WiMAX Technology for Broadband Wireless Access 2007 phần 4 ppsx

Finally, the burst profi les of OFDM and OFDMA PHY, an important building block of IEEE 802.16 MAC layer, are described in Section 6.5.. The most robust burst profi le or, equivalently,

Trang 1

The Physical Layer of WiMAX

6.1 The 802.16 Physical Transmission Chains

The modulation and OFDM transmission aspects, described in the previous chapter, are the major building blocks of the WiMAX PHYsical Layer In this chapter, some elements of the transmission chains of WiMAX are described for both OFDM and OFDMA PHYs

6.1.1 The Global Chains

The PHY transmission chains of OFDM and OFDMA are illustrated in Figures 6.1 and 6.2 The blocks are the same with the small difference that OFDMA PHY includes a repeti-tion block The modulation is one of the four digital modulations described in the previous chapter: BPSK, QPSK, 16-QAM or 64-QAM The modulated symbols are then transmitted

on the OFDM orthogonal subcarriers In the following, WiMAX channel coding building blocks are described

The building blocks of channel coding are described in Section 6.2 A possible FEC code

is the Turbo Code Turbo Code theory and the basic elements of its use in WiMAX can be found in Section 6.3 The Transmission Convergence Sublayer (TCS), which can be applied

in OFDM PHY, is described in Section 6.4 Finally, the burst profi les of OFDM and OFDMA PHY, an important building block of IEEE 802.16 MAC layer, are described in Section 6.5

6.2 Channel Coding

The radio link is a quickly varying link, often suffering from great interference Channel coding, whose main tasks are to prevent and to correct the transmission errors of wireless systems, must have a very good performance in order to maintain high data rates The 802.16 channel coding chain is composed of three steps: randomiser, Forward Error Correction (FEC) and interleaving They are applied in this order at transmission The corresponding operations at the receiver are applied in reverse order Error detection is realised with HCS and CRC (see Chapter 8)

WiMAX: Technology for Broadband Wireless Access Loutfi Nuaymi

Trang 2

6.2.1 Randomisation

Randomisation introduces protection through information-theoretic uncertainty, avoiding long sequences of consecutive ones or consecutive zeros It is also useful for avoiding non-centred data sequenes Data randomisation is performed on each downlink and uplink burst

of data If the amount of data to transmit does not fi t exactly the amount of data allocated, padding of 0FF (‘ones’ only) is added to the end of the transmission block The Pseudo-

Random Binary Sequence (PRBS) generator used for randomisation is shown in Figure 6.3 Each data byte to be transmitted enters sequentially into the randomiser, with the Most Sig-nifi cant Byte (MSB) fi rst Preambles are not randomised The randomiser sequence is applied only to information bits

The shift-register of the randomiser is initialised for each new burst allocation For OFDM PHY, on the downlink, the randomiser is reinitialised at the start of each frame with the se-quence: 1 0 0 1 0 1 0 1 0 0 0 0 0 0 0 The randomiser is not reset at the start of burst 1 At the start of subsequent bursts (starting from burst 2), the randomiser is initialised with the vector

(CC, Turbo Code, …)

leaving Modulation

To OFDMA Part: IFFT, CP, Subchannels etc (see Chapter 5)

tition

Repe-Figure 6.2 OFDMA PHY transmission chain

15 14 13 12 11 10 9 8

MSB LSB

Figure 6.3 PRBS generator used for data randomisation in OFDM and OFDMA PHY (From IEEE

Rando-FEC encoder

(CC, Turbo Code or other)

Interleaving Modulation

To OFDM Part: IFFT,

CP, etc (see Chapter 5)

Figure 6.1 OFDM PHY transmission chain

Trang 3

shown in Figure 6.4 This PRBS generates a Pseudo-Noise (PN) sequence of length 215 1

The frame number used for initialisation refers to the frame in which the downlink burst is transmitted BSID is the BS identity and DIUC the burst profi le indicator (see Chapter 9) For other cases (uplink, OFDMA), the details can be found in the standard The bits issued from the randomiser are then applied to the FEC encoder

6.2.2 Forward Error Correction (FEC) Codes

For OFDM PHY, the FEC encodings are:

• Concatenated Reed–Solomon Convolutional Code (RS-CC) This code is mandatory on both the uplink and downlink It consists of the concatenation of a Reed–Solomon outer code and a rate-compatible convolutional inner code (see below)

• Convolutional Turbo Codes (CTC) (optional)

• Block Turbo Coding (BTC) (optional) For Turbo Coding, see Section 6.3 below

The most robust burst profi le or, equivalently, the most robust coding mode must be used when requesting access to the network and in the FCH burst (see Chapter 9 for FCH burst).For OFDMA PHY, the FEC encodings are:

• (Tail-biting) Convolutional Code (CC) This code is mandatory according to the 802.16 standard According to WiMAX profi les, only the Zero-Tailing Convolutional Code (ZT CC) is mandatory

• Convolutional Turbo Codes (CTC) This code is optional according to the 802.16 standards [1,2] Yet, according to mobile WiMAX profi les, the CTC is mandatory

• Block Turbo Coding (BTC) (optional)

• Low Density Parity Check (LDPC) codes (optional)

RS-CC encoding will now be described WiMAX Turbo coding, BTC and CTC, will be described in the following section

6.2.2.1 RS-CC (Reed–Solomon Convolution Code)

For OFDM PHY, the RS-CC encoding is performed by fi rst passing the data in block format through the RS encoder and then passing it through a convolutional encoder (see Figure 6.5)

b3

BSID

OFDM randomizer DL initialization vector

LSB MSB

Figure 6.4 OFDM randomiser downlink initialisation vector for burst 2, … ,N (From IEEE Std

Trang 4

Reed–Solomon codes are used in many communications systems and other applications The RS error correction works by adding some redundant bits to a digital data sequence This

is done by oversampling a polynomial constructed from the uncoded data The polynomial

is evaluated at several points and then these values are sent (or recorded) By sampling the polynomial more often than needed, the receiver can recover the original polynomial in the presence of a relatively low number of errors

A Reed–Solomon code is specifi ed as RS(N,K) with T-bit symbols The data points are sent

as encoded blocks The total number of T-bit symbols in an encoded block is N 2T1 Thus

a Reed–Solomon code operating on 8-bit symbols has N 281 255 symbols per coded

block The number K, K N, of uncoded data symbols in the block is a design parameter

Then, the number of parity symbols added is N K symbols (of T-bits each) The RS decoder

can correct up to (N K)兾2 symbols that contain an error in the encoded block.

The RS encoder of OFDM PHY is denoted as an (N, K) (255, 239) code, which is capable

of correcting up to eight symbol errors per block This Reed–Solomon encoding uses GF(28), where GF is the Galois Field operator The Reed–Solomon encoder and decoder require Galois fi eld arithmethics The following polynomials are used for the OFDM RS systematic code, an RS code that leaves the data unchanged before adding the parity bits:

Code generator polynomial: g(x) (x  m 0) (x m 1) (x m 2)…(x m 2T 1), m 02HEX ;

Field generator polynomial: p(x) x8 x4 x3 x2 1

The coding rate of the OFDM PHY RS encoder is then 239/255 (very close to one) The standard indicates that this code can be shortened and punctured to enable variable block sizes and variable error-correction capabilities

The convolution code has an original coding rate of 1/2, as shown in Figure 6.6 The tional encoder is a zero-terminating convolutional encoder A single 0 00 tail byte is appended

convolu-to the end of each burst, needed for decoding algorithm normal operation Puncturing patterns defi ned in the standard can be used to realise the following different code rates: 2兾3, 3兾4 and 5兾6

For OFDMA, the convolutional encoder is also the one shown in Figure 6.6 The HARQ procedure (described in Chapter 8), in its IR (Incremental Redundancy) variant, uses four different FEC blocks for each uncoded FEC block This is realised using different puncture patterns Each FEC block is identifi ed by an SPID (SubPacket IDentifi er)

The tail-biting convolutional code encoder of OFDMA (simply known as CC) works as lows: the convolutional encoder memories are initialised by the (six) last data bits of the FEC

fol-block being encoded (the packet data bits numbered b n 5,…,b n) This OFDMA PHY volutional encoder may employ the Zero-Tailing Convolutional Coding (ZT CC) technique

con-In this case, a single 0 00 tail byte is appended at the end of each burst This tail byte is

appended after randomisation

Outer code :

Reed-Solomon (RS) encoder

Inner code :

Convolutional code (CC) encoder

Data from randomiser

Encoded data

Figure 6.5 Illustration of the RS-CC encoder of OFDM PHY

Trang 5

di-• Distribute the coded bits over subcarriers A fi rst permutation ensures that adjacent coded bits are mapped on to nonadjacent subcarriers.

• The second permutation insures that adjacent coded bits are mapped alternately on to less or more signifi cant bits of the constellation, thus avoiding long runs of bits of low reliability

6.2.4 Repetition

Repetition was added by the 16e amendment for OFDMA PHY The standard indicates that it can be used to increase the signal margin further over the modulation and FEC mechanisms

In the case of repetition coding, R 2, 4 or 6, the number of allocated slots (Ns) will be a

whole multiple of the repetition factor R for the uplink For the downlink, the number of the allocated slots (Ns) will be in the range of R K, R K (R 1), where K is the number of

required slots before applying the repetition scheme For example, when the required number

of slots before the repetition is 10 ( K) and the repetition of R 6 is applied for the burst

transmission, then the number of the allocated slots (Ns) for the burst can be from 60 slots to

65 slots

X output

Y output

1 bit delay

1 bit delay Data in

Figure 6.6 Convolutional encoder of rate 1/2 (From IEEE Std 802.16-2004 [1] Copyright IEEE

Trang 6

The binary data that fi ts into a region that is repetition coded is reduced by a factor R

com-pared to a nonrepeated region of the slots with the same size and FEC code type After FEC and bit-interleaving, the data are segmented into slots, and each group of bits designated to

fi t in a slot is repeated R times to form R contiguous slots following the normal slot ordering

that is used for data mapping

This repetition scheme applies only to QPSK modulation It can be applied in all coding schemes except HARQ with CTC

6.3 Turbo Coding

Turbo codes are one of the few FEC codes to come close to the Shannon limit, the theoretical limit of the maximum information transfer rate over a noisy channel The turbo codes were proposed by Berrou and Glavieux (from ENST Bretagne, France) in 1993 The main feature

of turbo codes that make them different from the traditional FEC codes are the use of two error-correcting codes and an interleaver Decoding is then made iteratively taking advantage

of the two sources of information

Data transmission is coded as follows (see Figure 6.7) Three blocks of bits are sent The

fi rst block is the m-bit block of uncoded data The second block is n兾2 parity bits added in

sequence for the payload data, computed using a convolutional code The third subblock is

another n兾2 parity bits added in sequence for a known permutation of the payload data, also

computed using a convolutional code Hence, two different redundant blocks of parity bits

are added to the sent payload The complete block has m n bits of data with a code rate of

m 兾(m n), as shown in the fi gure.

The data decoding process is the major innovation of turbo codes The likelihood is used in order to take advantage of the differences between the two decoders The turbo code inven-tors like to make the parallel with solving crosswords through both vertical and horizontal approaches

Each of the two convolutional decoders generates an hypothesis, with derived likelihoods,

for the m-bits sequence, called the a posteriori probability (APP) The hypothesis and the

received sequence (recalculated) parity bits are compared and, if they differ, the decoder exchanges the derived likelihoods it has for each bit in the hypotheses An iterative process

is run until the two convolutional decoders come up with the same hypothesis for the m-bits

sequence The number of steps is usually of the order of 10

Coded Sequence

(m+n bits)

Convolutional Code 1

Convolutional Code 2

Trang 7

6.3.1 Convolutional Turbo Codes (CTC)

Different classes of turbo codes exist Convolutional Turbo Codes (CTC) are defi ned as tional FEC for OFDM and OFDMA PHY For OFDMA PHY, the CTC can be used for the support of the optional Hybrid ARQ (HARQ, see Chapter 8) According to mobile WiMAX profi les, the CTC is mandatory for OFDMA PHY A brief overview of the CTC defi ned for OFDMA PHY is proposed here

op-The CTC encoder, including its constituent encoder, is depicted in Figure 6.8 It uses a double binary Circular Recursive Systematic Convolutional Code The bits of the data to be encoded are alternatively fed to A and B, starting with the MSB of the fi rst byte being fed

to A The encoder is fed by blocks of k bits or N couples (k 2N bits) For all the frame sizes,

k is a multiple of 8 and N is a multiple of 4 Further, N is limited to 8 N兾4 1024.

A

CTC interleaver

Constituent encoder

C1 1

2

switch

Y1W1 Y2W2

Systematic part C2

S3

B

Figure 6.8 OFDMA PHY Convolutional Turbo Code (CTC) encoder (From IEEE Std 802.16-2004

Trang 8

The encoding block size depends on the number of subchannels allocated and the tion specifi ed for the transmission Concatenation of a number of subchannels must be per-formed in order to make larger blocks of coding where it is possible, with the limitation of not passing the largest block under the same coding rate The concatenation rule should not

modula-be used when using HARQ A table providing the nummodula-ber of subchannels concatenated as a function of the number of subchannels is given in the standard

Figure 6.9 shows a block diagram of CTC subpacket generation The CTC encoded word with a coding rate of 1兾3 goes through the interleaving block and puncturing is per-

code-formed FEC structures proposed in the standard [1] puncture the mother codeword to ate a subpacket with various coding rates: 1兾2, 2兾3, 3兾4 and 5兾6 The subpacket may also be

gener-used as HARQ packet generation (with different SPIDs) The length of the subpacket is chosen according to the needed coding rate, refl ecting the channel condition (this is link adaptation)

6.3.2 Block Turbo Codes (BTC)

Block Turbo Codes (BTC) are defi ned as an optional FEC for OFDM and OFDMA PHY The BTC is also optional in WiMAX profi les

For OFDM and OFDMA PHY, the BTC is based on the product of two simple component codes, which are binary extended Hamming codes or parity check codes The codes are not the same for the two PHYs BTC component codes of OFDM are shown in Table 6.1 The

Table 6.1 BTC component codes of OFDM PHY (From IEEE Std

Figure 6.9 Block diagram of subpacket generation (From IEEE Std 802.16-2004 [1] Copyright IEEE

1/3 CTC Encoder

Interleaver

Puncturing block

Trang 9

component codes are used in a two-dimensional matrix form, which is depicted in Figure

6.10 The k x information bits in the rows are encoded into n x bits by using the component

block (n x , k x) code specifi ed in the standards for the respective composite code After

encod-ing the rows, the columns are encoded usencod-ing a block code (n y , k y), where the check bits of the

fi rst code are also encoded The overall block size of such a product code is n n x n y, the

total number of information bits k k x k y and the code rate is R R x R y , where R i k i /n i,

i x, y Data bit ordering for the composite BTC matrix is defi ned such that the fi rst bit in

the fi rst row is the LSB (Least Signifi cant Byte) and the last data bit in the last data row is the MSB

To match a required packet size, BTCs may be shortened by removing symbols from the BTC array In the two-dimensional case, rows, columns, or parts thereof, can be removed until the appropriate size is reached

6.4 Transmission Convergence Sublayer (TCS)

The Transmission Convergence Sublayer (TCS) is defi ned in the OFDM PHY Layer and the Non-WiMAX SC PHY Layer The TCS is located between the MAC and PHY Layers If the TCS is enabled, the TCS converts MAC PDUs of variable size into proper-length FEC blocks, called TC PDU An illustration of a TC PDU is shown in Figure 6.11 A pointer byte is added

at the beginning of each TC PDU, as illustrated in the fi gure This pointer indicates the header

of the fi rst MAC PDU

The TCS is an optional mechanism for the OFDM PHY It can be enabled on a burst basis for both the uplink and downlink through the burst profi le defi nitions in the uplink and downlink channel descriptor (UCD and DCD) messages respectively The TCS_ENABLE parameter is coded as a TLV tuple in the DCD and UCD burst profi le encodings (see Chapters 8 and 9 for TLV and UCD/DCD) At SS initialisation, the TCS capability is negotiated between the BS and SS through SBC-REQ/SBC-RSP MAC mes-sages as an OFDM PHY specifi c parameter The TCS is not included in the OFDMA PHY Layer

pre-Figure 6.10 BTC and shortened BTC structure (From IEEE Std 802.16-2004 [1] Copyright IEEE

n x

n y

k x

k y

Trang 10

6.5 Burst Profi le

The burst profi le is a basic tool in the 802.16 standard MAC Layer The burst profi le tion, which changes dynamically and possibly very fast, is about physical transmission Here the parameters of the burst profi les of WiMAX are summarised The burst profi les are used for the link adaptation procedure The use of burst profi les and the link adaptation procedure will be seen in more detail in Chapters 9 and 10

alloca-6.5.1 Downlink Burst Profi le Parameters

The burst profi le parameters of a downlink transmission for OFDM and OFDMA PHYsical

layers are proposed in Table 6.2 The parameter called FEC code is in fact the Modulation

and Coding Scheme (MCS) For OFDM PHY, there are 20 MCS combinations of tion (BPSK, QPSK, 16-QAM or 64-QAM), coding (CC, RS-CC, CTC or BTC) and coding rate (1/2, 2/3, 3/4 and 5/6) The most frequency-use effi cient (and then less robust) MCS

modula-Figure 6.11 Format of the downlink Transmission Convergence sublayer PDU (From IEEE Std

Transmission Convergence sublayer (TC) PDU

Second MAC PDU that starts

in this TC packet

P = 1 Byte pointer field

Table 6.2 Downlink burst profi le parameters for OFDM and OFDMA PHYsical layers

MCSs in OFDM PHY and 34 MCSs in OFDMA PHY (as updated in 802.16e)

DIUC mandatory exit threshold The CINR at or below where this burst profi le can no

longer be used and where a change to a more robust (but also less frequency-use effi cient) burst profi le is required Expressed in 0.25 dB units See Chapter 9 for DIUC

DIUC minimum entry threshold The minimum CINR required to start using this burst

profi le when changing from a more robust burst profi le Expressed in 0.25 dB units

TCS_ enable (OFDM PHY only) Enables or disables TCS

Trang 11

is 64-QAM (BTC) 5/6 For OFDMA PHY, there are 34 MCS combinations of modulation (BPSK, QPSK, 16-QAM or 64-QAM), coding (CC, ZT CC, CTC, BTC, CC with optional in-terleaver) and coding rate (1/2, 2/3, 3/4 and 5/6) The Downlink Interval Usage Code (DIUC)

is the burst usage descriptor (see Chapter 9), which includes the burst profi le

6.5.2 Uplink Burst Profi le Parameters

The burst profi le parameters of an uplink transmission for an OFDM PHY and an OFDMA PHY are proposed in Tables 6.3 and 6.4 respectively

6.5.3 MCS Link Adaptation

The choice between different burst profi les or, equivalently, between different MCSs is a powerful tool Specifi cally, choosing the MCS most suitable for the state of the radio chan-nel, at each instant, leads to an optimal (highest) average data rate This is the so-called link

Table 6.3 Uplink burst profi le parameters for the OFDM PHYsical Layer

FEC type and modulation type There are 20 MCSs in OFDM PHY

Focused contention power boost The power boost in dB of focused contention

carriers (see Chapter 10)

Table 6.5 Received SNR threshold assumptions [1], Table 266 (From

Table 6.4 Uplink burst profi le parameters for the OFDMA PHYsical Layer

FEC type and modulation type There are 52 MCSs in OFDMA PHY

Ranging data ratio Reducing factor, in units of 1 dB, between the power used

for this burst and the power used for CDMA ranging (see Chapter 11); encoded as a signed integer

Trang 12

adaptation procedure In the following chapters the MAC procedures that can be used for the implementation of link adaptation are described The link adaptation algorithm in itself is not indicated in the 802.16 standard It is left to the vendor or operator.

The order of magnitudes of SNR thresholds can be obtained from Table 6.5, proposed in the standard [1] for some test conditions These SNR thresholds are for a BER, Bit-Error Rate, measured after the FEC, that is smaller than 10–6

Trang 13

Part Three

WiMAX Multiple Access (MAC

Layer) and QoS

Management

Trang 14

sub-of WiMAX/802.16 Among other functions sub-of the CS is the optional Payload Header pression (PHS), the process of suppressing repetitive parts of payload headers at the sender and restoring the headers at the receiver The classifi cation and mapping made by a QoS management module allow full advantage to be taken of the different PHYsical layer features presented in the two previous chapters of this book.

Sup-In the present version of the 802.16-2004 standard, two CS specifi cations are provided and described in Section 5 of standards [1] and [2] The fi rst CS specifi cation is the ATM CS The Asynchronous Transfer Mode (ATM) CS is a logical interface that associates different ATM services with the MAC CPS SAP The ATM CS accepts ATM cells from the ATM layer, performs classifi cation and, if provisioned, PHS Then the ATM CS delivers CS PDUs to the appropriate MAC SAP

The other available CS specifi cation is the packet CS The packet CS is used for the port of all packet-based protocols such as the Internet Protocol (IP), IPv4, IPv6, Point-to-Point Protocol (PPP) and the IEEE standard 802.3 (Ethernet) Classifi cation and, if provisioned, PHS are also defi ned for the packet CS

trans-The standard states that other CSs may be specifi ed in the future For the moment, no implementation of the ATM CS is planned, although it is detailed in the standard In the rest

of this chapter only the packet CS will be considered

7.2 Connections and Service Flow

The CS provides any transformation or mapping of external network data received through the CS Service Access Point (SAP) into MAC SDUs received by the MAC Common Part Sublayer (CPS) through the MAC SAP (see Figure 7.1) This includes classifying external

Định dạng
Số trang	29
Dung lượng	310,25 KB