Waqas Ali Khan, Talha Ali Khan, Muhammad Arif Ali and Shahid Abbas Final Year Students of Electronics Department NED University of Engineering & Technology Karachi, Pakistan (waqas035, talha080, arifali137, shahid.nedian)@hotmail.com AbstractWireless Personal Area Network ( WPAN ) system with ranges of 4-10 m become popular in replacing wire system, technologies such as Bluetooth and ZigBee used for this purpose but due to its low data rate of 1 Mbps which is not enough for large file transfer and applications like video streaming there is a need of technology which address all these issues, Ultra Wideband is such a emerging wireless technology which support data rates up to 0.5 Gbps. UWB with its Multi-Band Orthogonal Frequency Division Multiplexing ( MB-OFDM ) infrastructure is an ultimate choice for the transmission of high speed data by dividing its wide frequency spectrum into sub 14-bands. The architecture of MB-OFDM transmitter baseband is presented in this paper and its modeling in MATLAB is discussed in detail. Modeling of UWB PHY transmitter baseband in MATLAB is helpful for designing its commercial product by optimizing various parameters. I. INTRODUCTION In February 2002, the Federal Communications Commission (FCC) allocated a large spectrum from 3.1 GHz to 10.6 GHz for use of commercial applications of UWB. Since then great efforts are made in standardization of UWB devices. The IEEE 802.15.3 Task Group ( TG3a ) is set for proposing final draft for UWB standards, the committee finals two drafts one proposed UWB using direct sequence CDMA supported by Xtreme Spectrum and Deca Wave and other is UWB using MB-OFDM base structure supported by Intel and Texas but deadlock arises between these two. So group of more than 170 companies which includes Intel, Texas and Motorola and many others forms a forum with the name of WiMedia Alliance for the development of UWB system using multi-band ofdm structure which announces its High Rate Ultra Wideband MAC and PHY standards which is also supported by ECMA by releasing two international ISO based specifications ECMA 368 and ECMA 369 for Ultra Wideband technology. Also, in 2007 IEEE 802.15 Task Group (TG4a) approved a standard which support UWB infrastructure using Direct sequence CDMA spread spectrum approach. Ultra Wideband with its absolute bandwidth of more than 500 MHz and with its low transmitting power provides wealth of advantages better than any communication system because these two features provides high accuracy and friendly coexistence with other communication system, to fulfill these requirements many wireless protocols have been designed but Orthogonal Frequency Division Multiplexing (OFDM) has become the ultimate choice for modulation scheme for high data rate digital transmission due to its spectral efficiency and robustness against multipath interference. In multi-band OFDM approach the large swaths of ultra wideband frequency spectrum is divided into fourteen smaller subbands, each of 528 MHz which is further grouped into five bands each of which comprises of three subbands and the last one consists of two subbands as see in Fig. 1. The MB-OFDM UWB system also supports multiples modes of operation: a mandatory mode in which first three subbands are included (Mode 1) and an optional mode which consist of seven subbands (Mode 2). The remaining paper is organized as follows Section 2 provides a brief overview on the architecture of the baseband portion of MB-OFDM transmitter. The modeling of building blocks of MB-OFDM transmitter is discussed in detail in Section 3. Finally some concluding remarks and future work is discussed in Section 4.
II. ARCHITECTURE OF MB-OFDMTRANSMITTER BASEBAND Basic building blocks of MB-OFDM transmitter baseband is shown in Figure 2. Data from MAC is coming to side stream scrambler which function isto randomizes the data stream to remove repeated patterns and spaces, after which incoming data are encoded to reduce the bit error rate. The main purpose isto correct the error that are arises due to the channel, convolutional encoder is suitable for this purpose with coding rate of 5/8 which is acquired by puncturering some data. The stream of data then shall be interleaved prior to modulation to provide robustness against burst of errors. After coding and interleaving serial binary data divided into group of two bits which is converted into complex valued constellation points which becomes input to the IFFT which is the core of baseband portion of transmitter. A 128 point IFFT is used to create 122 subcarriers out of which 12 are pilot subcarriers, 100 data subcarriers and 10 are guard subcarriers, five at each end. The rest six are set to null subcarriers. To compensate for multipath propagation and to eliminate ISI a Zero pad suffix of length 37 is added. Table 1. summarizes the parameters related to ultra wideband MB-OFDM modulation. 2009 International Conference on Emerging Technologies 978-1-4244-5632-1/09/$26.00 2009 IEEE 194 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply. Figure 1. Division of UWB Frequency Spectrum III . TRANSMITTER BASEBAND DESIGN IN MATLAB Matlab is a high level technical computing language which provide interactive envoirment for algorithm development and modeling of communication system. Also integration of Simulink with Matlab offer various tools for simulating and testing any communication system. UWB with transfer data rate of 200 Mbps is modeled. Below is the description of the UWB Model proposed in Simulink. A. Input of the Transmitter / Scrambler Input source for the scrambler is the data from MAC Layer which includes MAC Header , HCS and MAC frame body, since the coming data is in the pattern of 1s and 0s so it can be modeled by any data source which can create a pattern of 1s and 0s. A Maximum Length Sequence (MLS) is used for this purpose which is the type of pseudorandom binary sequence. They are polynomial rings generated using maximal linear feedback shift registers and are so called because they are periodic and reproduce every binary sequence that can be reproduced by the shift registers. The polynomial used for this purpose is defined by expression as follows; f(x) = x 10 + x 3 + 1 (1)
In Matlab there is a builtin block of PN SEQUENCE GENERATOR which has a parameters of generator polynomomial and initial states of shift registers which in our case become [ 1 0 0 0 0 0 0 1 0 0 1 ] and [ 0 0 0 0 0 0 0 0 0 1]. B. Scrambler A scrambler (often referred to as a randomizer) is a device that manipulates a data stream before transmitting. The purpose of scrambling is to eliminate the dependence of a signals power spectrum upon the actual transmitted data and making it more disperse to meet maximum power spectral density requirements, because if the power is concentrated in a narrow frequency band, it can interfere with adjacent channels. A side-stream scrambler shall be used to whiten only portions of the PLCP header , i.e., the MAC header and HCS, and the entire PSDU. In addition, the scrambler shall be initialized to a seed value specified by the MAC at the beginning of the MAC header and then re-initialized to the same seed value at the beginning of the PSDU. The polynomial generator, g(D), for the pseudo-random binary sequence (PRBS) generator shall be; g(D) = 1 + D 14 + D 15 (2) Where D is a single bit delay element and integers in the power of D represent connection of shift register with modulo 2 adder. Using this generator polynomial, the corresponding PRBS, x n , is generated as;
x n = x n-14 x n-15 (3) The scrambled data bits s n are obtained as follows;
s n = b n x n (4) where b n are unscrambled data bits. In Matlab there is a block of scrambler with a parameter of scramble polynomial and initial states. Scramble polynomial is defined by 2 and the initial states represents the states of scrambler register when simulation starts. C. FEC Encoder ( Convolutional Encoder ) Convolutional Coding is a part of the Forward Error Correction (FEC) done in communication systems. The purpose of forward error correction (FEC) is to improve the capacity of a channel by adding some carefully designed redundant information to the data being transmitted through the channel. The process of adding this redundant information is known as channel coding . Convolutional codes operate on serial data, one or a few bits at a time. The two parameters which defined convolutional codes are : the code rate and the constraint length, the code rate is defined as the ratio of number of bits into the convolutional encoder to the number of channel symbols output by the convolutional encoder in a given encoder cycle. The constraint length, k, denotes the length of the convolutional encoder. Figure 2. Basic Architecture of MB-OFDM Transmitter Baseband Scrambler FEC Encoder Puncturer Interleaver Mapper OFDM Modulator (IFFT & ZPS ) 195 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply. TABLE 1. PARAMETERS OF MB-OFDM MODULATION Parameters Values NSD : Number of Data Subcarrier 100 NSP : Number of Pilot Subcarrier 12 NSG : Number of Guard carriers 10 NST : Number of total Subcarrier 122 Number of Null carriers used 6 Number of Samples per zero padding suffix 37 Total Number of samples per symbol 165 Subcarrier Modulation QPSK TSYM : Symbol Interval 312.5 ns Code Rate 1/3. 1/2, 5/8, 3/4, 11/32 In the proposed design of our model of UWB for 200 Mbps according to the defined standard the code rate is 5/8 which is obtained by combining five encoder with a rate of /3 (known as mother rate) and by puncturering some output bits. The constraint length parameter k is 7. The generator polynomial define for UWB are g 0 = 133 8 , g 1 = 165 8 and g 2 = 171 8 , these numbers are in octet system which is converted to binary number system where in binary number if bit is 1 then it shows the connection of that register with modulo addition and if bit is 0 then register is not connected with shift registers as shown in Figure 3. The bit denoted as A shall be the first bit generated by the encoder, followed by the bit denoted as B, and finally, by the bit denoted as C. The Convolutional Encoder block in Matlab has a parameter of Trellis structure which is define by poly2trellis command which has two arguments constraint length and code generator polynomial which is defined above. Poly2trellis command create a trellis diagram which is best representation for convolutional encoding. The operation mode is selected Terminate trellis by appending bitsby pop down menu which terminate the trellis at the all zero state by appending tail bits at the end of each input frame. D. Puncturer The principle of puncturing is a process by which a few parity bits are deleted in order to improve the code rate. In Matlab the puncture block creates an output vector by removing selected elements of the input vector and preserving others. The input can be a real or complex vector of length K. The block determines which elements to remove or preserve by using the binary Puncture vector parameter. If Puncture vector (k) = 0, then the kth element of the input vector does not become part of the output vector and If Puncture vector (k) = 1, then the kth element of the input vector is preserved in the output vector. Since the input is frame base therefore puncture vector must be multiple of it. Figure 3. Convolutional encoder : rate R = 1/3, constraint length k = 7 E. Interleaver In MB-OFDM system we perform two types of interleaving: symbol interleaving and tone interleaving as see in Figure 4. In Symbol interleaving bits of stream are interleaved over three OFDM symbol to acquire different level of frequency diversity because each OFDM symbol is transmitted in different frequency bands. The symbol interleaver works on 3N CBPS bits where N CBPS is the number of coded bits per OFDM symbol which in our case is 200. If A(i) and B(j) represent the input and output bits of the symbol interleaver respectively at position j, the relationship between the two is given by; B(j) = A{ floor( i/N CBPS ) + 3mod( i, N CBPS ) } (5) Where i, j = 0, 1, 2 .N CBPS . The output B(j) from symbol interleaver is passed through tone interleaver. Tone interleaver is then used to interleave bits in each OFDM symbol to exploit frequency diversity across tones and provide robustness against narrow band interferers. If C(i) and D(j) represents inputs and outputs of tone interleaver respectively, the relationship between them is given by; D(j) = C{ floor( i/N ) + 10mod( i, N ) } (6) Where N = N CBPS /10 i, j = 0, 1, 2, N CBPS Matrix interleaver blocks has a parameters of number of columns and number of rows which in our case become 100 and 3 respectively for symbol interleaver and for tone interleaver it becomes 10 and 10. Buffer block is also used to scale up the matrix size at the beginning of symbol interleaver and at the end of matrix interleaver, buffer block is also used in the mid position to scale down the dimension of data matrix to make correct input dimensions for tone interleaver. F. SubCarrier Modulator (QPSK Modulator Baseband) For data rates less than or equal to 200 Mbps preferred subcarrier modulation is QPSK and for data rate greater than 200 Mbps Dual carrier modulation (DCM) is preferred. Since our proposed UWB system is for 200 Mbps so we use QPSK baseband modulator. The serial binary data from the interleaver shall be grouped into two bits and converted into complex numbers representing QPSK Constellation points. The conversion shall be performed according to the Gray- coded constellation mappings. The values, d, are formed by multiplying the resulting (I + jQ) value by a normalization factor of KMOD, as described in (7); Figure 4. A block diagram of the various stages of the bit interleaver 196 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply. d = ( I + jQ )*K MOD (7) where K MOD = Normailzation Factor = 1 / 2 For QPSK, b o determines the I value and b 1 determines the Q value. Matlab has builtin QPSK baseband modulator block with parameters Input type, Constellation ordering and phase offset which we set Bit , Gray and pi/4 respectively in our case. G. OFDM Modulator This is the core block of MB-OFDM Transmitter Base band. In this block construction of an OFDM signal and IFFT operation takes place. OFDM Signal can be described by the (8); S n [k]=1/N FFT [ C D,n [l]exp(j2*pi*M D [l]k/N FFT )+ C G,n [l] exp(j2*pi*M G [l]k/N FFT )+ C P,n [l]exp(j2*pi*M P [l]k/N FFT )] (8) where limit of first is [0,N SD 1] , limit of second is [0, N SG 1] and the last one has a limit [0,N SP 1] and k [0, N FFT 1]. C D,n [l] , C G,n [l] and C P,n [l] are the complex numbers placed on the l th data,guard and the pilot subcarriers of n th OFDM symbol and M D [l],M G [l] and M P [l] define a mapping from the indices [0,N SD 1], [ 0, N SG 1] and [0,N SP 1] to the logical frequncy offset indices as defined in Table 3. In baseband portion data is transmitted from Physical Service Data Unit (PSDU) to PLCP Protocol Data Unit (PPDU), which comprises of PLCP Preamble, PLCP Header and PSDU, PLCP Header and PSDU is modeled as the input source of the transmitter because they need scrambling now we modeled PLCP Preamble which comprises of Packet Synchronization Sequence, Frame Synchronization Sequence and Channel Estimation Sequence which is modeled in Matlab by creating a matrix of [123x9], [128x3] and [128x6] with values defined in [5]. After constellation mapping of a signal its enter into the Modulator Subsystem where its first passes through the Group data for OFDM Symbol block which reshape the signal after then its indices is converted into logical frequency offset with the help of SELECT ROW block, also Pilot signal and Null carriers are added to it which is modeled by using DSP CONSTANT block and their values are defined in the parameters of the block, all these signals are TABLE 2. QPSK CARRIER MAPPING Input bit ( b0b1 ) I-out Q-out 00 -1 -1 01 -1 1 10 1 -1 11 1 1 then Concatenate into single dimension by using MATRIX CONCATENATE block. Channel Estimation Sequence block is then concatenate with the output of the MATRIX CONCANTENATE block. After the construction of signalinverse fast fourier transform is applied by using builtin block of the IFFT in Matlab, Packet Synchronization Sequence and Frame sequence is added concatenate with the output of the IFFT by again using MATRIX CONCATENATE block. Zero padding suffix and guard interval is added in the last so we get a [ 165 x 1 ] length of an OFDM signal which is our requirement. TABLE 3. FREQUNCY OFFSET INDICES H. Simulation Results Transmitted OFDM signal can be viewed using SCOPE block whose snapshot is shown in Fig. 6.
IV . CONCLUSION In this model, we proposed architecture of MB-OFDM UWB baseband transmitter. Power Spectrum of Transmitter shows in Figure 6. that it fulfill the FCC standards for UWB System. By changing code rate and subcarrier modulator block, we can model this system for various data rates for M D [l] M G [l] M P [l] l 56 ; l = 0 -61 + l ; l =0.4 -55 + 10*l ; l = 0, 1, 2, 3,,11 l 55 ; 1 l 9 52 + l ; l =5.9 l 54 ; 10 l 18 l 53 ; 19 l 27 l 52 ; 28 l 36 l 51 ; 37 l 45 l 50 ; 46 l 49 l 49 ; 50 l 53 l 48 ; 54 l 62 l 47 ; 63 l 71 l 46 ; 72 l 80 l 45 ; 81 l 89 l 44 ; 90 l 98 l 43 ; l = 99 Figure 5. Implementation of Bit Interleaver in Matlab 197 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply. UWB system. Thus this model provides good basis for implementing an ASIC for proposed design in future. REFRENCES [1] D. M. W. Leenaerts. (2006). Transceiver Design for Multiband OFDM UWB. EURASIP Journal on Wireless Communications and Networking . [2] MATLAB 7, The Language of Technical Computing. (n.d.). Retrieved from http://www.mathworks.com/applications/tech_computing/technical_liter ature.html [3] A Tutorial on Convolutional Coding with Viterbi Decoding, Spectrum Applications. (n.d.). Retrieved from http://home.netcom.com/~chip.f/viterbi/tutorial.html [4] A. Batra et al. (March2003). Multi-band OFDM Physical Layer Proposal for IEEE 802.15 Task Group 3a. [5] Fixed-Point Modeling in an Ultra Wideband (UWB) Wireless Communication System. (2004, May ). MATLAB Digest . [6] Fredrik Kristensen, Peter Nilsson, Anders Olsson. (n.d.). Flexible baseband transmitter for OFDM. Retrieved from www.imec.be/pacwoman/publications/WP8- Lund-CSS03-11-07-2003- V1.0.pdf [7] ( 3rd Edition , December 2008.). High Rate Ultra Wideband PHY and MAC Standard [8] Implementing WiMax OFDM Timing and Frequency Offset Estimation in Lattice FPGAs. (2005). White papers . [9] Maximum Length Sequence. (n.d.). Retrieved 2009, from wikipedia.org: http://en.wikipedia.org/wiki/Maximum_length_sequence [10] Min-Su Kim, HoJun Kim, Jong Tae Kim,. (2008). High-level Modeling of UWB PHY for IEEE 802.15.4a. International Confrence and Hybird Information Technology. [11] Nishant Kumar. (2004). MAC and PHY Layer Design for Ultra- WideBand Communications. Master Thesis, Virginia Polytechnic Institute and State University, Electrical and Computer Engineering. [12] Xu Jinsong, Lu Xiaochun, Wu Haitao, Bian Yujing, Zou Decai, Zou Xiaolong, Wang Chaogang. (2008). Implementation of MB-OFDM Transmitter Baseband Based on FPGA. Circuits and Systems for Communications, 2008. ICCSC 2008. 4th IEEE International Conference on, (pp. 50-54). Shanghai. 198 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply. Figure 6. OFDM Transmitted Signal 199 Authorized licensed use limited to: NED UNIV OF ENGINEERING AND TECHNOLOGY. Downloaded on June 07,2010 at 10:35:18 UTC from IEEE Xplore. Restrictions apply.