Broadband Wireless Access (BWA) enables a variety of applications ranging from high bandwidth backhaul for data networks to simple wireless DSL for residential customers. Although this is a large and diverse market, most of these wireless systems are implemented using proprietary or vendor specific technologies. There is almost no inter-operability that would allow networks to use base stations from one vendor, and subscriber stations from another. This has significantly limited network operator choices and is one reason this market has had relatively low growth compared with other wireless technologies, like cellular and wireless LAN. The recent introduction of a new IEEE standard, 802.16-2004, and the formation of an industry group, the WiMAX Forum, is about to energize this market. Inter-operability, operator choices and lower price points will create growth opportunities to extend BWA into many new applications. As part of the new 802.16 standard, a new physical layer air interface has been defined that uses 256 carrier orthogonal frequency-division multiplex (OFDM). This article provides an overview of this new physical layer and a detailed technical explanation of some of the key building blocks in the channel structure.
The 802.16-2004 standard describes four different air interfaces. One of these interfaces is optimized for non-line-of-sight, RF frequencies less than 11 GHz and distances up to 30 km. Although the standard has officially named this PHY layer as WirelessMAN-OFDM, many people refer to it as the WiMAX air interface. The basic characteristics of the air interface are 256 carrier OFDM, bandwidths that range from 1.25 to 20 MHz and carrier frequencies up to 11 GHz.
WiMAX Air Interface
The basic OFDM symbol is based on a 256 point FFT; however, in WiMAX, only 200 of the subcarriers are actually used. The outermost subcarriers are not used to allow for guard bands and natural decay of the subcarrier energy. Also, the center frequency carrier is set to zero and not used. The center carrier is susceptible to interference caused by RF carrier feed-through. The remaining 200 carriers are allocated as 192 carriers for data and 8 carriers as pilots (see Figure 1). The pilot carriers are always BPSK modulated and the data carriers are BPSK, QPSK, 16QAM, or 64QAM.
Fig. 1 OFDM subcarriers.
The system can be configured to use any bandwidth from 1.25 to 20 MHz and, regardless of the bandwidth, the symbols always contain 200 carriers. For narrow bandwidth systems, this implies that the subcarriers are very closely spaced, which provides a relatively long symbol period (symbol period is defined as 1/subcarrier spacing). These closely spaced subcarriers and long symbols help overcome channel impairments such as multipath. This long symbol period is a key differentiator between WiMAX systems and Wireless LAN systems (relatively short symbols), which provides WiMAX with significant advantages for long distances and non-line-of-sight applications.
WiMAX systems can be deployed as TDD, FDD, or half-duplex FDD. Figure 2 shows a typical frame in a TDD configuration where the base station and subscriber equipment each transmit on the same RF frequency, separated in time. The base station transmits a downlink subframe, followed by a short gap called TTG, and then individual subscribers transmit the uplink subframes. The subscribers are accurately synchronized such that their transmissions do not overlap each other as they arrive at the base station. Following all uplink subframes, another short gap called RTG is allocated before the base station can again start transmitting. Notice that each uplink subframe is preceded by a preamble. This is called a “short preamble” and allows the base station to synchronize on each individual subscriber.
Fig. 2 Downlink and uplink subframes.
Looking closer at the downlink, a downlink subframe always begins with a preamble, followed by a header, and one or more downlink bursts of data. These downlink bursts are generally made up of multiple symbols within the burst. Within each burst, the modulation type is constant; however, from burst to burst the modulation type can change. Bursts using robust modulation types such as BPSK and QPSK are required to be transmitted first, followed by less robust modulation types (16 and 64QAM). Downlink subframes containing all four types of modulation would need to be in order as BPSK followed by QPSK, 16QAM, and finally 64QAM.
Every transmission, on both the uplink and downlink, always begins with a preamble. This preamble allows receivers to synchronize with the transmitter and is used for channel estimation. The downlink transmission begins with a long preamble. The long preamble (see Figure 3) is made up of two symbols of QPSK modulation. The first symbol uses 50 of the available 200 carriers (every fourth subcarrier) and the second symbol uses 100 of the 200 carriers (all the even-numbered subcarriers). These preamble symbols are transmitted with 3 dB more power than all other symbols in the downlink subframe, making them easier for the receivers to demodulate and decode correctly. A “short preamble” is used at the beginning uplink bursts. The short preamble is a single symbol of 100 QPSK carriers (all the even-numbered subcarriers). When using extremely long downlink bursts that contain many symbols, it may be desirable to insert a midamble (short preamble) in between the downlink bursts. This short preamble helps receivers resynchronize and perform additional channel estimation.
Fig. 3 Long preamble.
Following the preamble is a frame control header (FCH). This FCH is implemented as a single symbol of BPSK modulation. This symbol contains 88 bits of overhead data that describes critical system information such as base station ID and the downlink burst profile that receivers need to decode the subframe. The FCH does not contain enough information to fully describe the network or downlink profile, but it does contain enough that receivers can start decoding the downlink bursts.
The downlink bursts contain user data as well as control and MAC level messages. The downlink bursts each contain one or more symbols. Each symbol in the burst contains between 11 and 107 bytes of payload data, depending on the modulation type and coding gain. Table 1 shows the seven different combinations of modulation type and coding gain. For each of these combinations a specific amount of payload data is required for each symbol.
The coding process to get from payload data to actual bits sent to the IQ mapper is shown in Figure 4. When necessary, padding adds bits so that the payload data is in block sizes of 11, 23, 35, 47, 71, 95, or 107 bytes. The randomizer XORs the data with a pseudo random bit sequence to toggle some of the 1s to 0s and some 0s to 1s. This randomizer eliminates long strings of 1s or 0s in the payload data. A single tail byte is added and the bits are ready for Reed-Solomon and Convolutional Coding. These coding steps provide forward error correction and are very common coding methods used in digital communication systems. This coding adds redundant data that helps identify and fix bits that are missing or corrupted.
Fig. 4 Channel coding
The final steps in coding involve interleaving, which is performed in two steps. The first step of interleaving is to rearrange the ordering of the bits to make certain that adjacent bits are not mapped onto adjacent carriers. This helps eliminate errors by reducing the chance that adjacent bits would be lost if a portion of the channel bandwidth is degraded with some type of spurious or band-limited noise. The second step of interleaving is to reorder the bits so that the original adjacent bits are alternately mapped into more or less reliable points on the IQ constellation. In complex modulation like 64QAM, each IQ point represents multiple bits of data, and some of these bits are easier to detect (and therefore more reliable) than other bits. After interleaving, the coded bits are mapped to the IQ constellation, starting with carrier number –100 on up to carrier number +100.
To simplify transmitter and receiver designs, all symbols in the FCH and DL data bursts are transmitted with equal power. Because symbols use four different modulation types (BPSK, QPSK, etc.), it is necessary to scale each such that the average symbol power from each symbol is approximately equal. Figure 5 shows an actual measured IQ constellation of a single frame that contains symbols of BPSK, QPSK, 16QAM and 64QAM. This diagram shows each modulation type is scaled differently, and because the individual IQ points do not align, it is possible to see all 86 discrete IQ points (64QAM+16QAM+4 QPSK+2 BPSK). Measurements like this help designers quickly identify trouble areas with amplitude scaling or IQ modulation. Recall that the preamble bursts are 3 dB higher than these FCH and downlink burst symbols. The preamble is decoded and used for channel estimation as well as the reference for these IQ measurements. However, they are not shown in this display of the IQ constellation.
Fig. 5 Agilent 89600 IQ measurement of WiMax downlink frame.
Conclusion
The IEEE 802.16 technical groups and the WiMAX Forum have been working together to define an air interface optimized and scalable for a wide range of point-to-point and point-to-multipoint applications. This air interface leverages well know techniques of OFDM modulation and channel coding that provides a robust system that can be implemented in a variety of price and performance configurations. The air interface is the basic first step in device inter-operability that should enable higher volumes, lower prices and significant growth in the BWA market.
This technical overview of the air interface has hopefully prepared the reader with an overall understanding of the system and key building blocks, and should prepare them for a more detailed dive into specific elements of the implementation.
Allen Henley is a strategic planning engineer in Agilent’s test and measurement group. He provides technical and business leadership for emerging new technologies and is currently Agilent’s representative in the WiMAX Forum.
