Latest Trends in Worldwide Digital Terrestrial Broadcasting and Application to the Next Generation Broadcast Television Physical Layer

This paper summarizes key technologies and identifies trends in recent terrestrial broadcast standards. For robust service in urban canyons and mobile environments, Orthogonal Frequency Division Multiplexing (OFDM) has proven extremely advantageous and is employed in ISDB-T, DVB-T2, and DTMB standards. Strong Forward Error Correction (FEC) codes are also needed to increase signal robustness. Low Density Parity Check (LDPC) codes are strong codes that offer coding which enables reception approaching Shannon limit Eb/No levels. They have been adopted in DVB-T2 and DTMB. Another trend is the expanding use of terrestrial broadcast for mobile reception, such as the ATSC-M/H standard. To increase flexibility one innovative technology known as Multiple Physical Layer Pipes (M-PLP) was introduced in DVB-T2, where the physical layer is divided into separate logical signals or “pipes”. Based on these recent trends, key technologies suitable for Next Generation Broadcast Television (NGBT) are explored. Cutting edge technologies examined include the expansion of modulation constellations to 1024QAM, adoption of recent LDPC FEC coding schemes and the use of Multiple Input Multiple Output (MIMO) technology. We conclude that by using a combination of proven and cutting edge technology, ATSC can develop a world leading standard. Introduction Terrestrial broadcasting has evolved from analog to digital and there has been a flurry of second generation digital broadcasting standardization activity worldwide. Consumers have a desire for even higher definition TV, more content and to be able to reliably receive television programs wherever they are. In the near future, new and innovative services such as higher definition television (8K/4K), 3D video and multi subchannel audio are envisaged. To achieve these desirable new features the physical layer specification must provide increased spectral efficiency, robustness as well as a high level of flexibility for the broadcaster. Each new broadcast standard has made progress towards this goal by applying the latest technological advances to the physical layer. Recent Trends in Digital Terrestrial Television (DTT) With larger screen sizes available in the market, higher picture quality is desired. The current industry standard for high quality television is High-Definition (HD) television, which offers up-to 1920x1080 resolution at 60 interlaced frames per second. In the next ten years even higher picture quality is envisaged for broadcast television, for example 4K2K/60p and even 8K4K/60p. In order to provide for this high quality picture, the current HD data rate will increase 8 times for 4K2K/60p, and increase by a factor of 32 compared to current HD television for 8K4K/60p, if using equivalent video compression codecs. Another very recent trend is 3D television. On the physical layer approximately twice the bit rate is desirable to transmit 3D television at the same level of quality. In addition, since spectrum is a scarce resource, the most spectrally efficient technologies should be employed. Recent years have also seen a proliferation of mobile display and communication devices and consumers desire to receive television on these devices. Successful deployments in Japan using the ISDB-T (One-Seg) standard [3] and in Korea using the T-DMB standard [8] show that under certain conditions, mobile television can be a hit. The European mobile TV standard DVB-H [6] has been widely deployed, but with mixed results. The US has recently standardized mobile-DTV with ATSC A/153 (ATSC-M/H) [7]. To successfully broadcast mobile TV extremely robust modulation and coding methods should be examined. Technology Trends for Robustness to Multi-path Fading The terrestrial propagation channel is subject to a number of perturbations such as thermal and manmade noise, time variation and frequency selective fading. All these effects influence the selection and setting of broadcast parameters. Furthermore as it is impossible to guarantee a line-of-sight reception in a terrestrial channel, multi-path propagation is often unavoidable. Multi-path propagation causes inter-symbol interference (ISI) and frequency-selective fading. There exist two different options to minimize the effects of multi-path fading. First is to limit the broadcasters’ choice of transmission scheme and parameters, such as by choosing OFDM and an extremely large guard interval. The second option is to design receiver algorithms to estimate the channel impulse response and reduce its effects using efficient channel equalization. Table 1 illustrates the key parameters of recent DTT standards. They can broadly be categorized into two types; (i) single carrier (SC) and (ii) multi carrier (MC) or OFDM. In OFDM the available bandwidth is divided into a number of narrowband sub-carriers. The partial allocation of the data payload to each sub-carrier protects it against frequency-selective fading. This is due to the fact that the channel over each sub-carrier can now be considered as flat-fading resulting in a lowcomplexity equalization using a simple one-tap equalizer. In SC transmission systems such as ATSC and DTMB (C1) [2] channel equalization is relatively complex, usually deploying multi-tap decision-feedback adaptations. Two different OFDM schemes have been adopted in recent digital TV standards. In cyclic prefix OFDM (CP-OFDM) as used in DVB-T2 [1] and ISDB-T, the cyclic prefix is some samples of the OFDM symbol which are copied to the front of the symbol. In DTMB (C3780) [2], unique PN sequences are inserted at the front of each OFDM symbol. The creation of CP-OFDM and PN-OFDM symbols is shown in Fig.1. In CP-OFDM-based systems such as DVB-T2 and ISDB-T it is possible to cancel ISI completely by insertion of a cyclic prefix guard interval, provided that the channel delay spread is smaller or equal to the guard interval duration. Pseudorandom number sequences used in DTMB (3780) have good autocorrelation properties which assist in channel estimation, however ISI cannot be completely cancelled due to the interference caused by the delayed path’s PN sequence into the first OFDM symbol. Table 1. Key Features of Recent Worldwide DTT Transmission Standards Fig. 1. Guard Interval Insertion, (a): Cyclic Prefix (CP-OFDM), (b): PN Sequence (PN-OFDM) Key Features of Recent World-Wide DTT Transmission Standards Standard DVB-T2 [1] DTMB [2] (C1) (C3780) ISDB-T [3] ATSC A/53 [4] Country of Origin Europe China Japan US Year 2008 2006 2003 2005 Frequency Range (MHz) 470-862 and 174-230 470-862 470 – 770 54-72,76-88,174216,470-698 Channel Spacing,B (MHz) 1.7, 5, 6, 7, 8, 10 8 6, 7, 8 6 Type Multi-Carrier CP-OFDM Single Carrier Multi-Carrier PN-OFDM Multi-Carrier CP-OFDM Single Carrier Number of SubCarriers (K = 1024) 1K ,2K, 4K, 8K, 16K, 32K 1 3780 2K, 4K, 8K 1 Sub-Carrier Modulation QPSK, 16QAM, 64QAM, 256QAM 4QAM-NR, 4QAM, 16QAM, 32QAM, 64QAM QPSK, 16QAM, 64QAM, DQPSK 8VSB Useful Symbol Length, Tu (us) 112 to 3,584 (8 MHz Channel) 500 252 to 1008 0.092917 Guard Interval, Tg (fraction of Tu) 1/128, 1/32,/1/16, 19/256, 1/8, 19/128, 1/4 1/4, 595/3780, 1/9 1/4, 1/8, 1/16, 1/32 Sub-Carrier Spacing (Hz) 279 to 8,929 (8 MHz Channel) 2k (8MHz) 992 to 3968 (6 MHz Channel) Net Bit Rate, R (Mbit/s) 40.2 (Typical), 50.3 (Max.) 4.81 32.49 3.651 23.23 19.392658 Link Spectral Efficiency, R/B (bit/s/Hz) 0.87 6.65 0.60 4.1 0.61 3.87 3.23 Inner FEC LDPC {1/2, 3/5, 2/3, 3/4, 4/5, 5/6} LDPC {0.4, 0.6, 0.8} Convolutional Code {1/2, 2/3, 3/4, 5/6, 7/8} Convolutional Code (2/3) Outer FEC BCH BCH (762, 752) Reed Solomon (204, 188) Reed Solomon (207,187) Time Interleaving Depth (ms) 80 ms (U.K. mode) 200 500 0 to 400 4 Main Unique Features Rotated Constellations, FEFs, Multi-Pipes, MISO TDS-OFDM (PN Sequence Insertion) Segmented OFDM Low C/N Requirement To mitigate the frequency selective effects of multi-path propagation the receiver needs to perform accurate estimation of the channel followed by efficient channel equalization. In DVB-T2 and ISDB-T, pilot sub-carriers, at regular grids which shift positions from one OFDM symbol to the next, are inserted in between the data carriers. The receiver makes use of the known pilot sub-carriers to first estimate the frequency response of the channel at these locations. Secondly, it extends the frequency response to full bandwidth by some form of interpolation such as frequency-only or time-frequency interpolation. In DTMB (C3780) the PN sequences are used to estimate the time-domain channel response which is then used for equalization of the data frame. Table 1 shows that the most recent DTT standard, DVB-T2, is the most spectrally efficient at 6.65 bit/s/Hz. This system is currently used in the UK for transmitting free-to-air HDTV with a net data rate of greater than 40 Mbps. This high spectral efficiency is achieved due to the high number of sub-carriers, large constellation size (256QAM) and the minimum number of pilot symbols within the standard. FEC Trends Error control coding in a broadcasting scenario typically comes in the form of forward error correction (FEC) coding. The ratio of information to code symbols is reflected by the code rate. To meet the demand for high spectrum efficiency, FEC techniques are becoming more important and much research has been undertaken. Low density parity-check (LDPC) codes are known to achieve high capacity, high throughput, and have been adopted in DVB-T2 and DTMB. LDPC are the codes of choice for all recent second generation broadcast formats. Density evolution is one powerful tool to analyze the asymptotic behavior of an LDPC ensemble (a set of LDPC codes of a certain degree profile and of infinite length) and it is usually used to guarantee the performance of the codes. Quasi-cyclic structure with unit circulant matrices in a parity-check matrix is very effective to reduce matrix ROM size and the randomness of memory accesses. This helps to maximize decoder parallelism so that higher throughputs can be realized. In order to simplify the encoding procedure, an accumulator structure is set in parity-check matrices, which enables calculation of parity bits with simple accumulators. DVB-T2 adopts LDPC codes of length N=64800 (Normal mode) and 16200 (Short mode) with quasicyclic and accumulator structure in their parity-check matrices. Unit circulant matrix size l is equal to 360. In order to increase data rates, constellations of high order up to 256QAM are employed. Fig. 2. One of DVB-T2 parity check matrices with quasi-cyclic and accumulator structure, in which each black dot means “1” and blank means “0” in the matrix (N=64800, rate=3/4, l=360). Two diagonal lines in the right hand side construct accumulators for encoding. DTMB adopts LDPC codes of length N=7493 with quasi-cyclic structure, but without accumulators in its parity-check matrices. Generator matrices with the same cyclic structure are defined for encoding procedures, whose unit matrix size l is equal to 127. Constellations for the standard are up to 64QAM. Fig. 3. One of DTMB parity check matrices with quasi-cyclic structure, in which each black dot means “1” and blank means “0” in the matrix (N=7493, rate=0.6, l=127). First 2921 columns correspond to parity bits. Figure 4 shows the comparison of spectrum efficiency based only on FEC differences between DVB-T2 and DVB-T [6], the 1st generation European digital video broadcasting standard. DVB-T employed convolutional codes as its inner codes and Reed-Solomon codes as its outer codes. DVB-T2 obtains more than approximately 1.5 times as much capacity as DVB-T by employing LDPC codes with spectral efficiency approaching the Shannon limit.