The Advantages of SOFDMA for WiMAX

The Advantages of SOFDMA for WiMAX Vladimir Bykovnikov Intel Corporation Abstract SOFDMA has several advantages when used in NLOS wireless networks. The paper outlines these advantages and shows the evolutionary path from OFDM to SOFDMA. Copyright 2005 Intel Corporation. All rights reserved. Page 1 of 10

Table of Contents Table of Contents...2 1. Overview...2 2. Wireless Channel...2 3. OFDM...3 3.1 Signal Structure...3 3.2 Multipath Immunity...3 3.3 Diversity...4 3.4 Easy Support of Multiple Antennas...4 4. OFDMA...5 4.1 Subchannels and Multiple Access...5 4.2 Subcarrier Permutation Schemes...5 4.3 Sub-channelization Gain...6 4.4 Efficient Scheduling...7 4.5 Frequency Reuse Factor of One...7 5. SOFDMA...7 5.1 SOFDMA Profiles...7 5.2 Other Complementary Features of SOFDMA...8 6. Summary...8 7. Acknowledgements...9 8. References...9 Copyright 2005 Intel Corporation. All rights reserved. Page 2 of 10

1. Overview This paper provides a high-level overview of the advantages of SOFDMA. It is assumed that a reader already has some background in wireless communication and the goal is to understand the advantages of SOFDMA the air interface defined for portable/mobile WiMAX systems [1-3]. This paper starts with a description of the wireless communication channel and the challenges faced in non-line-of-sight (NLOS) links. It then describes the signal structure and features of orthogonal frequency division multiplexing (OFDM), orthogonal frequency division multiple access (OFDMA), and Scalable OFDMA (SOFDMA). This paper is structured to show the evolutionary path from OFDM to SOFDMA in the hope that it will help the reader understand advantages of each multiplexing scheme step by step. Please note that OFDMA has all the advantages described in the OFDM section, and SOFDMA has all the advantages of both OFDMA and OFDM. 2. Wireless Channel In non-line-of-sight (NLOS) wireless channels electromagnetic waves from the transmit antenna travel via several different paths until they reach a receiver (Fig. 1). There may or may not be a direct line-of-sight (LOS) path between the transmitter and the receiver. These multiple paths are referred to as multipath propagation. Multipath presents a challenge for any communication system and results in additional complexity of system design. Fig. 1. NLOS multipath propagation The length of each path is different so signals coming to the receiver over each path experiences different time delays, resulting in delay spread. The channel is characterized by this delay spread which depends on the terrain type, environment (e.g. urban, suburban, rural), and other factors [4]. The delay spread can be up to 10 μs in typical urban environments [5]. In the frequency domain large delay spreads translate into frequency-selective channels. Signals on some frequencies arrive at the receiver in phase signals at some other frequencies arrive out of phase [6]. That results in frequency selective fading as shown in Fig. 2. NLOS channels may also vary in time significantly, due to moving transceivers in mobile communications. Also time variation of NLOS channels is caused by other moving objects in the paths of signals. This results in time selective fading as shown in Fig. 2. Copyright 2005 Intel Corporation. All rights reserved. Page 2 of 10

Fig. 2. Frequency- and time-selective fading 3. OFDM 3.1 Signal Structure OFDM stands for Orthogonal Frequency Division Multiplexing. An OFDM signal is comprised of a number of orthogonal subcarriers (also referred to as tones). Each subcarrier can be modulated independently by BPSK, QPSK, 16QAM, or 64QAM modulation and carry different pieces of information. All subcarriers within an allocation are modulated using the same modulation but the modulation can be adapted for each subscriber according to the signal-to-noise ratio (SNR) at the receiver. Besides data subcarriers there are pilot subcarriers, DC subcarrier, and guard-band subcarriers. Pilot subcarriers are used to measure channel characteristics in order to aid signal recovery at the receiver. Guard-band subcarriers are nulled in order to satisfy spectrum mask requirements and to reduce interference with adjacent channels. The structure of OFDM signals is illustrated in Fig. 3. Fig. 3. Structure of OFDM signals (frequency domain) OFDM signals can be easily constructed in the frequency domain and transformed to the time domain with the relatively simple procedure. The term OFDM is frequently followed by the number that depicts the potential number of subcarriers (including guard-band subcarriers) in the signal (e.g. OFDM-256). 3.2 Multipath Immunity One of the reasons for choosing OFDM for WiMAX is its natural immunity to multipath. This is the key to success for NLOS wireless communication systems. As shown above, in NLOS channels, signals from the transmitter travel via multiple paths and arrive at the receiver with different delays. This leads to intersymbol interference (ISI) which is an issue for many traditional single-carrier broadband systems (higher data rates in single-carrier systems translate to smaller symbol durations and more severe ISI). OFDM makes the symbol duration significantly larger (e.g. 1024 times larger for OFDM-1024 than for singlecarrier systems with the same data rate). That greatly reduces the relative size of the delay spread compared to symbol duration in time domain, and greatly reduces the effect of that ISI on communications. Additionally, OFDM encodes data bits using forward error correction (FEC) coding and distributes them Copyright 2005 Intel Corporation. All rights reserved. Page 3 of 10

across several subcarriers. If frequency-selective fading causes errors in the reception of a few subcarriers, the data bits in those subcarriers are recovered using FEC. Finally, OFDM lends itself easily to the use of diversity techniques to combat frequency-selective fading as described in the following section. In these ways, OFDM eliminates the need for the complex equalizers that are common in CDMA and TDMA systems today. 3.3 Diversity In wireless communication fading reduces link range and hence additional margin is required in the link budget to ensure proper signal recovery in the receiver. For example for SUI-3/omni channel [4] in order to have a probability of fading less than 10-3, a narrowband system sacrifices about 30 db of link budget. OFDM is a wide-band system that with the help of coding (forward error correction) and interleaving provides significant fade-margin improvement (frequency diversity). The improvement depends on channel conditions and bandwidth (e.g. for 2-MHz bandwidths in SUI-3/omni channels [4] the improvement is about 15 db). With natural frequency diversity of OFDM, even if some data is received in error on the subcarriers with frequency-selective fading, the entire packet can often be reconstructed by using the data from the rest of subcarriers and applying error correction. OFDM also makes easy use of diversity techniques to provide further improvement in fade margin such as multiple antennas (spatial diversity), repetition (time diversity), and space-time coding (STC) (a combination of spatial and time diversity). The combination of frequency, time, and space diversity methods results in significant fade margin improvement. 3.4 Easy Support of Multiple Antennas There is a theoretical limit (the Shannon limit) to the data rate of a channel that depends on the bandwidth and signal-to-noise ratio (SNR) of that channel. Please note that the limit applies to a single channel. However if the receiver and/or transmitter have more than one antenna and there is significant multipath in the communication channel, signals from/to antennas travel several different paths. Each combination of transmit and receive antennas can be considered as a separate channel. In an environment with severe multipath, where the channels are independent, the throughput can be greatly enhanced in accordance with the number of transmitter-receiver combinations in the system. Thus, the system throughput can significantly exceed the theoretical limits for single-antenna systems! This concept is known as MIMO (Multiple Input Multiple Output). Also multiple antennas can be used in an adaptive antenna system (AAS) to increase the link budget by forming a beam on the transmit side in the direction of a subscriber (beam-forming) and/or by combining signals from multiple spatially-separated antennas in the receiver (MRC Maximum Ratio Combining, an effective spatial-diversity combining method ). Multiple antennas represent a great potential to increase spectral efficiency/link budget and it can be implemented in two different ways in the time domain or in the frequency domain. Time-domain implementation requires complex time/space processing and a lot of computational resources that currently is hardly viable for portable/mobile devices. Frequency-domain implementation is much simpler because it reduces computation to simple weighting of complex numbers. In contrast, CDMA systems utilize time-domain processing that is a barrier for MIMO and AAS. OFDM systems naturally utilize Copyright 2005 Intel Corporation. All rights reserved. Page 4 of 10

frequency-domain processing, which greatly reduces the cost and unlocks the enormous potential for MIMO and AAS. 4. OFDMA 4.1 Subchannels and Multiple Access OFDMA (Orthogonal Frequency Division Multiple Access) is a multiple access method based on OFDM signaling that allows simultaneous transmissions to and from several users along with the other advantages of OFDM. While OFDM-256 is used in fixed WiMAX (based on IEEE 802.16-2004, [1]), OFDMA mode has many advantages to be considered for nomadic/mobile usage and is used in IEEE 802.16(e) In OFDMA, subcarriers are assigned to subchannels that in turn can be allocated to different users. This provides high-granularity bandwidth allocation as illustrated in Fig. 4. Fig. 4. Two-dimensional scheduling in OFDMA 4.2 Subcarrier Permutation Schemes There are two approaches to allocating subcarriers to subchannels: 1) distributed subcarrier allocation, and 2) adjacent subcarrier allocation. The two approaches are shown on Fig. 5. In the distributed-subcarrier-allocation approach, a subchannel uses different subcarriers randomly distributed across the channel bandwidth. The distributed-subcarrier approach maximizes frequency diversity and averages inter-cell interference. This is the best approach for mobile environment where channel characteristics change fast. In the adjacent-subcarrier-allocation approach, a subchannel uses adjacent subcarriers which can be adaptively selected by the scheduler (subcarriers with the highest signal-to-interference-plus-noise ratio (SINR) are chosen, and subcarriers in deep fades are avoided). This approach creates a loading gain, and it s easier to use with beam-forming AAS. The limitation of this approach is that it requires relatively stable conditions, where the characteristics of the channel change slowly (low-speed or nomadic usage). Copyright 2005 Intel Corporation. All rights reserved. Page 5 of 10

SINR SINR SINR SINR SINR Fig. 5. Adjacent and distributed subcarrier allocation OFDMA implements several permutation schemes to support both approaches. The partial usage of subcarriers (PUSC) and full usage of subcarriers (FUSC) permutation schemes defined in the standard have the advantages of distributed subcarrier allocation and are well suited for mobile, fast-moving subscribers. On the other hand, the adaptive modulation and coding (AMC) scheme uses adjacent subcarriers that maximizes efficiency for stationary/nomadic users by creating loading gain and enabling beam-forming AAS. The different schemes can be adaptively used for different subscribers within the frame thereby maximizing overall system efficiency. Adjacent and distributed subcarrier allocation schemes are compared in Table 1. PUSC and AMC are recommended in WiMAX. Table 1: Comparison of adjacent and distributed subcarrier allocation schemes Parameter Adjacent subcarrier allocation (AMC) Distributed subcarrier allocation (PUSC, FUSC) Gain Sub-channelization gain and loading gain Sub-channelization gain and benefits of frequency diversity Scheduling Requires advanced scheduler that allocates Simplified scheduler, doesn t use info about the subchannels according to channel characteristics Almost no data lost channel Requires more redundancy (overhead) for forward error correction Efficiency in multipath channel Channel Can be used in stationary channel Can be used in fast-changing channel AAS & MIMO Easier implementation More complicated implementation Usage Fixed, portable, nomadic, pedestrian speed Mobile fast-moving subscribers 4.3 Sub-channelization Gain Sub-channelization gain is one of the unique OFDMA features that can significantly improve uplink link budgets. Imagine a situation where a subscriber uses 1 of 16 subchannels and concentrates all its transmitter power in the subchannel. In this situation the power of each subcarrier in the subchannel will Copyright 2005 Intel Corporation. All rights reserved. Page 6 of 10

be increased by 10*log 10 (16) = 12 db. This improvement in uplink budget can be used to increase uplink link range, compensate for indoor penetration loss, or save power for the subscriber. It comes at almost no cost in terms of system capacity, since the rest of the subchannels can be used by other subscribers. Sub-channelization is a feature of OFDMA. It is also supported in the OFDM-256 PHY on the uplink defined in IEEE-802.16-2004 [1]. This brings the advantage of sub-channelization gain to fixed WiMAX systems and enables the use of indoor CPEs and nomadicity. 4.4 Efficient Scheduling With OFDMA, packets can be scheduled across both frequency (subchannels) and time (symbols). It provides an added dimension of flexibility resulting in higher granularity in resource allocation compared to time-division multiplexing (TDM) systems (e.g. OFDM). More degrees of freedom in scheduling improve fairness, quality of service (QoS), and bandwidth efficiency (an improvement of about 25-50% can be achieved). 4.5 Frequency Reuse Factor of One OFDM works well in the channels with relatively high SINR. In multi-cell deployments, in order to avoid inter-cell interference, basic OFDM requires directional antennas or relatively high frequency-reuse factors and careful radio-frequency (RF) planning. OFDMA with its various subcarrier allocation schemes (FUSC and PUSC) improves performance in multi-cell deployments by averaging the interference across multiple cells. The interference becomes a function of cell loading and can be significantly reduced by efficient scheduling. OFDMA systems, on the other hand, are very flexible in terms of RF planning and support a variety of frequency reuse schemes. Two of the most promising schemes are 1x3x1 and 1x3x3. Both schemes use three-sector base-stations and require only one RF channel for all sectors and base-stations hence opening the door for operators who have limited amount of spectrum. Frequency reuse 1x3x1 eliminates the need for any frequency planning. That is a significant advantage especially for heavy urban areas where RF planning is very difficult. Scheme 1x3x3 uses different (orthogonal) sets of tones (caled segments ) for each sector of a base-station thereby reducing inter-cell interference and minimizing outage area. This scheme also simplifies RF planning one need only assign segments to sectors while using the same RF channel among all base-stations. 5. SOFDMA 5.1 SOFDMA Profiles When designing OFDMA wireless systems the optimal choice of the number of subcarriers per channel bandwidth is a tradeoff between protection against multipath, Doppler shift, and design cost/complexity. Increasing the number of subcarriers leads to better immunity to the inter-symbol interference (ISI) caused by multipath (due to longer symbols); on the other hand it increases the cost and complexity of the system (it leads to higher requirements for signal processing power and power amplifiers with the capability of handling higher peak-to-average power ratios). Having more subcarriers also results in narrower subcarrier spacing and therefore the system becomes more sensitive to Doppler shift and phase noise. Calculations [7] show that the optimum tradeoff for mobile systems is achieved when subcarrier spacing is about 11 khz. Copyright 2005 Intel Corporation. All rights reserved. Page 7 of 10

Unlike many other OFDM-based systems such as IEEE 802.11a/g WLANs, the 802.16 standard [1] supports variable bandwidth sizes for NLOS operations. In order to keep optimal subcarrier spacing, the FFT size should scale with the bandwidth. This concept is introduced in Scalable OFDMA (SOFDMA) [2,7]. Possible SOFDMA profiles are shown in Table 1. Please note that in order to reduce system complexity and facilitate interoperability the decision was made to limit the number of profiles for WiMAX. At the present time only two FFT sizes, 512 and 1024, are recommended in WiMAX. Table 1: OFDMA scalability parameters Parameters Values System bandwidth (MHz) 1.25 5 10 20 FFT size (N FFT ) 128 512 1024 2048 Subcarrier frequency spacing 11.1607 khz Useful symbol time (T b =1/ f) 89.6 μs Besides the fixed (optimal) subcarrier spacing, SOFDMA specifies that the number of subcarriers per subchannel should be independent of bandwidth too. This results in the property that the number of subchannels scales with FFT/bandwidth. The basic principals of SOFDMA are clearly outlined in [7]: Subcarrier spacing is independent of bandwidth. The number of subcarriers scales with bandwidth. The smallest unit of bandwidth allocation, based on the concept of subchannels, is fixed and independent of bandwidth and other modes of operation. The number of subchannels scales with bandwidth and the capacity of each individual subchannel remains constant. 5.2 Other Complementary Features of SOFDMA In addition to variable FFT sizes, SOFDMA supports features such as Advanced Modulation and Coding (AMC), Hybrid Automatic Repeat Request (H-ARQ), high-efficiency uplink subchannel structures, Multiple-Input-Multiple Output (MIMO) in DL and UL, as well as other OFDMA default features such as a variety of subcarrier allocation and diversity schemes. Discussing these features is beyond of the scope of the paper. An overview of these features can be found in [7] and detailed descriptions can be found in the IEEE 802.16 standards [1-3]. 6. Summary There are many challenges in NLOS communication channels (e.g. delay spread, intersymbol interference, time-selective fading, and frequency-selective fading caused by multipath propagation). OFDM is extremely well suited to overcoming those challenges. OFDMA is a multiple-access method that allows simultaneous transmissions to and from several users and provides several more advantages (e.g. subchannelization gain, loading gain, more efficient scheduling, and a frequency reuse factor of one). OFDMA is also very well suited for use with AAS and MIMO which can significantly improve throughput, increase link range, and reduce interference. SOFDMA uses a subchannel structure that scales with bandwidth, providing an optimal tradeoff between protection against multipath and Doppler shift, and design cost/complexity. All of these advantages contribute to a spectral efficiency that is much higher than competing systems, and the potential for even greater improvements in the future. Copyright 2005 Intel Corporation. All rights reserved. Page 8 of 10

7. Acknowledgements This paper benefited greatly by helpful reviews and discussions with Fu-sheng Cheng, Kuo-Hui Li, Greg DesBrisay, Jitendra Raichura, Andrew Tang, Shailender Timiri, Masud Kibria, Kranti Singh, and Hassan Yaghoobi. 8. References [1] IEEE Std 802.16-2004 Part 16: Air Interface for Fixed Broadband Wireless Access Systems, IEEE New York, 2004. [2] IEEE P802.16e/D9 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, New York, June 2005. [3] IEEE P802.16-2004/Cor1/D3 Corrigendum to IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access Systems, IEEE, New York, 2005. [4] Channel Models for Fixed Wireless Applications, IEEE 802.16 Broadband Wireless Access Working Group, http://ieee802.org/16, IEEE, New York, 2003. [5] Donald C. Cox., 910 MHz Urban Mobile Radio Propagation: Multipath Characteristics in New York City, IEEE Transactions On Communications, vol. com-21, no. 11, November 1973. [6] Morais, Douglas H., Fixed Broadband Wireless Communications: Principles and Practical Applications, 1 st ed., Prentice Hall, January 2004. [7] Hassan Yaghoobi, Scalable OFDMA Physical Layer in IEEE 802.16 Wireles MAN, Intel Technology Journal, Volume 8, Issue 3, 2004. Copyright 2005 Intel Corporation. All rights reserved. Page 9 of 10