unicast polls on a regular basis. • Best Effort Service (BE). Services of this class do not receive any QoS guarantees at all. They are only served if sufficient capacity is available. Examples for such lowpriority applications are those which create a low amount of data that may be delivered with a considerable delay, that is, Email, instant messaging or chat applications. If required, it is possible to modify the QoS parameters of a services class or to determine its own set of QoS paWiMAX — Worldwide Interoperability for Microwave Access

der to validate the subscriber’s identity. The authentication is based on X.509 certificates, which are issued by the manufacturer of the subscriber station and which are encrypted with the subscriber’s secret key. The certificate is passed to the network (6) and is decrypted there with the subscriber’s public key. If the validation is successful, the base station sends an authentication response (7), which contains an authorization key for ciphering subsequent messages. This authorization key is encrypted with the subscriber’s public key and can only be decrypted with her secret key at the subscriber station. Registration and IP Connectivity The subscriber station can now register with the network and be configured for IP operation. Upon sending a registration request message (8), it receives information about the used IP version, supported protocols for retransmission of erroneous data (Automatic Repeat Request, ARQ) and other capabilities needed for medium access (9). Finally, further operations are executed for IP connectivity (10), among them the allocation of an IP address by using the Dynamic Host Configuration Protocol (DHCP), the exchange of current date and time via the Internet Time Protocol and the download of operational parameters by using the Trivial File Transfer Protocol (TFTP). Connection Setup After a subscriber station is known to the network, service flows can be established in both directions, for which a series of management messages is exchanged. The service flows may be initiated by the subscriber station or by the base station. In the former case, which is shown in Figure 19, the subscriber station sends a request message to the base station (11), which addresses the convergence sublayer the service flow refers to (that is, IP or ATM) and which contains the desired QoS parameters. After reception of this message, the base station acknowledges its reception (12) and checks whether the requesting subscriber is allowed at all to request service flows with the specified QoS configuration (13). If this check is successful, this is indicated to the subscriber station by sending another message (14). The connection setup is completed if the subscriber station then acknowledges this message (15). In addition to this setup procedure, an existing service flow can be reconfigured (regarding its QoS parameters) or deleted, for which similar procedures exist. Also, it is possible to establish several service flows in parallel.

6.1 Handover
Basically, the handover process as performed in most mobile networks can be subdivided into the three following phases: measurements, decision and execution. In Mobile WiMAX, all of them are initialized by the subscriber station, but supported by the base stations involved in the handover procedure, that is, the serving base station and possible target base stations. As stated in Section 2.1, Mobile WiMAX supports hard as well as soft handover. The three handover phases for both types are explained in the following. Hard Handover A hard handover is characterized by the fact that the connection to the serving base station is released before another one is established to the new base station ("breakbefore-make"). Measurements are made by the subscriber station and refer to observing the signal-to-noise ratio (SNR) of downlink transmissions from the serving base station as well as from possible target base stations. The SNR expresses the ratio between the reception power of the intended signal and that of other interferences. If the SNR of the serving base station gets low, the error rates increase, and a handover to another base station should be performed. In order to measure the SNR of possible target base stations, the measuring subscriber station must be aware of their existence and the configuration of the associated radio channels. Therefore, the serving base station periodically
Subscriber station Serving Target #1 Target #2 base station

Network topology advertisement

1

2

Scanning request Scanning response 3

4 Scanning interval

Regular downlink transmissions Regular downlink transmissions

6 Mobility Support
This section gives an overview of functions for mobility support in Mobile WiMAX systems. The focus is on the different types of handover, modes for power saving, procedures of location management and a reference model for a WiMAX network architecture.
8

5

Handover request Capacity & QoS checks Handover response 7
6

9 Network entry, registration, ...

Figure 20. Hard handover
14 WiMAX — Worldwide Interoperability for Microwave Access

broadcasts a so-called network-topology-advertisement message, which contains a list of all neighbouring base stations together with their configuration, see Step (1) in Figure 20. Basically, this configuration is an aggregation of the DCD and UCD fields broadcast by the respective base stations in their downlink frames. After analyzing the network-topologyadvertisement message, the subscriber can switch between different neighbouring base stations and obtain the SNRs of their downlink transmissions. However, for listening to the transmissions of neighbouring base stations, the subscriber station must interrupt the reception of the serving base station. For this purpose, it requests the serving base station for the assignment of a so-called scanning interval (2). During this interval, transmissions to and from the requesting subscriber station are interrupted. The beginning and length of the scanning interval are returned in a scanning-response message to the subscriber station (3), whereupon it starts the scanning of neighbouring base stations (4). After the scanning is complete, the subscriber station compares the measured SNRs with the SNR of the serving base station and decides whether or not a handover is necessary. This decision process also includes the identification of potential target base stations, which are also selected under consideration of the transmission quality experienced during the scanning interval. However, this decision cannot be made by the subscriber station solely. The list of potential target base stations must first be sent to the serving base station (5), which then checks whether the identified base stations have enough capacity at all to serve the subscriber station and to maintain the QoS parameters of its service flows (6). After the results of this check are available, the serving base station returns the list of remaining target base stations or proposes new ones (7). If the subscriber station does not accept one of the chosen base stations, it returns a negative acknowledgement. This negotiation process can then be repeated for several times until a suitable target base station is determined (8). If the subscriber station does not return a negative acknowledgement within a specified time period, the serving base station acts on the assumption that the subscriber station has switched to one of the proposed target base stations and releases all connections. The subscriber station, on the other hand, registers with the new target base station

then (9) and for this purpose executes network entry, registration and all following steps as explained in Section 5.4. Soft Handover During a soft handover, the subscriber station maintains several connections to different base stations simultaneously ("make-before-break"), see Figure 21. Measurements are performed in the same manner as for the hard handover during scanning intervals, see Step (1)-(4) in Figure 20. However, handover decision and execution are handled differently. The base stations a subscriber station is connected to are managed in its active set. At the beginning, the active set only contains the base station the subscriber station has initially registered with, which is called anchor base station. The active set can be extended if the subscriber station measures an SNR from another base station that exceeds a pre-defined threshold value. If this happens, the subscriber station requests the anchor base station for updating the active set, which can be accepted or denied depending on similar capacity checks as performed for the hard handover, see Step (6) in Figure 20. Analogously, a base station can be removed from the active set if its SNR falls below another threshold value. Furthermore, if the SNR of the anchor base station is lower than that of another base station for a certain period of time, the subscriber station can request to change the anchor. Maintaining connections to several base stations simultaneously means that the subscriber station receives the same data for several times over different paths. In order to improve the reception quality, all signals are combined to an aggregated signal, which usually shows a much better SNR when compared to that of a single signal. This feature is called macro diversity and is also implemented in UMTS networks. In the uplink, the signals of the subscriber station are received by all base stations of the active set. Instead of summing up the signals, only that with the best quality is selected and further processed, which is called selective diversity. Making soft handovers possible is a difficult task and requires a careful design and planning of WiMAX networks as well as complex and sophisticated coordinations during their operation. The realization of macro and selective diversity makes it necessary that neighbouring base stations operate at the same frequencies and follow the same structure of data bursts within the transmission frames. Furthermore, in order to avoid interferences, the frames must be exactly synchronized in time, for which GPS receivers mounted at
Ready for reception/ transmission Handover Standby Handover Suspended Paging + Location Update

Active mode

Sleep mode

Idle mode

Figure 21. Multiple connections to different base stations during soft handover
WiMAX — Worldwide Interoperability for Microwave Access

Figure 22. Power saving modes

15

the base stations deliver a common time basis. Finally, there is additional management overhead, for example, the coordination of uplink and downlink maps and the assignment of common CIDs and SFIDs, to name only a few.

6.2 Power Saving
A typical problem of mobile devices is the lack of sufficient battery resources, and that is why for Mobile WiMAX new operational modes are defined, see Figure 22. These are called sleep and idle mode and consume considerably less power than the conventional active mode. A subscriber station turns from the active into the sleep mode if no data is to be sent in the various service flows it maintains with the base station. This may happen, for example, if a service flow is used for transferring web pages, and the subscriber remains on a certain web page over a longer period of time before requesting the next one. The sleep mode is characterized by alternating listening and sleep periods. In a sleep period, the subscriber station is deactivated and does neither monitor the downlink transmission frames from the base station nor does it transmit in the uplink. From time to time, however, the subscriber station changes into a listening period in order to check whether data from the network has arrived. If so, it then returns to the active mode. If data arrive from the network during the sleep periods, the base station has to buffer it until the next listening period occurs. The start and length of sleep and listening periods are negotiated between subscriber station and base station before starting into the sleep mode. In the idle mode, the subscriber station is suspended from the network, but remains available for the case that network-initiated data is to be delivered, for example, an incoming VoIP session or push email. The subscriber station does neither transmit nor receive, similar to the sleep periods in the sleep mode, and hence saves its power resources. It only awakens for listening to so-called paging intervals, during which it is informed about incoming data and other procedures of the location management, which will be described in the next section. The idle periods between two paging intervals can alternatively be used for scanning neighbouring base stations if the transmissions in the paging intervals from the serving base station get too weak. 6.3 Location Management A subscriber station being in active or sleep mode always performs a handover when it leaves the coverage area of the serving base station. However, as demonstrated in Section 6.1, performing a handover goes along with a complex sequence of control information exchanged between subscriber and base station, which significantly burdens the air interface as well the battery resources of the subscriber station. Therefore, handovers are not performed when the subscriber station is in idle mode. However, in this case it must still be available for network-initiated traffic, even if the subscriber moves around and leaves the coverage area of the serving base station. While in the other modes, the change to another base station is recognized when perform16

ing a handover, the idle mode uses classical mechanisms known from the location management of cellular networks like GSM. These mechanism enable to locate a subscriber station within the topology of a network, or, strictly speaking, to determine the access network and base station a subscriber is currently connected to. Location management in Mobile WiMAX is based on paging areas, which comprise the coverage areas of several base stations. The different paging areas may overlap, that is, a single base station may be assigned to different paging areas. In idle mode, a subscriber's location is known to the network only with the granularity of paging areas. The serving base station is not known. Therefore, if network-initiated traffic needs to be delivered, the serving base station of the target subscriber must be determined, which is done via paging. It is realized by broadcasting the subscriber station's identifier during the paging intervals of the idle mode (see last section). Paging is performed by all base stations that belong to the paging area the subscriber is registered with. Upon recognizing that it is paged, the subscriber station returns to the active mode and re-registers with its serving base station. In this way, the serving base station has been determined and the data can be routed over this base station for transmission to the subscriber. The advantage of this concept is that a handover is not necessary, and the subscriber stations only need to inform the network when crossing the boundaries of the paging area, which is denoted as location update. For this purpose, each base station broadcasts in the paging interval the identifiers of the paging areas it belongs to. If a subscriber station recognizes that it receives the identifier of another paging area than in the previous paging interval, it has obviously entered another paging area and issues a location update. This location update contains the identifier of the paging area and the subscriber station's identity, both of which are stored in the databases of the network and used for paging. In addition to the location update on crossing the boundaries of a paging area, subscriber stations issue location updates on a periodic basis or on request by the network.
Visited
Network Service Provider

Network

Home
Network Service Provider

Access
Provider R2 R2

Subscriber Station

R1

Access Service Network

R3

Connectivity Service Network

R5

Connectivity Service Network

R4

Anderes ASN
Control data User data

Application Service Provider or Internet

Application Service Provider or Internet

Figure 23. WiMAX Network Reference Model
WiMAX — Worldwide Interoperability for Microwave Access

Access Service Network
R6

R1

Base Station
R8

ASN Gateway
R4

R6 R3

R1

Base Station

R6

ASN Gateway

R4

Figure 24. Components of the Access Service Network 6.4 WiMAX System Architecture Initially, WiMAX was planned as access technology that simply connects subscriber stations to a network. However, the migration towards a system with full mobility support requires additional components for realizing and supporting mobility-related procedures like handover, location management, roaming as well as authentication, authorization, accounting (AAA). A system with standalone base stations, as it might be sufficient for Fixed or Nomadic WiMAX, is not sustainable and cannot fulfil the demands of the functions mentioned before. Instead, base stations must be connected among each other as well as to other components such as databases or gateways. The WiMAX forum has recognized these demands and released an architecture for WiMAX End-to-End Network Systems in December 2005 [4]. Figure 23 gives a rough overview of the WiMAX Network Reference Model (NRM), which is a logical representation of this architecture. It incorporates the four administrative domains of the Subscriber, Network Access Provider (NAP), Visited Network Service Provider (V-NSP) and Home Network Service Provider (H-NSP). The different domains are interconnected by reference points, which are an abstract representation of communication links, interfaces, protocols and transactions that are needed to exchange control data (represented by the dotted line in Figure 23) or user data (represented by the solid line). The NRM describes an AllIP network, that is, all reference points depicted in the figure are realized on top of the Internet protocol. This approach is very similar to what is envisaged for the next releases of UMTS networks. A subscriber station is connected to an Access Service Network (ASN) of the NAP, which consists of several base stations and Access Service Network Gateways (ASN-GW). The reference point R1 describing that connection comprises all functions of the physical and medium access layer described in the last sections. As depicted in Figure 24, the base stations of an ASN are connected among each other via reference point R6, for example, in order to exchange control information needed for handover support. The ASNGW provides typical bridging and routing functions and transfer user data between base stations and the Connectivity Service Network (CSN), which is operated by the VNSP or H-NSP, and vice versa. In addition, it controls and manages the load of base stations, which is part of the radio
WiMAX — Worldwide Interoperability for Microwave Access

resource management. The CSN of a V-NSP and H-NSP respectively provides connectivity services to other public and non-public networks, offers high-level services, for example, multimedia and location-based services, and manages subscriber data. Furthermore, it centrally coordinates the assignment of IP addresses to subscriber stations and provides mobility support for subscribers moving between different ASNs. Finally, it is also responsible for accounting and charging subscribers for the usage of services. The separation between the CSNs of H-NSP and V-NSP has been arranged to support roaming subscribers in a similar manner as known from GSM networks. Subscribers registering with a V-NSP do not need to maintain a dedicated subscription with that provider. Instead, the V-NSP accounts the H-NSP for all services the subscriber has accessed.

7 Conclusion
WiMAX belongs to the class of wireless Metropolitan Area Networks and provides access to fixed, nomadic and mobile subscribers. While Fixed and Nomadic WiMAX represent an interesting alternative to wired access technologies like DSL and cable modem, the role of Mobile WiMAX in the area of other mobile networks is unclear so far. From a today's perspective, Mobile WiMAX bridges the gap between WLAN, which realizes high data rates but almost no mobility support, and cellular networks like GSM and UMTS, which provide world-wide coverage but suffer from low data rates and capacity shortages. In the long-term, however, Mobile WiMAX and cellular systems might merge into what is today referred to as 4G networks, which are characterized by the coexistence of heterogeneous radio technologies, for example, UMTS, WLAN and WiMAX, operated under a common core network. Alternatively, WiMAX networks as defined by the NRM might be realized by a homogenous approach and then provide interoperability with GSM, UMTS or other wireless systems. This white paper could give only a rough overview of WiMAX. A detailled description can be found in [5] and [6]. References
[1] [2] WiMAX Forum, Web Site, http://www.wimaxforum.org/ IEEE Standard for Local and Metropolitan Area Networks; Part 16: Air Interface for Fixed Broadband Wireless Access Systems, IEEE 802.16-2004, New York, October 2004 IEEE Standard for Local and Metropolitan Area Networks; Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems, Amendment for Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands, IEEE 802.16e-2005, New York, February 2006 WiMAX Forum; Mobile WiMAX — Part 1: A Technical Overview and Performance Evaluation, February 2006 Johannes Maucher and Jörg Furrer; WiMAX — Der IEEE802.16-Standard: Technik, Anwendung, Potential, Heise Zeitschriften Verlag, November 2006 Loutfi Nuaymi; WiMAX: Technology for Broadband Wireless Access, John Wiley & Sons, January 2007 17

[3]

[4] [5]

[6]

http://www.samsungmobile.com/ Samsung Telecommunication Europe Samsung House • Am Kronberger Hang 6 • 65824 Schwalbach/Ts. • Deutschland/Germany