Theory and applications of ofdm and cdma wideband wireless communications phần 7 ppsx

246 OFDM Bit index 1/Code rate Subband samples Scale factors code rate 8/18 code rate 8/14 code rate 8/24 Header CRC, PAD code rate 8/19 Figure 4.81 Example for an error protection profile for the audio data rate 192 kbit/s. by far the biggest one. The first bits inside a frame are the header, the bit allocation (BAL) table, and the scale factor select information (SCFSI). An error in this group would make the whole frame useless. Thus, it is necessary to use a strong (low-rate) code here. The next group consists (mainly) of scale factors. Errors will cause annoying sounds (so-called birdies), but these can be concealed up to a certain point on the audio level if they are detected by a proper mechanism. The third group is the least sensitive one. It consists of subband samples. Subband sample errors cause a kind of gurgling sound. Often this will not even be noticed in a noisy car environment. A last group consists of programme-associated data (PAD) and the cyclic redundancy check (CRC) for error detection in the scale fac- tors (of the following frame). This group requires approximately the same protection as the second one. The distribution of the redundancy over the audio frame defines such an error protection profile. For DAB audio transmission, 64 different protection profiles have been specified (ETS 300 401) that correspond to different audio data rates from 32 kbit/s and 384 kbit/s and allow 5 different protection levels from PL1 (the strongest) to PL5 (the weakest) corresponding to five average code rates. Each of them requires (approximately) the same SNR for distortion-free audio reception. Table 4.4 gives the detailed definition of the protection profile corresponding to Figure 4.81. The last column shows the number of encoded bits. Note that for each frame, the trellis will be closed by tail bits. These are Table 4.4 Example for an error protection profile profile (PL3) for the audio data rate 192 kbit/s Audio data bits Code rate Encoded bits Group 1 352 R c = 8/24 1056 Group 2 768 R c = 8/18 1758 Group 3 3392 R c = 8/14 5936 Group 4 96 R c = 8/19 228 Tail bits 6 R c = 8/16 12 OFDM 247 always six zero bits that are encoded by R c = 1/2. In this example, the total number of encoded bits per frame is 8960. This corresponds to 140 capacity units of 64 bits (see the following table). For data transmission, eight different protection levels with equal error protection (EEP) have been specified with code rates R c = 1/4, R c = 3/8, R c = 4/9, R c = 1/2, R c = 4/7, R c = 3/4, and R c = 4/5. The code rates 3/8 and 3/4 are constructed by a composition of two adjacent RCPC code rates. The EEP protection profiles allow fixed data rates that are integer multiples of 8 kbit/s or 32 kbit/s. The paper (Hoeher et al. 1991) gives some insight into how the channel coding for DAB audio has been developed. It reflects the state of the research work on this topic a few months before the parameters were fixed. We finally note that the UEP protection profiles for audio have been designed in such a way that one has a kind of graceful degradation. This means that if the reception becomes worse, the listener first hears the gurgling sound from the sample errors before the reception is lost. These errors can be noticed at a BER slightly above 10 −4 with headphones in a silent environment. In the noisy environment of a car, up to 10 −3 may be occasionally tolerated. Multiplexing All the UEP and EEP channel coding profiles are based on a frame structure of 24 ms. These frames are called logical frames. They are synchronized with the transmission frames, and, for audio data subchannels, with the audio frames. At the beginning of one logical frame, the coding starts with the shift registers in the all-zero state. At the end, the shift register will be forced back to the all-zero state by appending six additional bits (tail bits) to the useful data for the traceback of the Viterbi decoder. After encoding, such a 24 ms logical frame builds up a punctured code word. It always contains an integer multiple of 64 bits, which is an integer number of CUs. Whenever necessary, some additional puncturing is done to achieve this. A data stream of subsequent logical frames that is coded independently of other data streams is called a subchannel. For example, an audio data stream of 192 kbit/s is such a possible subchannel. A PAD data stream is always only a part of a subchannel. After the channel encoder, each subchannel will be time-interleaved independently as described in the next subsection. After time interleaving, all subchannels are multiplexed together into the MSC (see Figure 4.82 for an example). There is an elementary 24 ms time period in the MSC that is called a common interleaved frame (CIF). For TM II and TM III, each transmission frame carries one CIF. For TM I and TM IV, each transmission frame carries four or two subsequent CIFs, respectively. The multiplex configuration of the DAB system is extremely flexible. For each subchan- nel, the appropriate source data rate and the error protection can be individually chosen. The total capacity of 864 will be shared by all these subchannels. Table 4.5 shows an example (taken from reality) of how the capacity may be shared by different subchannels (which are loosely called programmes in that table). Time interleaving For DAB, time and frequency interleaving has been implemented. To spread the coded bits over a wider time span, time interleaving is applied for each subchannel. It is based on the 248 OFDM Audio encoder 1 Channel encoder 1 interleaver Time Audio Channel encoder 2 encoder 2 Time interleaver Subch 1 Subch 2 Time interleaver Subch n Channel encoder nencoder Data Multiplexer MSC Figure 4.82 Example for an error protection profile for the audio data rate 192 kbit/s. Table 4.5 Example for multiplex configuration Programme Content Bit rate Capacity Protection Audio 1 Pop music 160 kbit/s 116 CU PL3 Audio 2 Classical music 192 kbit/s 140 CU PL3 Audio 3 Classical music 224 kbit/s 168 CU PL3 Audio 4 Traffic info 80 kbit/s 58 CU PL3 Data 1 Visual service 72 kbit/s 54 CU PL3 Audio 5 Information 192 kbit/s 116 CU PL4 Audio 6 Information 128 kbit/s 96 CU PL3 Audio 7 Pop music 160 kbit/s 116 CU PL3 Sum 864 CU convolutional interleaver as explained in Subsection 4.4.2. With the notation introduced in that subsection, B = 16 has been chosen and N is the number of coded bits of one logical frame. First, the code word (i.e. the bits of one logical frame) will be split up into small groups of 16 bits. The bits with number 0 to 15 of each group will be permuted according to the bit reverse law (i.e. 0 → 0, 1 → 8, 2 → 4, 3 → 12, , 14→ 7, 15 → 15). Then, in each 16 bit group, bit number 0 will be transmitted without delay, bit number 1 will be transmitted with a delay of N serial bit periods T S , that is, by the duration of one logical frame of T L = NT S =24 ms. Bit number 2 will be transmitted with a delay of 2T L = 2 · 24 ms, and so on, until bit number 15 will be transmitted with a delay of 15T L = 15 · 24 ms. At the receiver side, the deinterleaver works as follows. In each group, bit number 0 will be delayed by 15T L = 15 · 24 ms, bit number 1 will be delayed by 14T L = 14 · 24 ms, , bit number 14 will be delayed by T L = 24 ms and bit number 15 will not be delayed. Afterwards, the bit reverse permutation will be inverted. The deinterleaver restores the bit stream in the proper order, but the whole interleaving and deinterleaving procedure results OFDM 249 in an overall decoding delay of 15T L = 15 · 24 ms = 360 ms. This is a price that has to be paid for a better distribution of errors. A burst error on the physical channel will be broken up by the deinterleaver, because a long burst of adjacent (unreliable) bits before the deinterleaver will be broken up so that two bits of a burst have a distance of at least 16 after the deinterleaver and before the decoder. The time interleaving is defined individually for each subchannel. This has been done because the receiver usually will decode only one subchannel and should therefore not process any data that belong to other subchannels. At the transmitter, it is more convenient to process all the subchannels together. The DAB system has been designed in such a way that both are possible. It is an important fact that the size of the capacity unit of 64 bits is an integer multiple of the period of B = 16 bits. As a consequence, each subchannel has a logical frame size N that is an integer multiple of B = 16 bits. Thus, we may interchange the order of time interleaving and multiplexing in Figure 4.82 and get the same bit stream for the MSC. The time interleaving will only be applied to the data of the MSC. The FIC has to be decoded without delay and will therefore only be frequency interleaved. Frequency interleaving and modulation Because the fading amplitudes of adjacent OFDM subcarriers are highly correlated, the modulated complex symbols will be frequency interleaved. This will be done with the QPSK symbols before the differential modulation. We explain it by an example for TM II with K = 384 subcarriers: A block of 2K = 768 encoded and time-interleaved bits have to be mapped onto the 384 complex modulation symbols for one OFDM symbol of duration T S . The first 384 bits will be mapped to the real parts of the 384 QPSK symbols, the last 384 bits will be mapped to the imaginary parts. To write it down formally, the bits p i,l (i = 0, 1, ,2K − 1) of the block corresponding to the OFDM symbol with time index l will be mapped onto the QPSK symbols q i,l (i = 0, 1, ,K − 1) according to the rule q i,l = 1 √ 2  1 − 2p i,l  + j  1 − 2p i+K,l  ,i= 0, 1, ,K −1. The frequency interleaver is simply a renumbering of the QPSK symbols according to a fixed pseudorandom permutation. The QPSK symbols after renumbering are denoted by x k,l (k =±1, ±2, ±3, ,±K/2). Then the frequency-interleaved QPSK symbols will be differentially modulated according to the law s k,l = s k,l−1 · x k,l . The complex numbers s k,l are the Fourier coefficients of the OFDM with time index l in the frame. Performance considerations Sufficient interleaving is indispensable for a coded system in a mobile radio channel. Error bursts during deep fades will cause the Viterbi decoder to fail. As already discussed in detail, OFDM is very well suited for coded transmission over fading channels because it allows time and frequency interleaving. Both interleaving mechanisms work together. 250 OFDM An efficient interleaving requires some incoherency of the channel to achieve uncorrelated or weakly correlated errors at the input of the Viterbi decoder. This is in contrast to the requirement of the demodulation. A fast channel makes the time interleaving more efficient, but causes degradations because of fast phase fluctuations. As discussed in the example at the end of Subsection 4.4.1, the benefit of time interleaving is very small for Doppler frequencies below 40 Hz. On the other hand, this is already the upper limit for the DQPSK demodulation for TM I. For even lower Doppler frequencies corresponding to moderate or low car speeds and VHF transmission, the time interleaving does not help very much. In this case, the performance can be saved by an efficient frequency interleaving. Long echoes ensure efficient frequency interleaving. As a consequence, SFNs (single frequency networks) support the frequency interleaving mechanism. If, on the other hand, the channel is slowly and frequency-flat fading, severe degradations may occur even for a seemingly sufficient reception power level. To compare with the theoretical DQPSK bit error rates discussed in Subsection 4.4.1, we performed several simulations of the DAB system. For the delay power spectrum DAB HT2 that was defined during the evaluation process is based on real channel measurements. It is the superposition of three exponential delay power spectra delayed by τ 1 = 0 µs, τ 2 = 20 µs, τ 3 = 40 µs with normalized powers P 1 = 0.2,P 2 = 0.6,P 3 = 0.2 and respective delay spreads τ m1 = 1 µs, τ m2 = 5 µs, τ m3 = 2 µs. The overall delay spread is τ m ≈ 14 µs. Figure 4.83 shows BER simulations for the DAB transmission mode II system with a 256 kbit/s data stream with EEP compared with the DQPSK union bounds. The maximum Doppler frequency for the isotropic spectrum is 64 Hz, which leads to ν max T S = 0.02. Time 0 5 10 15 20 10 − 4 10 −3 10 −2 10 −1 10 0 ←R c = 8/10 64 Hz ←R c = 8/12 64 Hz ←R c = 8/16 64 Hz ←R c = 8/32 64 Hz SNR [dB] BER Figure 4.83 Simulated BER for the DAB system for ν max T S = 0.02 and R c = 8/10, 8/12, 8/16, 8/32 and a frequency-selective channel. OFDM 251 0 5 10 15 20 10 −4 10 −3 10 −2 10 −1 10 0 ←R c = 8/10 10 Hz ←R c = 8/12 10 Hz ←R c = 8/16 10 Hz ←R c = 8/32 10 Hz SNR [dB] BER Figure 4.84 Simulated BER for the DAB system for ν max T S = 0.003 and R c = 8/10, 8/12, 8/16, 8/32 and a frequency-flat channel. interleaving alone cannot be sufficient because closely related bits are only separated by 24 ms. To separate them, the Doppler frequency would have to exceed, significantly, 40 Hz, which would lead to unacceptable high values of ν max T S . The simulated curves fit quite well with the theoretical curves, which indicates that both interleaving mechanisms together lead to a sufficient separation of the bits on the physical channel. The weakest protection profile shows some degradations. This can be understood by the fact that the DAB EEP profiles have exactly those fractional code rates including the coded tail bits. The tail bits are coded by R c = 1/2. The corresponding 12 coded bits are saved by using the next weakest code for the last 96 bits in the data stream, which leads to a poorer performance there. It can be verified by computer simulations that this effect becomes smaller for higher data rates and more severe for lower data rates. Figure 4.84 shows BER simulations for the DAB transmission mode II system with a 256 kbit/s data stream with EEP compared with the DQPSK union bounds. The maximum Doppler frequency for the isotropic spectrum is 10 Hz, which leads to ν max T S = 0.003. For a radio frequency of 230 MHz, this corresponds to a vehicle speed of 48 km/h. The delay power spectrum is the GSM typical urban spectrum, which is of exponential type with τ m = 1 µs. Neither time interleaving nor frequency interleaving is sufficient for this channel. Significant degradations compared to the other channel can be observed. 4.6.2 The DVB-T system The European Digital Video Broadcasting (DVB) system splits up into three different transmission systems 14 corresponding to three different physical channels: a cable system 14 Further extensions are currently being defined. We do not discuss them here. 252 OFDM (DVB-C), a satellite system (DVB-S) and a terrestrial system (DVB-T). Because the re- quirements of the three channels are very different, different coding and modulation schemes have been implemented. Common to all three systems is an (outer) Reed–Solomon (RS) code to achieve the extremely low bit error rates that are required for the video data stream and that cannot be reached efficiently by convolutional coding alone. For the DVB-C stan- dard, an AWGN channel with very high SNR can be assumed so that the Reed–Solomon code alone is sufficient. Both DVB-S and DVB-T need an inner convolutional code. This is necessary for the first one because of the severe power limitation of the satellite channel. For the second one, the terrestrial channel is typically a fading channel for which convolu- tional codes are usually the best choice because they can take benefit from the channel state information. All three systems use QAM modulation. For DVB-S, only 4-QAM (= QPSK) is used for reasons of power efficiency. Both other systems have higher-level QAM as possible options. DVB-C and DVB-S use conventional single carrier modulation. DVB-T uses OFDM to cope with long echoes and to allow SFN coverage. We concentrate on the discussion of the terrestrial system. The physical channel is similar to that of the DAB system. We may have runtime differences of the signal of several ten microseconds, which are due to echoes caused by the topographical situation. For both systems, SFNs are a requirement at least as one possible option. One significant difference in the requirements is that the DAB system has been especially designed for mobile reception. For the DVB-T system, portable – but not mobile – reception was required when the system parameters were chosen. DVB-T is intended to replace existing analog television signals in the same channels. Depending on the country and the frequency band (VHF or UHF band), there exist TV chan- nels of 6 MHz, 7 MHz and 8 MHz nominal bandwidth. The DVB-T system can match the signal bandwidth to these three cases. Similar to the DAB system, transmission modes have been specified to deal with different scenarios. For each of the three different bandwidth options, there exist two such parameter sets. They are called 8k mode and 2k mode, corre- sponding to the smallest possible (power of two) FFT length 8192 and 2048, respectively. The OFDM symbol length of the 8k mode is similar to that of the DAB transmission mode I and thus intended for SFN coverage. Because of the long symbol duration, it is more sensi- tive against high Doppler frequencies. The OFDM symbol length of the 2k mode is similar to that of the DAB transmission mode II. It is suited for typical terrestrial broadcasting sit- uations, but not for SFNs. It may thus preferably be used for local coverage. Let us denote again the OFDM Fourier analysis window by T , the total symbol length by T S and the guard interval by . In contrast to the DAB system, there exist several options for the length of the guard interval:  = T/4,  = T/8,  = T/16 and  = T/32. Table 4.6 shows the OFDM symbol parameters for the 8k mode and Table 4.7 for the 2k mode, both with the Table 4.6 OFDM Parameters for the DVB-T 8k mode and  = T/4 Channel t s TT S  Max. frequency 8192 t s 10, 240 t s 2024 t s 8MHz 7/64 µs 896 µs 1120 µs 224 µs ≈800 MHz 7MHz 1/8 µs 1024 µs 1280 µs 256 µs ≈700 MHz 6MHz 7/48 µs ≈1195 µs ≈1493 µs ≈299 µs ≈600 MHz OFDM 253 Table 4.7 OFDM Parameters for the DVB-T 2k mode and  = T/4 Channel t s TT S  Max. frequency 2048 t s 2560 t s 512 t s 8MHz 7/64 µs 224 µs 280 µs56µs ≈3200 MHz 7MHz 1/8 µs 256 µs 320 µs64µs ≈2800 MHz 6MHz 7/48 µs ≈299 µs ≈373 µs ≈75 µs ≈2400 MHz guard interval length  = T/4. All time periods are defined as a multiple of the sampling period t s = f −1 s that is different for the three different TV channel bandwidths. For each mode, the three bandwidths can be obtained by a simple scaling of that sampling frequency. The number of carriers is given by K + 1 = 6817 for the 8k mode and by K + 1 = 1705 for the 2k mode. The spacing f K/2 − f −K/2 between the highest and the lowest subcarrier is approximately given by 7607 kHz for the 8 MHz channel, 6656 kHz for the 7 MHz channel and by 5705 kHz for the 6 MHz channel. The frequency in the last column is the optimistic upper limit for the maximum frequency that can be used for a vehicle speed of 120 km/h if a very powerful channel es- timation with Wiener filtering has been implemented and if an appropriately strong channel coding and modulation scheme has been chosen. The pilot grid for DVB-T is the diagonal one of Figure 4.36. The parameters of the 7 MHz system correspond approximately to the numerical example given in Subsection 4.3.2. For the 8k mode, according to that example, the channel will be sampled with a sampling frequency of approximately 200 Hz. Owing to the sampling theorem, the limit for the Doppler frequency is then given by 100 Hz. This corresponds to 900 MHz radio frequency for a vehicle speed of 120 km/h. In practice, one should be well below the limit given by the sampling theorem. For a good channel estima- tion, 700 MHz should be possible. This value corresponds to approximately 78 Hz Doppler frequency or ν max T S = 0.1. As we have seen in Subsection 4.5.3, this value can be tolerated, for example, for 16-QAM and code rate R c = 1/2, but not for higher spectral efficiencies. For 64-QAM and code rate R c = 1/2, the maximum frequency should be 25% lower. Because the DAB transmission modes I and II have similar symbol length as the 8k and 2k modes of DVB-T, a direct comparison of the sensitivity against high Doppler frequencies are possible. We conclude that the DVB-T system allows approximately twice the carrier frequency (or vehicle speed) compared to the DAB system. From the discussion in Subsection 4.5.3, we further conclude that at the highest possible value for the DAB system, the DVB-T system with 16-QAM has a similar performance as the DAB system at approximately twice the spectral efficiency. In both cases, R c = 1/2 has been assumed. Baseline transmission system The baseline DVB-T transmission system is depicted in Figure 4.85. Packets of 188 bytes length will first be encoded to code words of length 204 by the outer RS(204, 188, 17) code. This code has Hamming distance 17 and can thus correct up to eight byte errors. This shortened RS code has been obtained from a RS(255, 239, 17) code by setting the first 51 systematic bytes to zero and not transmitting them. The code words are interleaved by a convolutional byte interleaver as described in Subsection 4.4.2 with parameters B = 12 and M = 17. Thus, N = BM = 204 is just the block length of one code word. The interleaver 254 OFDM RS encoder Convol. encoder Bit interl. Byte interl. mapper QAM Symbol interl. OFDM Figure 4.85 Simplified block diagram for the DVB-T signal generation. works in such a way that the byte number zero (i.e. the first one) in a block stays at the same position and in the same block. The byte number one is delayed by the block length N, that is, it will be transmitted in the next block at the same position within the block. The byte number two is delayed by 2N, that is, two blocks, and so forth until byte number 12, which stays inside the block at the same position and the whole procedure, will continue that way. This outer byte interleaver is necessary because at the receiver the inner decoder produces error bursts. These error bursts must be distributed over several code words because more than 8 bytes in one code word cannot be corrected. Following the discussion in Subsection 4.4.2 we observe that an error burst of 12 bytes (= 96 bits) length after the inner decoder will result in only one corresponding byte error inside one code word. The RS code can correct up to eight byte errors, that is, the error bursts may be eight times longer. The bit stream of the byte interleaved code words will be encoded by an inner encoder for the standard (133, 171) oct convolutional code and then modulated as discussed in Subsections 4.5.1 and 4.5.2. With optional puncturing, the code rates R c = 1/2, R c = 2/3, R c = 3/4, R c = 5/6andR c = 7/8 are possible. The output bit stream of the convolutional encoder will be interleaved by a (small) pseudorandom permutation and mapped on complex QAM symbols by a symbol mapper. Thus, exactly the concept of bit-interleaved coded modulation has been implemented in the DVB-T system. The options 4-QAM, 16-QAM and 64-QAM are possible. The QAM symbols are OFDM modulated. Each OFDM symbol carries 6048 QAM symbols in the 8k mode and 1512 QAM symbols in the 2k mode, respectively. The other complex symbols serve as pilot symbols for channel estimation. In addition to the diagonal grid of scattered pilot of Figure 4.36, there are continuous pilots that serve as references for frequency synchronization. All pilots are boosted by a factor of 4/3 in the amplitude compared to the QAM symbols. Table 4.8 shows the possible coding and modulation options and the corresponding data rates for  = T/4 and the 8 MHz system. To exploit the channel diversity in frequency direction, for each OFDM symbol, the QAM symbols are frequency interleaved by a pseudorandom permutation of length 6048 or 1512, respectively. In contrast to the DAB system, no time interleaving is applied. This is due to the fact that originally no mobile reception was intended. A set of 68 OFDM symbols are grouped together to a transmission frame, and four such frames build a hyperframe. There are some significant differences to the DAB system. First, there is no correspondence between certain parts of the data stream and certain OFDM symbols in the frame. DAB allows different code rates for different parts of the signal. This is not possible for DVB-T. The information that is necessary to identify the overall code rate and the guard interval length are transmitted on special TPS (transmission parameter signaling) carriers. Channel coding aspects The DVB-T channel coding scheme consists of an inner convolutional code and an outer Reed–Solomon code. The outer symbol interleaver is a frequency interleaver that has the OFDM 255 Table 4.8 Transmission options and data rates for DVB-T for guard in- terval length  = T/4 Modulation Code rate Bits per symbol R b Useful R b QPSK R c = 1/2 1 5.4 Mbit/s 4.98 Mbit/s QPSK R c = 2/3 1.33 7.2 Mbit/s 6.64 Mbit/s QPSK R c = 3/4 1.5 8.1 Mbit/s 7.46 Mbit/s QPSK R c = 5/6 1.67 9.0 Mbit/s 8.29 Mbit/s QPSK R c = 7/8 1.75 9.45 Mbit/s 8.71 Mbit/s 16-QAM R c = 1/2 2 10.8 Mbit/s 9.95 Mbit/s 16-QAM R c = 2/3 2.67 14.4 Mbit/s 13.27 Mbit/s 16-QAM R c = 3/4 3 16.2 Mbit/s 14.93 Mbit/s 16-QAM R c = 5/6 3.33 18.0 Mbit/s 16.59 Mbit/s 16-QAM R c = 7/8 3.5 18.9 Mbit/s 17.42 Mbit/s 64-QAM R c = 1/2 3 16.2 Mbit/s 14.93 Mbit/s 64-QAM R c = 2/3 4 21.6 Mbit/s 19.91 Mbit/s 64-QAM R c = 3/4 4.5 24.3 Mbit/s 22.39 Mbit/s 64-QAM R c = 5/6 5 27.0 Mbit/s 24.88 Mbit/s 64-QAM R c = 7/8 5.25 28.4 Mbit/s 26.13 Mbit/s purpose to break up the correlations of the channel and provide the inner code with the diversity that can be obtained from the frequency selectivity of the channel. No similar mechanism is intended to take advantage from time variance of the channel. The bit inter- leaved coded modulation needs a (small) bit interleaver between the convolutional encoder and the symbol mapper. This is necessary in order to avoid closely related bits of the code word being affected by the same noise sample. Of course, it would have been possible to use a bigger bit interleaver for both purposes together. The outer byte interleaver has the purpose to break up long error bursts resulting from erroneous convolutional decoding. The combination of a convolutional inner code together with an outer RS code with an interleaver in between is a very powerful combination. The RS code is very efficient for burst error decoding as long as the bursts are not too long. It takes advantage from the fact that more than one bit error is inside one erroneous byte. Let P be the byte error probability and P b the bit error probability after the Viterbi decoder. We note that the worst case of only one average bit error in one erroneous byte corresponds to P = 8P b , two bit errors correspond to P = 4P b and four bit errors correspond to P = 2P b . The assumption of ideal interleaving means that the byte errors are uniformly distributed. The block error probability analysis of Subsection 3.1.2 can be generalized to the case that we deal with bits rather than with bytes. The probability for the block code word error probability is then given by P Block = N  i=t+1  N i  P i (1 − P) N−i . In these equations, N = 204 is the length of the code word, and t = 8 is the error correction capability. To obtain the residual bit error probability, we can argue as we did in Subsec- tion 3.1.2. We take into account that, for a given bit inside a byte, 128 of 255 possible byte [...]... Nsym of OFDM symbol form the so-called payload There are five different bursts with different preamble length: 1 The Broadcast burst: Preamble of length 16 µs The payload consists of Nsym = 496 OFDM symbols 2 The Downlink burst: Preamble of length 8 µs The payload consists of Nsym = 498 OFDM symbols 3 Uplink burst with short preamble: Preamble of length 12 µs The payload consists of Nsym = 4 97 OFDM. .. long preamble: Preamble of length 16 µs The payload consists of Nsym = 496 OFDM symbols 5 Direct link burst: Preamble of length 16 µs The payload consists of Nsym = 496 OFDM symbols The last 8 µs of the preamble is common to all bursts and serves as a reference for the channel estimation that is necessary for the coherent demodulation It consists of an OFDM reference symbol of length 2TS = 8 µs, which... regarded as the OFDM pioneer system One of the authors (Henrik Schulze) became involved in the DAB project in 19 87 (at Bosch Company in Hildesheim) and came in touch with OFDM through an internal project paper that was a draft version of (Alard and Lassalle 19 87) At that time, very few people understood that concept and thus OFDM was regarded as a wonder cure against everything by its supporters, and it was... achieve this goal, OFDM has been introduced as the basis for the transmission techniques Intensive discussion between these two groups led to widely harmonized parameters for OFDM, modulation and channel coding The corresponding IEEE 802.11a standard (IEEE 802.11a 1999) and HIPERLAN/2 standard (EN101 475 2001) were released in 1999 and 2000, respectively The channel coding schemes of all the OFDM systems... property for military communications because it helps to hide the signal and it makes the signal more robust against intended interference (jamming) Spreading is achieved – loosely speaking – by a multiplication of the data symbols by a spreading sequence of pseudorandom signs These sequences are called pseudonoise (PN) sequences or code signals We Theory and Applications of OFDM and CDMA  2005 John Wiley... exponentially with d and a decay constant of some tens of meters (the typical scale of obstacles) Figure 5.6 shows a typical diagram of the received level as a function of the distance r taking into account all three parts of the propagation model It can be seen that the received level varies in a region of about 70 dB Quality of service, coverage and handover gain The objective of network planning... perturbations of the audio quality 4.6.3 WLAN systems OFDM with a guard interval is applied within two systems for wireless communications between computers in a local area network The corresponding standards for these Wireless Local Area Networks (WLAN) are called: • the HIPERLAN/2 standard released by the European Telecommunications Standards Institute (ETSI) in 2000; • the IEEE 802.11a and IEEE 802.11g standard... one transmission mode of UMTS, which is also often called Wideband CDMA Its transmission bandwidth of about 5 MHz may be viewed as wide in some urban environments, but not in any indoor environment  270 CDMA UL DL MS MS MS MS BS BS MS MS MS BS BSC / RNC BSC / RNC MSC Figure 5.4 Architecture of a cellular mobile radio network Handover When an MS moves from one cell to another, a handover occurs One distinguishes... very serious candidate and, at the end of the project, an OFDM system was standardized in 1993 (see (EN300401 2001a) for a recent update of the standard) An exhaustive treatment of the DAB system that is also very helpful for the practical engineer can be found in (Hoeg and Lauterbach 2003) A comprehensive overview about multicarrier modulation and its history can be found in (Bingham 1990) and in (Gitlin... the case of a soft handover) For producing the solid line probability function of Figure 5 .7, an ideal power budget handover has been used selecting the BS with the best local mean signal level Looking at cp = 0.95, that is, op = 0.05, a gain of about 7 dB is obtained by the handover method This gain is called handover gain or macrodiversity gain When using a kind of maximum ratio combining of the macrodiversity . t s 8MHz 7/ 64 µs 896 µs 1120 µs 224 µs ≈800 MHz 7MHz 1/8 µs 1024 µs 1280 µs 256 µs 70 0 MHz 6MHz 7/ 48 µs ≈1195 µs ≈1493 µs ≈299 µs ≈600 MHz OFDM 253 Table 4 .7 OFDM Parameters for the DVB-T 2k mode and. multiplication of the data symbols by a spreading sequence of pseudoran- dom signs. These sequences are called pseudonoise (PN) sequences or code signals. We Theory and Applications of OFDM and CDMA Henrik. consists of N sym = 4 97 OFDM symbols. 4. Uplink burst with long preamble: Preamble of length 16 µs. The payload consists of N sym = 496 OFDM symbols. 5. Direct link burst: Preamble of length