Computer Communications 33 (2010) 113–123 Contents lists available at ScienceDirect Computer Communications journal homepage: www.elsevier.com/locate/comcom IEEE 802.16: History, status and future trends A. Bacioccola a,b, C. Cicconetti a,*, C. Eklund c, L. Lenzini a, Z. Li b, E. Mingozzi a a Dip. di Ingegneria dell’Informazione, University of Pisa, Italy Nokia Devices, Finland c Nokia Siemens Networks, Finland b a r t i c l e i n f o Article history: Available online 10 November 2009 Keywords: WiMAX Broadband wireless access Orthogonal frequency division multiple access IEEE 802.16m a b s t r a c t Over the last few years, IEEE 802.16 has been established as one of the most promising solutions for broadband wireless metropolitan area networks. In 2007, the standard was included as one of the radio access technologies for IMT-2000. The forthcoming version of the standard, IEEE 802.16m, is currently under evaluation by ITU as a radio technology for IMT-Advanced. In this paper we present a historical overview of the standard and provide a detailed technical discussion of the most relevant features introduced in the 2009 release and in the upcoming IEEE 802.16m. In particular, the downlink control signaling of these two versions is investigated in detail, and the results of a numerical analysis are illustrated for comparison purposes. Ó 2009 Elsevier B.V. All rights reserved. 1. Introduction – historical overview of IEEE 802.16 and WiMAX releases In this first section we summarize the evolution of IEEE 802.16 since its first release in 2001, up to IEEE 802.16m version, expected to be completed in 2010. A good overview of the IEEE 802.16-2009 system and the current WiMAX release can be found in the papers published in the special issue on ‘‘Mobile WiMAX” of the IEEE Communications Magazine (June 2009) [22–25]. An additional status update on WiMAX was published on October 2009 in the IEEE Communication Magazine [26–28]. This work is complementary to them, since we address in details the MAC and PHY changes introduced by the new IEEE 802.16m standard, which are only sketched in the papers cited. We therefore present in Section 2 a technical overview of IEEE 802.16m, with special emphasis on those aspects of the medium access control and physical layers that will be improved with respect to the current release of IEEE 802.16-2009. The issue of the MAC signaling for downlink data transmission is analyzed separately in Section 3. Lastly, Section 4 provides the conclusions. wireless access (BWA) for fixed users, as an alternative to cabled access networks, such as a digital subscriber line (DSL) links [2]. For this reason, the original IEEE 802.16 defines a point-to-multipoint (PMP) network architecture where resources are shared by a central node called base station (BS) to a set of subscriber stations (SS). In fact, the PMP operational mode fits a typical fixed BWA scenario, where multiple subscribers are served by one centralized service provider to access external networks (e.g., the Internet) or services (e.g., digital video broadcasting – DVB). From its first release, the medium access control (MAC) layer was connection-oriented and supported quality of service (QoS) [3]. Moreover, the standard was designed to evolve as a set of air interfaces based on a common MAC protocol, but with physical layer specifications dependent on the spectrum of use and the associated regulations. The standard, as approved in 2001, addresses frequencies from 10 to 66 GHz in line-of-sight (LOS) operations using single carrier transmission only. In 2003, a new version of the standard, IEEE 802.16a-2003 [4], was published with support for non-LOS operations in frequencies from 2 to 11 GHz. 1.2. The introduction of mobility support 1.1. IEEE 802.16-2001: the origins The first version of the IEEE Standard 802.16-2001 [1], completed in October 2001 and published on 8 April 2002, defined the WirelessMAN™ air interface specification for wireless metropolitan area networks (MANs). The intention behind the first release of the standard was to define a technology for broadband * Corresponding author. Tel.: +39 0502271452. E-mail address: claudio.cicconetti@iet.unipi.it (C. Cicconetti). 0140-3664/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.comcom.2009.11.003 The subsequent milestone in standard development was IEEE 802.16-2004 [5] which introduced support for two additional physical layers: orthogonal frequency division multiplexing (OFDM) and orthogonal frequency division multiple access (OFDMA). In 2005, a new version of the standard was released to enable combined fixed and mobile operations in licensed bands. The aforementioned standard, IEEE 802.16e-2005 [6], was defined as an amendment to IEEE 802.16-2004 and added several features related to mobile operations and mobile stations (MS), including 114 A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 power-saving, idle mode, handover and an improved OFDMA physical layer. After the 2005 release, the standard development continued to define the management information base (MIB) for MAC and PHY (IEEE 802.16f [7]) and the management plane and procedures (IEEE 802.16g [8]), to improve the co-existence for licenseexempt operation (IEEE 802.16h [9]), to introduce relay capabilities (IEEE 802.16j-2009 [10]), and to refine the MAC and PHY procedures for mobile operations (IEEE 802.16-2009 [11]). The latter is also known as the 2009 release, and brings the following major changes: half-duplex mobile terminal operations in OFDMA frequency division duplexing (FDD), load balancing, robust header compression (ROHC), enhanced mechanism for resource allocation (e.g., persistent allocation), support for location based services (LBSs) and multicast and broadcast services (MBSs) [12]. The clean up for IEEE 802.16-2009 also involved incorporating the IEEE 802.16f and IEEE 802.16g amendments, and removing some stale features, such as the mesh mode. and service demands in many various environments, the IMT-Advanced systems will support low-to-high mobility applications and a wider range of data rates than IMT-2000 systems. Lastly, significant improvement in performance and quality of service (QoS) will be made, so as to enable high quality multimedia applications within a wide range of services and platforms. In 2007 the task group of IEEE 802.16 started a new project called IEEE 802.16m, described in the following sections, which will undergo the IMT-Advanced evaluation process being carried out by the ITU [16,20,21]. The IEEE announced on October 6th, 2009 to have submitted a candidate radio interface technology for IMT-Advanced standardization in the Radiocommunication Sector of the International Telecommunication Union (ITU-R). The proposal [31] documented that IEEE 802.16m met ITU-R’s challenging and stringent requirements in all four IMT-Advanced ‘‘environments”: Indoor, Microcellular, Urban, and High Speed. 1.3. IEEE 802.16 and the WiMAX Forum 1.5. The future: IEEE 802.16m The IEEE 802.16 specifications were designed with the focus on flexibility, thus leaving several parts as optional and allowing various BS and MS implementations. Moreover, the standard only deals with the MAC and physical layers, without defining the over-the-air upper layer signaling nor the overall network architecture and protocols. These two factors were the main catalyst for the establishment, in 2001, of the WiMAX Forum (http:// www.wimaxforum.org/). Since its birth, the goal of the WiMAX Forum has been to enable conformity and inter-operability of SSs and BSs based on IEEE 802.16. Following the publication of [6], in 2007 the WiMAX Forum released its first set of specifications. The Mobile WiMAX Release 1.0 is composed of several documents that together define the network reference model, the reference points, the core network elements, and all the functions and procedures to enable an all-IP end-to-end network architecture [13,14]. In addition to developing complementary technical specifications to the IEEE, the WiMAX Forum has also established certification laboratories, and manages conformance and inter-operability testing to ensure that all WiMAX-certified products across different implementations work seamlessly with one another. Since June 2008, the WiMAX Forum has been working on a new version of the Mobile WiMAX, called Release 1.5, based on the latest IEEE 802.16-2009 standard. This release is aimed at enabling mobile WiMAX in new spectrum bands, including those for FDD operation, addressing the most recent MAC improvements, and introducing advanced network capabilities. The interested reader can find details of the WiMAX evolution in [12]. The aim of IEEE 802.16m is to amend both IEEE 802.16-2009 and IEEE 802.16j-2009 standards, which specifies relay capabilities, in order to design an advanced air interface for operation in licensed bands, also providing support for legacy equipment. Since the beginning of 2007, the IEEE 802.16m technical working group has been working on the following documents: the system requirement document (SRD), the evaluation methodology document (EMD), the system description document (SDD), and the draft IEEE 802.16m amendment. The aim of the SRD [17] is to describe the high-level requirements of the new standard, including: legacy support capabilities, design complexity, supported services, operating frequencies and bandwidths, duplexing schemes, advanced PHY techniques (e.g., advanced antenna techniques), and support for emergency and military services. The aim of EMD [18] is to provide a common framework for the methodology that should be employed to perform both link-level and system-level simulations, and to specify the requirements of the respective simulation models, along with their associated parameters, in order to evaluate different technical proposals for inclusion in the IEEE 802.16m standard. The SDD [19] is targeted at providing a high-level description of each feature which will be included in the amendment document. 1.4. IMT-2000 and IMT-Advanced The IMT-2000 is the global standard for third generation (3G) wireless communications as defined by the ITU. In 1999 the ITU approved five radio interfaces for IMT-2000 as part of the ITU-R M.1457 specifications, but WiMAX was included only in 2007 [15]. The goal of IMT-2000 systems is to provide global access to IMT frequency bands and access to a wide range of telecommunication services, supported by the fixed telecommunication networks (e.g., PSTN/ISDN/IP), and to other services that are specific to mobile users. To meet the ever-increasing demand for wireless communication, IMT-2000 is being continuously enhanced. More recently, the ITU defined the set of requirements for IMTAdvanced mobile systems [16], which go beyond those of IMT2000. IMT-Advanced systems will provide access to a wide range of telecommunication services, including advanced packet-based mobile services. Then, in accordance with the ever-increasing user 2. Technical overview of IEEE 802.16m standard The system reference model for IEEE 802.16m is reported in Fig. 1. The MAC layer is divided into three sub-layers: the service-specific convergence sub-layer (CS), the MAC common part sub-layer (CPS), and the security sub-layer. The MAC CS provides a means of transformation of the external network data, received through the CS service access point (SAP), into MAC service data units (SDUs). The MAC SAP is the interface responsible for delivering MAC SDUs to the MAC CPS. The security sub-layer provides the functions necessary for authentication, secure key exchange, and encryption. The MAC CPS of IEEE 802.16m is different from the previous releases of the standard, since a soft classification has been introduced: radio resource control and management functions, and medium access control functions. However, no SAP interface is required between the two logical blocks in the MAC CPS, to allow for a custom cross-layer interaction between these modules. The PHY SAP is implementation-specific and it is up to the vendor to decide its actual internal structure. It provides the means to transfer data, PHY control, and statistics between the MAC CPS and the PHY layer. The IEEE 802.16m physical layer is based on OFDMA and is designed to work in licensed bands below 6 GHz. A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 Fig. 1. IEEE 802.16m system reference model. Lastly, as the network devices are part of a larger network and therefore will require interfacing with other entities, a network control and management system (NCMS) abstraction has been introduced in the standard [11] as depicted on the right hand of Fig. 1. The NCMS abstraction allows the IEEE 802.16m PHY/MAC layers to be independent of the network architecture, the transport network, and the protocols used at the backend. In the rest of this section we will analyze the features introduced in IEEE 802.16m in more detail, firstly focusing on the overall network architecture. 2.1. Network architecture In this section, we describe the high-level network architecture focusing on those aspects defined by the IEEE 802.16 standards. A description of the complete network architecture, including an operator backbone network can be found in [12,14]. Fig. 2 shows the high-level network as defined by the IEEE 802.16 standards. In the IEEE 802.16-2009 system model, the MSs are user devices, 115 including mobile phones, PDAs, and laptops, and the BSs provide the MSs with network access. In addition to them, IEEE 802.16m defines a new type of network node, called relay station (RS). An RS, when present, can be either mobile or fixed, and it communicates with BSs using in-band wireless resources. MSs can then be provided with network access also via one intermediate RS, when direct communication with a BS is either not possible or inconvenient (e.g., because of obstruction or shadowing). The IEEE 802.16m relay architecture will be based on IEEE 802.16j-2009 [25], but substantial improvements on the latter will be brought, which are at the moment of writing still undefined. Furthermore, IEEE 802.16m introduces support for a Femto cell architecture, self organizing networks (SON), and self-optimization procedures. The Femto cell architecture is targeted at small office home office (SOHO) deployments. A Femto BS is a BS with reduced transmit power typically connected to the core network through classical Internet connections (e.g., DSL or cable). After being successfully attached to the network, a Femto BS enters the operational state. As depicted in Fig. 3, a Femto BS supports two operational modes: normal mode and low-duty mode. In the low-duty mode, availability intervals alternate with unavailability intervals similar to the MS power-saving mechanism. However, this low-duty mode is mainly intended to reduce interference with neighbor cells, rather than to save energy. During an availability interval, the Femto BS becomes active on the air interface for synchronization and signaling purposes (e.g., paging or ranging) and for providing the MSs with data transmission opportunities. During an unavailability interval, instead, the Femto BS does not transmit on the air interface, thus it is not reachable by any MS. However, the Femto BS can use these intervals to synchronize with the overlay macro BS or to measure the interference from neighbor cells. The Femto BS can enter low-duty operation mode if one of the following two conditions holds: all the MSs attached to the Femto BS are in idle or sleep mode, or if no MSs are in its service range. Normally, Femto BSs operate in licensed spectrum bands and the standard supports the full service continuity between a Femto BS and the overlay BS, so that an MS can seamlessly perform a handover from/to the Femto BS. Since association to a Femto BS should be favored over that to an overlay macro MS, because of the reduced wireless resources consumed, the following feature has Fig. 2. Overall WiMAX network architecture. 116 A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 Fig. 3. Femto BS operational state diagram. been also defined. A Femto BS may monitor the downlink and uplink signal associated with an MS, while it is being served by an overlay macro BS, and inform the macro BS that the MS is in its coverage and a handover can be accomplished. The IEEE 802.16m standard also provides a basic set of SON functions to automate the configuration of the BS parameters and to optimize the network performance, in terms of coverage and capacity. At the moment, the only functions defined are for measuring and reporting some air interface performance metrics from MS/BS, and adjusting the BS parameters accordingly. However, additional work on the SON specifications is still to be carried out by the WiMAX Forum in order to define the network implications of the SON architecture. Lastly, IEEE 802.16m also specifies a set of self-optimization procedures in order to analyze SON measurements from the BS/ MS and to fine-tune the BS parameters in order to optimize the network performance. This procedure can adjust parameters for QoS, network efficiency, throughput, cell coverage and cell capacity. Typically, the self-optimization procedure uses the following set of measurements from BS/MS: signal quality of the serving and neighbor BSs, interference level and other information from the neighbor BSs, handover status, time and location information of MS, load information of neighbor BS, etc. The three main scenarios for SON and self-optimization are: to the serving BS, when the MS performs a handover to a target BS, all the CIDs are no longer valid and they need to be reassigned by the target BS, with the only exception of some multicast and broadcast CIDs. To fix this lack of flexibility, the IEEE 802.16m MAC maintains the concept of connections, but it introduces a new set of logical identifiers to address an MS and its connections within a BS. In order to uniquely identify each MS attached to it, the BS assigns a 12 bits length station ID (STID) to the MS during network entry. Furthermore, a 4 bits length flow ID (FID) is assigned to each connection. The FID uniquely identifies a connection within the MS. During handover, the target BS assigns a new STID to the MS, but keeps all the FIDs untouched. To summarize, IEEE 802.16m has more stringent limits on the number of connections per MS compared to the 2009 release, i.e., 16 instead of  216 , but the BS and MS management of connection identifiers is very much simplified. As in the previous versions, the IEEE 802.16m MAC layer provides QoS by associating uni-directional flows of packets to service flows. Each service flow is mapped to one transport connection, which is identified by one FID. To support enhanced QoS, IEEE 802.16m improves the polling and granting mechanisms. More than one set of QoS parameters can be associated with each service flow. The MS and the BS negotiate these sets of QoS parameters when the service flow is configured. Both the BS and the MS can dynamically adapt the service flow parameters according to a new predefined set when the traffic characteristics or the QoS requirements change. For instance, in the case of Voice over IP (VoIP) applications, the BS and MS may negotiate two sets of QoS parameters at the connection setup. One set can be used during talkspurt periods, the other during silence periods. Thus, when the VoIP application moves from one state to another, the QoS requirements can be easily adapted. To reduce the latency between the time an MS requests bandwidth and the time the bandwidth is granted by the BS, IEEE 802.16m introduces an enhanced bandwidth request mechanism. The MS can request the bandwidth for its connections using either a regular 5-step approach or an optional quick 3-step approach as illustrated in Fig. 4. Steps 2 and 3 are only used in the 5-step approach. With the quick bandwidth request mechanism, the BS can allocate a pre-determined grant size to the MS based on a simple indication message instead of a full message containing the buffer status of the connection. Fig. 5 shows the overall IEEE 802.16 protocol structure, for BSs, RSs, and MSs. The MAC CPS is further divided into radio resource (1) coverage and capacity optimization, to detect and resolve the blind areas for reliable and maximized network coverage and capacity, when an MS cannot receive any acceptable signals from any BSs; (2) interference management and optimization, to keep the intercell interference at a reasonable level; and (3) load management and balancing, to tune the handover parameters of the BS and dynamically adjust the load with the neighbor BSs. 2.2. MAC layer The MAC layer is connection-oriented and each connection is defined as a uni-directional mapping between the BS and MS MAC peers. There are two types of connections: management and transport. Since its first version of the standard and up to the latest revision in 2009, each connection is identified by a 16 bit connection identifier (CID) and each MS attached to a BS is uniquely identified by its set of CIDs. Since the scope of a CID is local Fig. 4. Bandwidth request mechanism in IEEE 802.16m. A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 117 Fig. 5. IEEE 802.16m protocol structure. control and management (RRCM) and MAC functions, whose components are illustrated below. The radio resource management block includes functions for load balancing, admission control, and interference control. The mobility management block includes functions for intra-radio access technology (RAT) and inter-RAT handover, and it is responsible for deciding whether or not the MS should perform a handover. The network-entry management block is responsible for all the operations performed during initialization and network access, including: network discovery, ranging, basic capability negotiation, registration. The location management block defines the set of procedures and messages to support LBSs, which are included in the standard starting from the 2009 release, and help the development of an end-to-end LBS architecture for, e.g., assisted global positioning system (A-GPS). The idle mode management block includes the functions needed for location update during idle mode and to generate paging advertisement messages in the core network. The security management block handles the authorization, authentication, and key management functions. The system configuration management block distributes the configuration and system parameters to the MSs. The multicast and broadcast service (MBS) block handles the configuration information for MBS data transmission. The MBS support was also introduced in the 2009 release and IEEE 802.16m enhances this solution to allow an end-to-end MBS architecture in conjunction with the network specifications defined by the WiMAX Forum. The enhanced MBS (E-MBS) refers to the capability of the standard to efficiently support point-tomulti-point services, such as video streaming and TV broadcasting. Specifically, IEEE 802.16m supports both static and dynamic E-MBS services: with a dynamic E-MBS service the content is broadcast only if there are users who are receiving such content; with a static service the content is broadcast anyway. The service flow, connection management block allocates STID and FIDs during network entry, handover, and service flow creation procedures. The relay functions block is responsible for enabling multi-hop relay mechanisms and for defining the procedures to maintain a relay path between the BS and the access RS. The multi-carrier block enables the definition of a common MAC entity to control a physical layer spanning over multiple frequency channels (both contiguous and non-contiguous channels). Lastly, the self organization block is used for self-configuration and self-optimization. The MAC functions block includes several sub-blocks related to both control and data planes. We will focus the description on the control plane first. The multi-radio coexistence block allows a correct concurrent operation of IEEE 802.16m and other radios collocated in an MS. In a deployment scenario where the WiMAX radio can interfere with other collocated radios (e.g., Bluetooth and WiFi operating at 2.4 GHz), the standard defines several functions to facilitate such co-existence. The BS and MS can therefore negotiate resources so as to avoid the interference of different radios at the MS. Details of the radio co-existence support can be found in [30]. The data forwarding block performs forwarding functions. RSs are present on the path between ABS and AMS. The Data Forwarding block may cooperate with other blocks such as Scheduling and Resource Multiplexing block and MAC PDU formation block. The sleep mode management block handles sleep mode operations and is responsible for the generation of sleep signaling messages. The PHY control block handles PHY signaling, link adaptation techniques, and ranging adjustment, also including the management of inter-cell/sector interference by means of interference measurements and mitigation, transmit power control, and interference randomization. The quality of service (QoS) block enables the support of several classes of service and defines the set of QoS parameters for each connection. The scheduling and resource multiplexing block schedules user data based on the information obtained from the QoS block and channel quality reports received from the mobile stations. The control signaling block generates control messages to allocate resources in the system. The data plane of the MAC Functions includes the following modules. The ARQ block enables a reliable communication between the sender and the receiver based on an Automatic Repeat reQuest (ARQ) protocol. In IEEE 802.16m the ARQ protocol can be enabled only on a connection that is using hybrid ARQ. In fact, ARQ has been re-designed to leverage the HARQ mechanism to reduce the signaling overhead. The fragmentation/packing block performs fragmentation and/or packing of SDUs based on the output of the scheduling and resource multiplexing block. The MAC PDU formation block eventually constructs the MAC protocol data units (PDUs) to be passed to the PHY. 118 A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 Fig. 6. Frame structure in TDD mode with type 1 and type 3 sub-frames for bandwidth of 5 MHz, 10 MHz, and 20 MHz (DL:UL ratio 5:3 and 1/8 CP length). 2.3. Physical layer As already mentioned, IEEE 802.16m is designed to work on bands below 6 GHz in NLOS environments. The standard only supports OFDMA, with both time division duplexing (TDD) and FDD. The latter also supports devices having half-duplex capabilities, i.e., which are not able to transmit and receive at the same time. The channel bandwidth varies from 5 MHz to 20 MHz, including 5, 7, 8.75, 10, and 20 as possible values. The basic frame structure, depicted in Fig. 6, consists of a superframe broadcast every 20ms. A superframe is divided into four equally-sized 5 ms radio frames and each radio frame further comprises up to eight sub-frames. Each superframe starts with a preamble used for physical synchronization. There are two types of preambles: primary A-Preamble and secondary A-Preamble. A primary A-Preamble is transmitted on the first sub-frame of the second radio frame, while a secondary A-Preamble is transmitted at the beginning of all the other radio frames. There are three1 types of sub-frames whose parameters and usage scenarios are reported in Table 1. A data carrier, as reported in Table 1, consists of one OFDMA symbol and one subcarrier. In FDD, resources between uplink (UL) and downlink (DL) are multiplexed in the frequency domain. Thus, every 20 ms, a DL superframe and an UL superframe are transmitted at two separate frequencies. On the other hand, in TDD multiplexing occurs in time, as illustrated in Fig. 6. Each radio frame begins with a set of DL sub-frames followed by a set of UL sub-frames. A switching point is inserted in each DL/UL transition to allow the MSs to switch their radio from receiver to transmitter, or transmitter to receiver, mode. The first sub-frame of each superframe contains the super frame header (SFH). The SFH is divided into primary SFH (P-SFH) and secondary SFH (S-SFH). The P-SFH has a fixed size and is required every superframe, whereas the S-SFH has a variable size and does not need to be transmitted every superframe. In addition, the S-SFH has a modular structure, i.e., it is made up of several IEs, to allow for flexible broadcasting of system information. In the 2009 release, the content of the SFH was broadcast in two messages for downlink and uplink configuration, i.e. the downlink channel descriptor (DCD) and uplink channel descriptor (UCD), 1 IEEE 802.16m defines a forth type of sub-frame which is used only for 8.75 MHz. It consists of 9 OFDMA Symbols by 18 subcarriers. Table 1 Sub-frame types. Size OFDMA BWs LRU size (OFDMA Pilots Data carriers symbols (MHz) symbols  subcarriers) per LRU per LRU Type I 6 Type II 7 Type III 5 All All All 6  18 7  18 5  18 6 6 6 102 120 84 respectively. The major drawback of this approach is that these messages cannot be fragmented; hence all the system parameters have to be transmitted with the same periodicity irrespectively of their importance. The modular structure of the S-SFH, instead, allows the BS to broadcast system parameters as soon as they change. This is particularly efficient since the different modules in Fig. 5, each mapped to a respective IE, may have different periodicities. However, as in IEEE 802.16-2009, the system parameters are regularly broadcasted to reduce the network entry latency. To support legacy IEEE 802.16-2009 MSs, IEEE 802.16m introduces a special frame structure. Two different schemes are supported: time division multiplexing (TDM) and frequency division multiplexing (FDM). These two schemes differ only in the multiplexing of the uplink sub-frame. Fig. 7 illustrates the legacy support frame structure for 5 MHz, 10 MHz, and 20 MHz system bandwidth in FDM mode. In downlink, the legacy and new systems are multiplexed in a time division fashion, with their frames having an offset of a fixed number of sub-frames as shown on the bottom left corner of Fig. 7. Focusing on the IEEE 802.16m system point of view, if the legacy UL sub-frame is carried within radio frame n the corresponding legacy DL sub-frame was carried within radio frame n  1. In other words, the legacy downlink sub-frame can be seen as located at the end of the IEEE 802.16m radio frame, whereas the corresponding legacy uplink sub-frame can be seen as located in the very subsequent IEEE 802.16m radio frame. In the uplink direction, the legacy and new systems can be multiplexed either in a frequency division or in a time division manner. Legacy MSs ‘‘see” the 16m DL(UL) sub-frames as an unused part of the frame, in both downlink and uplink. The IEEE 802.16m standard has introduced the support for multi-carrier operations, which allow the BS to use one MAC layer to control several PHY layers (or carriers). The set of carriers may be allocated either in a contiguous or in a non-contiguous spectrum, thus allowing wide-band configurations where there is no A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 119 Fig. 7. Frame structure for legacy support (FDM). contiguous spectrum available. The multi-carrier operations are only allowed among ‘‘advance-mode” sub-frames, where legacy operations are not supported. Furthermore, a multi-carrier architecture allows the BS to concurrently support different types of MSs, in which each type has different Radio Frequency (RF) capabilities. For example, in a scenario with three carriers (5 MHz, 5 MHz, 10 MHz) the following types of MSs can be supported: single carrier 5 MHz, single carrier 10 MHz, multi-carrier 10 MHz, and multi-carrier 20 MHz. Details on the multi-carrier frame structure can be found in [23]. We will now complete the overview of the frame structure focusing on the physical structure of the sub-frame. Each subframe spans two dimensions: time and frequency. In the time domain, a sub-frame is made up of a number of consecutive OFDM symbols, whereas the frequency domain consists of a number of adjacent physical subcarriers. In terms of resource allocation, each sub-frame can be divided into a set of physical resource units (PRUs). In the time domain a PRU occupies all the OFDM symbols within a sub-frame (e.g., a row in the time domain) and 18 subcarriers on the frequency domain. The actual size of a PRU depends on the sub-frame type. For instance, in type I sub-frames, a PRU consists of 6 OFDMA symbols  18 subcarriers. Any PRU contains both data and pilot subcarriers. To support several subcarrier permutations schemes, the standard introduces the concept of Logical Resource Units (LRUs). An LRU has the same size as a PRU, but it consists of a set of logical, rather than physical, subcarriers in frequency. The actual mapping between a logical subcarrier and a physical subcarrier depends on the LRU permutation scheme. There are two permutations schemes currently defined in the standard: contiguous (i.e., localized or AMC in the legacy system) and non-contiguous (i.e., distributed or PUSC in the legacy system). An LRU employing the contiguous permutation scheme is called a contiguous resource unit (CRU). In a CRU, the mapping between logical and physical subcarriers is done such that two adjacent logical subcarriers are also physically adjacent. An LRU employing the distributed permutation scheme is called distributed resource unit (DRU). In a DRU, the mapping between logical and physical subcarriers is done through an algorithm for randomization. An analysis of the data subcarriers per LRU per sub-frame type is reported in Table 1. Each sub-frame can be divided into one or more frequency partitions (FPs), where each partition consists of a set of physical resource PRUs across all the OFDM symbols available in the subframe. Each frequency partition can contain both CRUs and DRUs. The division between CRUs and DRUs is frequency partition specific and is the same for all the sub-frames in a super frame (i.e., the actual information is broadcast in the SFH). In order to achieve either a better frequency diversity gain or a better frequency-selective scheduling gain, the LRUs in one FP are further divided into DRUs and CRUs. An example of mapping downlink resources to resource units is illustrated in Fig. 8. The FPs were introduced to efficiently support different scenarios, including, for example, fractional frequency reuse (FFR) and MBSs. FRR is based on subcarriers across the whole frequency band being grouped into frequency partitions, which can employ different reuse factors. The basic concept is illustrated by means of the simple example in Fig. 9, where there are three sectors and four different FPs within each sector (A–D). As can be seen, subcarriers across the whole bandwidth are grouped into FPs with different frequency reuse factors. The received signal quality can be improved at MSs in the frequency partitions with higher frequency reuse factor, due to lower interference levels. This will be helpful for the MSs located around cell boundaries or, anyway, suffering severe inter-cell interference. On the other hand, the BS may apply lower frequency reuse factors for the other frequency partitions to serve the MSs which do not experience significant inter-cell interference. It should be pointed out that compared to the traditional orthogonal approach (i.e., only one sector transmits with non-zero power while the others are silent), the major difference is the dynamic and adaptive nature of FRR: the resource allocation, in fact, 120 A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 Fig. 8. Downlink subcarrier to resource mapping. Table 2 Spectral efficiency. MCS IEEE 802.16-2009 Bits/slot Fig. 9. Basic concept of FFR. can be tuned in many aspects, including the reuse factor in partition, the FP transmission power, and, most important, interference-based measurements collected at the MSs or the BS. We will now continue our discussion on the physical layer focusing on several key elements of L1 techniques. The support for multiple antennae (MIMO) technology has been greatly improved in IEEE 802.16m and it plays a key role in the new wireless system design. It offers significant performance improvement in data throughput and coverage without additional bandwidth or transmission power. According to the IEEE 802.16m SRD [17], the minimum antenna configuration in downlink is 2  2 and in uplink is 1  2. IEEE 802.16m supports both single-userMIMO (SU-MIMO) and multi-user-MIMO (MU-MIMO). In SUMIMO, only one user is scheduled per RU, whereas multiple users can be scheduled in one RU with MU-MIMO. Depending on the application scenarios, five different MIMO modes have been introduced in IEEE 802.16m, as described in [24]. In both SISO and MIMO modes, link adaptation is used by the BS to follow the channel variations of MSs, for both downlink and uplink. Specifically, the BS adapts the modulation and coding scheme (MCS) level based on physical layer measurements, for uplink, and channel quality indicator (CQI) messages conveyed by the MSs, for downlink. The best MIMO mode is also selected based on the same factors. In Table 2 we report the theoretical best-case spectral efficiency, derived through inspection and simple derivations, without taking into account the MAC overhead. The figures represent an QPSK-1/2 QPSK-3/4 16-QAM-1/2 16-QAM-3/4 64-QAM-1/2 64-QAM-2/3 64-QAM-3/4 64-QAM-5/6 48 72 96 144 144 192 216 240 Bits/s/Hz 0.6624 0.9936 1.3248 1.9872 1.9872 2.6496 2.9808 3.3120 IEEE 802.16m Bits/LRU Bits/s/Hz Type I Type III 102 153 204 306 306 408 459 510 84 126 168 252 252 336 378 420 0.7488 1.1232 1.4976 2.2464 2.2464 2.9952 3.3696 3.7440 upper bound to the system’s performance achievable in practice. Quite clearly, even in the SISO mode considered, IEEE 802.16m achieves a better spectral efficiency than that of IEEE 802.162009. This is because of the enhanced structure of pilot subcarriers for channel estimation at the MSs [23,29]. Lastly, the hybrid ARQ (HARQ) support has been enhanced too in IEEE 802.16m. H-ARQ is used on top of normal channel coding to improve the reliability of unicast data traffic. Basically, HARQ is based on a stop-and-wait protocol. Multiple concurrent channels for the same MS can be enabled. In the 2009 release the mandatory H-ARQ scheme is chase combining, where every retransmission of the data burst contains exactly the same information and is transmitted with the same MCS as the previous one. On the other hand, in IEEE 802.16m, incremental redundancy is used, where any retransmission can contain a new portion of the overall encoded data, not necessarily using the same MCS. In any case, chase combining can be seen as a special case of incremental redundancy. In addition, IEEE 802.16m also allows asynchronous downlink retransmissions, not supported in the 2009 release. 3. IEEE 802.16m vs. IEEE 802.16-2009: DL control analysis Due to the need to indicate to every MS that a certain amount of bandwidth is granted in a frame, the frequency/time location of their grant, and to convey any additional information needed, A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 e.g., for H-ARQ processing, both IEEE 802.16-2009 and IEEE 802.16m use so-called maps. Such special control messages are very important because they consume a significant portion of the available bandwidth. It is worth noting that downlink bandwidth is consumed for both downlink and uplink maps. The former is especially relevant from a designer’s perspective because the system is expected to have a much higher throughput, hence a greater overhead, in downlink than in uplink. In the first part of this section, we will describe the structure of the downlink map in IEEE 802.16-2009 and IEEE 802.16m, respectively. We will then carry out a numerical analysis to compare the overhead of these two systems. 3.1. DL control signaling in IEEE 802.16-2009 The generic frame structure of IEEE 802.16-2009 in TDD is reported in Fig. 7. We will describe only the downlink sub-frame. After the physical preamble, which is required to synchronize the MS receivers, the BS transmits the Frame Control Header (FCH) and the DL-MAP message, which are allocated within the downlink sub-frame in column-wise order. The remaining part of the downlink sub-frame is allocated as data regions. A data region may be visualized as a rectangle of OFDMA slots, as shown in Fig. 7. Each data region is specified through an Information Element (IE) into the DL-MAP. The greater the number of data regions therefore, the higher the overhead due to the DL-MAP transmission. As far as shaping is concerned, a given number of slots can generally be arranged into several different shapes in order to form one or more data regions. For example, assume that the BS has to transmit eight 121 slots of data to an MS. If only one data region is used, there are four possible shapes: 2  4, 4  2, 1  8 or 8  1. It is also possible for the BS to employ multiple data regions, e.g., two data regions of four slots each. Lastly, the BS can inflate the data regions so that more than eight slots are used, e.g., a single 3  3 data region, where one slot remains empty. However, not all alternatives are equivalent, since the MAC overhead varies, depending on the configuration. 3.2. DL control signaling in IEEE 802.16m Without loss of generality, we will describe the DL Control Structure where there is only one frequency partition in the subframe. The description can easily be extended to when there is more than one FP for each sub-frame. In fact, each FP presents the same control structure irrespectively of the number of frequency FPs per sub-frame. Fig. 10 shows the DL Control Structure for unicast control messages. Resources in each sub-frame are notified to the MSs at the beginning of the sub-frame through an Advanced MAP (A-MAP) region. Depending on a system parameter broadcast in the S-SFH, the A-MAP region can allocate resources either for one sub-frame (i.e., the sub-frame where the A-MAP is transmitted, Fig. 10a) or for two consecutive sub-frames (Fig. 10b). There might be one A-MAP region for each frequency partition and this is transmitted using distributed resource allocation. The A-MAP region addresses resources in both DRUs and LRUs and starts with a non-user specific part which contains information required to decode the assignment A-MAP IEs, including: the number of assignment A-MAP IEs within different MCS groups and Fig. 10. A-MAP design for IEEE 802.16m. 122 A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 Table 3 A-MAP types and respective sizes. Information element Min Max DL UL DL UL DL UL 52 50 51 52 54 52 55 55 51 53 60 60 basic assignment A-MAP basic assignment A-MAP group configuration A-MAP group configuration A-MAP persistent A-MAP individual persistent A-MAP individual other fields not yet specified at the moment of writing. The nonuser specific A-MAP has a fixed size and is transmitted with given MCS. There are several types of assignment A-MAP IE as reported in Table 3. The DL (UL) Basic Assignment A-MAP IE notifies DL (UL) resources to one MS. These two IEs are transmitted using separate encoding: each IE is, thus, encoded with a different MCS depending on the channel quality of the recipient MS. The identifier of the MS for which an IE is intended is not explicitly encoded. Rather, it is bit-masked with the 16 bits CRC. This reduces the overhead of specifying the destination address in each IE, but introduces an additional complexity when decoding the IE. This is because any MS has to perform a ‘blind decoding’ operation on the whole set of IEs, thus trying all the possible combinations until its IE is decoded. In fact, there is no indication in the control messages to specify when one A-MAP IE ends and another one starts. Other A-MAP IEs are also specified in the standard, but are not reported here. The DL (UL) Basic Assignment A-MAP IE has a constant size independently of the system bandwidth. The minimum and maximum IE sizes are reported in Table 3.2 The actual size of an IE depends on the MIMO scheme employed for data transmission. The A-MAP region is transmitted using a specific resource unit, called a minimum A-MAP logical resource unit (MLRU), which corresponds to half an LRU. Each A-MAP IE is transmitted using one or multiple MLRUs. The size of the A-MAP region is then rounded to the closest LRU as defined in Table 1. The minimum resource unit that can be assigned through an assignment A-MAP IE is equal to one LRU. 3.3. Performance evaluation In this section we compare two aspects that have been significantly improved in IEEE 802.16m, and are important for the overall performance of the system: the spectral efficiency and the downlink signaling overhead. We will first compare the overhead of the downlink map for the two systems. The results are obtained via a numerical analysis as follows. For IEEE 802.16m, we considered the case of type I and type III sub-frames separately. For IEEE 802.16-2009, we computed the overhead by considering both a single sub-map and the minimum map size with three sub-maps. Note that in practice finding the latter requires OðN 2 Þ operations, with N being the number of different MCSs. To make a fair comparison, the non-user specific part of the downlink map is not considered, which is pessimistic from the point of view of IEEE 802.16m. The unit of allocation was the frame for IEEE 802.16-2009, and the sub-frame for IEEE 802.16m, both with the same duration, i.e., 5 ms. The number of MSs per unit of allocation varied from 10 to 100. For each possible combination of the above factors we ran 1000 independent drops, whose results were averaged to produce the final results. A drop consisted of locating every MS in a random position within the cell, in a uniform manner. We considered a single hexagonal cell, with the BS located in a corner, which is a typical deployment with tri-sectored antennae. The BS-to-BS distance was 500 m. After positioning the MS, its path-loss and shadowing 2 Reserved and padding bits have not been taken into account. Fig. 11. Comparison of the downlink signaling overhead between IEEE 802.16-2009 and IEEE 802.16m. were computed according to the specifications in [18], in the ‘‘macro cell” scenario. The resulting SINR, assuming a BS transmission power of 43dBm and a constant co-cell interference of 26dB, was mapped to an MCS according to the minimum SNR requirements specified in the latest WiMAX mobile system profile specifications. To obtain comparable results the same SNR mapping and set of MCS were used for both IEEE 802.16-2009 and IEEE 802.16m, which, again, is pessimistic for the latter. In Fig. 11 we show the overhead comparison, in terms of the ratio between the radio resources required to transmit the A-MAP IEs (or sub-maps) and the unit of allocation size. Secondary effects, like padding and MLRU or slot alignment were not included. As can be seen, the curve of IEEE 802.16-2009 with a single sub-map is significantly higher than the others. If three sub-maps are used, the overhead improves significantly, but still lies above IEEE 802.16m. Note that this case is rather optimistic, since it assumes that the hardware is really able to compute the minimum map size, with OðN2 Þ operations, in real-time. With regard to IEEE 802.16m, the sub-frame type I was slightly more efficient than sub-frame type III, because the former had denser LRUs. In any case, the absolute value of the overhead was acceptable: even with a very high number of MSs, i.e., 100 different MSs served every 5 ms, the overhead was between 10% and 15%. 4. Conclusions In this paper we have first presented a historical overview of the IEEE 802.16 standard from the first version released in 2001 to the current draft version which is expected to be published in autumn 2010. We have then provided a detailed technical analysis of the PHY, MAC layer, and other relevant aspects of the new standard, including a detailed description of its relay architecture and support for self organizing networks and Femto cells. To better understand the technical impact of the new release, we have also presented a comparison of the downlink control overhead between IEEE 802.16-2009 and IEEE 802.16m. In fact, one of the biggest problems in IEEE 802.16-2009 and its previous revisions was the high signaling overhead. The simulation results have shown that the new version of the standard introduces several effective mechanisms to reduce the signaling overhead. References [1] IEEE 802.16-2001, ‘‘IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems,” Apr. 8, 2002. A. Bacioccola et al. / Computer Communications 33 (2010) 113–123 [2] C. Eklund, R.B. Marks, K.L. Stanwood, IEEE Standard 802.16: a technical overview of the wirelessman™ air interface for broadband wireless access, IEEE Communications Magazine (2002). [3] C. Cicconetti, L. Lenzini, E. Mingozzi, C. Eklund, Quality of service support in IEEE 802.16 networks, IEEE Networks (2006). [4] IEEE P802.16a-2003: ‘‘Amendment to IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Wireless Access Systems – Medium Access Control Modifications and Additional Physical Layers Specifications for 2–11 GHz”, April 2003. [5] IEEE 802.16-2004, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems”, October 2004. [6] IEEE 802.16e-2005, ‘‘Amendment for Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands”, February 2006. [7] IEEE 802.16f, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Management Information Base”, December 2005. [8] IEEE 802.16g, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Amendment 3: Management Plane Procedures and Services”, December 2007. [9] IEEE 802.16h/D9, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access – Improved Coexistence Mechanisms for License-Exempt Operation”, Draft D9, March 2009. [10] IEEE 802.16j, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Multihop Relay Specification”, May 2009. [11] IEEE 802.16, ‘‘IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems”, May 2009. [12] K. Etemad, Overview of mobile WiMAX technology and evolution, IEEE Communication Magazine (2008). [13] WiMAX Forum Network Architecture Stage 2, WiMAX Forum website, Available from: <http://www.wimaxforum.org/resources/documents/ technical/release1>, February 2009. [14] WiMAX Forum Network Architecture Stage 3, WiMAX Forum website, Available from: <http://www.wimaxforum.org/resources/documents/ technical/release1>, February 2009. 123 [15] Recommendation ITU-R M.1457: Detailed specifications of the radio interfaces of International Mobile Telecommunications-2000 (IMT-2000), November 2009. [16] Recommendation ITU-R M.1645: Framework and overall objectives of the future development of IMT-2000 and systems beyond IMT-2000, January 2003. [17] IEEE 802.16m System Requirements Document (SRD), Draft, 2009-09-24, IEEE 802.16m-07/002r9, Available from: <http://wirelessman.org/tgm/>. [18] IEEE 802.16m Evaluation Methodology Document (EMD), Draft, 2009-01-15, IEEE 802.16m-08/004r5, Available from: <http://wirelessman.org/tgm/>. [19] IEEE 802.16m System Description Document (SDD), Draft, 2009-09-24, IEEE 802.16m-09/0034r2, Available from: <http://wirelessman.org/tgm/>. [20] New Report ITU-R M.2134: Requirements related to technical performance for IMT Advanced radio interface(s), 2008. [21] Report ITU-R M.2135: Guidelines for evaluation of radio interface technologies for IMT-Advanced, 2008. [22] Woonsub Kim, Mobile WiMAX, The leader of the mobile internet era, IEEE Communication Magazine (2009). [23] Sassan Ahmadi, An overview of next generation mobile WiMAX technology, IEEE Communication Magazine (2009). [24] Qinghua Li, Xintian Eddie Lin, Jianzhong (Charlie) Zhang, Wonil Roh, Advanced of MIMO technology in WiMAX: from IEEE 802.16d/e/j to 802.16m, IEEE Communication Magazine (2009). [25] Jerry Sydir, Rakesh Taori, An evolved cellular system architecture incorporating relay stations, IEEE Communication Magazine (2009). [26] Kamran Eternad, Limel Wang, Multicast and broadcast multimedia services in mobile WiMAX networks, IEEE Communication Magazine (2009). [27] Muthalah Venkatachalam, Kamran Eternad, Wayne Ballantyne, Bonnie Chen, Location services in WiMAX networks, IEEE Communication Magazine (2009). [28] Yang Yang, Honglin Hu, Jing Xu, Guoqlang Mao, Relay technologies for WiMAX and LTE-advanced mobile systems, IEEE Communication Magazine (2009). [29] IEEE Std 802.16m/D2 (amendment to IEEE Std 802.16-2009, developed by Task Group ‘m’ Working Document), Available from: <http://wirelessman.org/pubs/ 80216m.html>. [30] Jing Zhu, Hujun Yin, Enabling collocating coexistence in IEEE 802.16 networks via perceived concurrency, IEEE Communication Magazine (2009). [31] Submission of a Candidate IMT-Advanced RIT based on IEEE 802.16 (Parts 1–4) (2009-10-07), Available from: <http://wirelessman.org/liaison/index.html>.