The compact disc (CD), introduced

Transcription

1 / TECHNICAL PAPER The Digital Versatile Disc (DVD): System Requirements and Channel Coding By Kees A. Schouhamer Immink The digital versatile disc (DVD) is a new optical recording medium with a storage capacity seven times higher than the conventional compact disc (CD). The major part of the capacity increase is achieved by the use of optics, shorter laser wavelength, and larger numerical aperture, which reduces the spot diameter by a factor of The track formed by the recorded pits and lands, as well as the track pitch, can be reduced by the same factor. The storage capacity is further increased by a complete redesign of the logical format of the disc, including a more poweiful error correction and recording code. The system requirements of the DVD and the related channel coding are outlined in this paper. The compact disc (CD), introduced more than a decade ago, is a very successful medium for the distribution and storage of digital audio and other digital information. It is anticipated that its storage capacity, 680 Mbytes, will be insufficient for future graphicsintensive computer applications and high-quality digital video programs. An extension of the compact disc family, the digital versatile disc (DVD), is a new optical recording medium with a storage 1 capacity seven times higher than the conventional CD. From a consumer vantage point, DVDs will be capable of storing highdefinition video/audio for featurelength movies and are expected to gradually replace videocassette tape and CD-ROM computer data disks once DVD becomes accepted as a commercial product. The DVD format will support a variety of uses. There will be at least two types of products introduced in 1996: the DVD movie player and the DVD-ROM. An audioonly version of the DVD is still under development. The single-layer disc can hold 4.7 Gbytes, which amounts to seven times more data than an audio CD. Typically, this is 1 ~oom for 135 min of widescreen (16:91aspect ratio) video Revised version of a paper presented at the!37th SMPTE Technical Conference in New Orleans, La. (paper no ) on September 7, An unedited version appeared in Moving Images: Meeting the Challenges, SMPTE, Kees A. Schouhamer Immink is with Philips Research Laboratories, Eindhoven, The Netherlands. Copyright 1996 by the Society of Motion Picture and Television Engineers, Inc. at MPEG-2 quality accompanied by multiple audio and subtitle channels. Applying multilayer technology, a storage capacity of up to 17 Gbytes was obtained. The major part of the capacity increase is achieved by the use of optics, shorter laser wavelength, and larger numerical aperture, which reduces the spot diameter by a factor of The track formed by the recorded pits and lands, as well as the track pitch, can be reduced by the same factor. The storage capacity of the DVD is also increased by a complete redesign of the logical format of the disc, including a more powerful Reed-Solomon product code (RS-PC) and recording code (EFMPlus). The next sections will discuss the various design issues and choices that had) to bemade. / Physical Aspects - The main physical parameters of the single-layer DVD are listed in Table 1. The parameters of the dual-layer disc are more conservative. Mechanical specifications such as outer diameter and center hole diameter of the DVD are equal to those of the CD, allowing full backward compatibility. The disc thickness was halved to 0.6 mm, which resulted in a higher storage capacity than possible with a disc of 1.2-mm thickness. Specifications of the obsolete 1.2 mm thick Multimedia CD disc can be found in Ref. 1. Mechanical instabilities, such as disc warp, of the 0.6 mm thick disc have been solved by back-to-hack bonding of two 0.6 mm discs. Clearly, overall thickness of the sandwich is the same 1.2 mm of the CD, and the sandwich has the same mechanical stability. By employing a red laser at 635-nm wavelength and a numerical aperture (NA) of 0.6, the read-out resolution is increased by a factor of The subsequent scaling of the track pitch and the pit length per bit increase the actual reference wavelength NA margins EFM- ClRC- other gains lcd) EFMPlus RS-PC no subcode no 3rd layer CRC Optics Design I Drive and Drive Precision mill Decodin~ B Electromcs Figure 1. Overview of factors determining the capacity increase. 483

2 0.6mm Figure 2. Dual-layer reading principle diagram. Figure 3. Double dual-layer reading principle diagram. physical data density by a factor of 3.5. The nomimil read-out reference speed goes up from 1.2 to 4 m/sec. An overview of the various factors that underlaid the capacity increase is shown in Fig. 1. The various sources that contributed to the capacity increase into optics, drive precision, and electronics have been partitioned. It can be seen that the drive and disc margins have become more stringent by a factor of It is anticipated that the manufacturing of players and discs can be done with greater precision than. was the case 15 years ago when the CD was conceived. A second method to increase the capacity by a factor of two is by using a dual-layer disc, where the layers are on one side of the disc. The principle of operation of the dual-layer disc is shown in Fig. 2. The dual-layer CD is similar to the standard CD. It adopts the same molded substrate for the first data layer. The second layer is made by overlaying a fully reflective aluminum layer on a partially reflective aluminum layer. The second layer is applied using the same principles as the 2P (photo polymerization) process. Focal plane servos are used to switch from one layer to the other. The two layers can 484 Info layer I: fully reflective / be accessed without flipping the disc. Continuous play of video signals is obtained by reading outwards on one layer, then inwards on the other. By bonding two dual discs, the structure depicted in Fig. 3 is obtained. The sandwich dual-layer disc has a capacity of 18 Gbytes. Disc Thickness Considerations The disc thickness played a major role during the discussions between the various electronics companies. In this section, we will answer the intriguing question of why a thin disc can hold more information than a thick disc. The number of data bits, Nb, that can be stored on the disc is given by (1) where A is the useful area of the disc surface, d is the diameter of the laser light spot on the disc, and 11 is the efficiency of the recording method. The (areal) information density is thus given by Nb /A = 11Aitf2 (2) The spot diameter d is one of the most relevant parameters of an optical recording system. The channel coding can give a higher value of 11 and thus of N b Some of the aspects of the recording channel that govern the specifications of the playe,r and disc will now be explained briefl~'.. The quality of the c. annel is determined by the player an ',1 the disc; these are mass-produced and the tolerances cannot be made unacceptably small. One example is given to illustrate the way in which such tolerances affect the design. Specifically, the choice of the spot diameter d and disc thickness D are discussed. Defining d as the halfvalue diameter for the light intensity we have d= 0.5 IJNA (3) where A is the wavelength of the laser light and NA is the numerical aperture of the objective. To achieve a high information density (Eq. 2), d must be as small as possible, and according to this equation there are two design parameters., The laser chosen for this syste~ is a state-of-the-art, solid-state laser, which is inexpensive and has a wavejength of A -:::: 635 nm. This means that in order to have a small spot diameter, we must try to make the numerical aperture as large as possible. With increasing NA, however, the manufacturing tolerances of the player and the disc rapidly become smaller. For example, the tolerance in the local skew of the disc (the disc tilt) relative to the objective-lens axis is proportional to D/NN. The tolerance for the disc thickness is proportional to D/NN, and the depth of focus, which governs the focusing tolerance, is proportional to DINA 2 In practice the most stringent tolerance condition arises from disc tilt. It is very difficult to manufacture players and discs where disc tilt angles are much smaller than 0.5. After considering all these factors in relation to one another, for the CD having a disc thickness of D = 1.2 mm, a value of NA = 0.45 was arrived at. As the disc skew tolerance is proportional to D, the second parameter entering the capacity arena is the disc thickness. For given disc tilt tolerance, halving the disc thickness allows an NA increase by a factor of "" 1.26, and a consequent decrease of the spot diameter. The disc capacity will therefore increase by a Jactor of = 1.6. The benefit of the potential increase in capacity by a factor of 1.6 as a rdult

3 of halving the disc thickness is only one side of the balance. On the other side of the balance we have a loss in compatibility with the conventional CD. Replication equipment should be modified, and also, objective lenses capable of reading both "thin" and "thick" discs could be more expensive than standard lenses optimized for one thickness. On September 15, 1995, after some debate, the world's electronics industries decided to adopt the sandwich 0.6-mm disc. Electronics Aspects The 1 main audio/video specifications of the DVD are listed in Table 2. In the CD system, a concatenation of two codes, namely EFM (Eight-to-Fourteen Modulation) and CIRC (Cross Interleaved Reed-Solomon Code) is used. CIRC is used for correction and detection of erroneously retrieved information, while EFM is used for transforming the digital audio bit stream into a sequence of binary symbols, called channel bits, which are suitable for storage on the disc. 2. Not only are the EFM and CIRC coding schemes useful for the CD for which they have been designed, but they have been extensively employed in a large variety of digital audio players and home-storage products such as CD-ROM, CD-~ and MiniDisc. As in CD-ROM, the Qser data is organized into sectors. Sixteen sectors make one ECC block. Under DVD format rules, sectors of 2048 user bytes are translated into 2418 bytes (2048 user + 52 sync + 16 header error code), which are in tum translated into 16 x 2418 channel bits. Thus, one user bit is translated into 4836/2048 = 2.36 channel bits. In a conventional CD, an audio bit is translated into 588/192 = 3.05 channel bits/ and we conclude that the "format efficiency" of the DVD is improved by 32%. The new format is even 47% more efficient with respect to CD ROM, which uses a "third" error correction layer. Even though consuming a significantly lower data overhead, the error correction code, RS-PC, can cope with longer bursts and burst sequences, and more random errors. A comparison of the format efficiency of the DVD with conventional CD audio is provided in Table 3. E"or Co"ection Code: RS-PC The errors found in both the CD and DVD systems are a combination of a random and bursty character, and in order to alleviate the load on the manufacturing process some form of error correction is required. As mentioned earlier, the CD system uses the CIRC error correction code (ECC). The interleaving scheme is tailored to the specific requirements of the compact disc audio system. In particular, the adopted "cross" -interleaving technique will make it possible to effectively mask errors if correction is found to be impossible. Depending on the magnitude of the error to be concealed, this is done by interpolating or by muting the audio signal. The concealment will make errors almost inaudible, and as a result, it offers a graceful degradation of the sound quality. Specifically, sharp temporary degradation of the audio signal, or "clicks," are avoided. The judicious positioning of the left and right stereo channel, as well as the audio samples on even- or odd-number instants within the interleaving scheme, are key parameters to the success of the concealment strategy. For the DVD a much more powerful error correction has been developed. Enhanced error correction capability is required for several reasons. First, the increased physical density implies that 485

4 of halving the disc thickness is only one side of the balance. On the other side of the balance we have a loss in compatibility with the conventional CD. Replication equipment should be modified, and also, objective lenses capable of reading both "thin" and "thick" discs could be more expensive than standard lenses optimized for one thickness. On September 15, 1995, after some debate, the world's electronics industries decided to adopt the sandwich 0.6-mm disc. Electronics Aspects \ The main audio/video specifications of the DVD are listed in Table 2. In the CD system, a concatenation of two codes, namely EFM (Eight-to-Fourteen Modulation) and CIRC (Cross Interleaved Reed-Solomon Code) is used. CIRC is used for correction and detection of erroneously retrieved information, while EFM is used for transforming the digital audio bit stream into a sequence of binary symbols, called channel bits, which are suitable for storage on the disc. 2 Not only are the EFM and CIRC coding schemes useful for the CD for which they have been designed, but they have been extensively employed in a large variety of digital audio players and home-~torage products such as CD-ROM, CD4J, and MiniDisc. As in CD-ROM, the luser data is organized into sectors. Sixteen sectors make one ECCblock. Under DVD format rules, sectors of 2048 user bytes are translated into 2418 bytes (2048 user + 52 sync + 16 header+ 302 error code), which are in tum translated into 16 x 2418 channel bits. Thus, one user bit is translated into 4836/2048 = 2.36 channel bits. In a conventional CD, an audio bit is translated into = 3.05 channel bits, 3 and we conclude that the "format efficiency" of the DVD is improved by 32%. The new format is even 47% more efficient with respect to CD ROM, which uses a "third" error correction layer. Even though consuming a significantly lower data overhead, the error correction code, RS-PC, can cope with longer bursts and burst sequences, and more random errors. A comparison of the format efficiency of the DVD with conventional CD audio is provided in Table 3. E"or Co"ection Code: RS-PC The errors found in both the CD and DVD systems are a combination of a random and bursty character, and in order to alleviate the load on the manufacturing process some form of error correctionisrequrred., As mentioned earlier, the CD system uses the CIRC error correction code (ECC). The interleaving scheme is tailored to the specific requirements of the compact disc audio system. In particular, the adopted "cross" -interleaving technique will make it possible to effectively mask errors if correction is found to be impossible. Depending on the magnitude of the error to be concealed, this is done by interpolating or by muting the audio signal. The concealment will make errors almost inaudible, and as a result, it offers a graceful degradation of the sound quality. Specifically, sharp temporary degradation of the audio signal, or "clicks," are avoided. The judicious positioning of the left and right stereo channel, as well as the audio samples on even- or odd-number instants within the interleaving scheme, are key parameters to the success of the concealment strategy. For the DVD a much more powerful error correction has been developed. Enhanced error correction capability is required for several reasons. Frrst, the increased physical density implies that 485

5 ,physical imperfections affect proportionally more bits. It is further anticipated, since the system margins are much tighter, that the random error rate of DVD is larger than that of conventional CD. Secondly, as we can no longer rely on the concealment techniques used in the audio CD, the reliability of the decoded data must be much higher. As DVD is a true multimedia disc, its data integrity must be comparable to that of computer data. The Reed-Solomon Product Code (RS-PC), like CIRC, uses a combina-. tion of two Reed-Solomon (RS) codes denoted by C 1 and C 2 4 In CIRC, C 1 is [32,28] and C 2 is a [28,24] code, where [n,k] denotes a code with k input and 11 output bytes. The redundant n-k bytes are generated under the rules of the RS code. The rate of CIRC is 24/32 = 3/4. In other words, under CIRC rules one redundant byte is added to three user bytes. The RS-PC code is a classical product code where, as in CD, two codes cooperate. A product code is also part of the format of the digital audio tape (OAT) audio recorder. The two codes form the two dimensions of a rectangle. In RS-PC, the C 1 and C 2 codes are significantly longer than in CIRC, namely [208,192] and [182,172]. As a result, the rate of RS PC is much higher than that of CIRC, namely. 172*192/(182*208) = Sixteen sectors of 2,048 bytes user data make one ECC block. Figures 4 and 5 depict schematically the principle of operation of the CIRC and RS-PC codes. The C 1 and C 2 codes cim be represented by the rows and columns of a matrix. The combination of C 1 and C 2 is designed in such a way that in RS-PC the parities generated by cl are also taken into account in code C 2 These "checks on checks" have the advantage that the error correcting capability of the codes in tandem is improved relative to the CIRC code where this "double" check is absent. 486 TECHNICAL PAPER The product - matrix, or structure of the RS-PC code - cures a difficulty of the CIRC code, namely, its convolutional structure. In a cross-interleaved structure the data are formed into an endless array and the codewords are produced on columns and diagonals. Block structures, such as in RS-PC, are much better adapted to the data world, which operates in small segments of information. The CIRC structure of interleaving was specifically designed for very long, nonblocked data segments, such as digital audio or video. Other examples of cross-interleaved structures are found in the professional audio and video recorder formats. The disadvantage of a product code structure relative to a CIRC structure is the requirement of twice the memory capacity. 4 As a result of the fallen price of solid-state, this is a less important requirement than in 1979, when the CD was designed. The maximum correctable burst length is approximately 500 bytes (2.4 mm) for CIRC, while it is 2200 bytes (4.6 mm) for RS-PC. RS-PC is capable of reducing a random input error rate of 2 x 10 2 to a data error rate of 10 15, which is a factor of 10 better than in CD. EFM Recording Code The EFM code (and the EFMPlus code, to be discussed later) is a member of the family of de-free runlengthlimited (RLL) codes. The number of 32 Figure 4. Block diagram of CIRC encoder. sequential like symbols in a (binary) sequence is known as runlength. A runlength-limited sequen~ is a sequence of - binary symbols characterized by two parameters, Tmin = (d::rl) and Tmax = (k+ 1 ), which stipulat~ the minimum and maximum runlength, respectively, that may occur in the sequence. The parameter d controls intersymbol interference when the sequence is transmitted over a bandwidth-limited channel. The maximum runlength parameter k ensures adequate frequency of transitions for synchronization of the read clock. There are two reasons why EFM suppresses the low-frequency components. First, the servo systems for track following and focusing are controlled by low-frequency signals, so that low~frequency components of the information signal could interfere with the servo systems. Second, low-frequency disturbances resulting from fingerp]\ints on the disc can be filtered out withqut distorting the data signal itself. Under EFM rules, the data bits are translated eight at a time into 14 channel bits, with minimum runlength parameter d = 2 and a maximum runlength parameter k = 10 channel bits (this means at least 2 and at most 10 successive zeros between successive ones). Part of the EFM coding table is presented in Table 4, which shows the decimal representation of the 8-bit source word (left column) and its 14- bit channel representation. The codewords are described in such, a way that a "one" represents a transition of either positive or negative polarity, and a "zero" represents the absence of a transition. At least two bits, called merging bits, are required to ensure that the runlength conditions continue to be satisfied when the codewords are cascaded. If the runlength is in danger of becoming too short, "zeros" are chosen for the merging 28.. L CIRC

6 / 1728 lob---- N ::: \0 I -25 '----'--'-"~LW...L.L--'-~-'--I...WU...WL-...i-.c...>~'-'-:'-':-' Freq. 1/fb Figure 5. Block diagram of RS-PC encoder. Figure 6. Spectrum of conventional EFM. Both axes are normalized for fixed user bit rate fb. bits; if it is too long a "one" is selected for one of them. If we do this, we still retain a large measure of freedom in the choice of the merging bits. This freedom is used for minimizing the low-frequency content of the signal. In itself, two merging bits would be sufficient for continuing to satisfy the runlength conditions. A third merging bit was added, however, to give sufficient suppression of the low-frequency components. The measure of the low-frequency content is the running digital sum (RDS), 1 which is the difference betwee~ the totals of pit and land lengths accumulated froin the beginning of the disc. The system now opts for the merging combination that makes "the RDS at the end of the second codeword as close to zero as possible. The power spectral density (PSD) function of conventional EFM has been obtained by computer simulation. Results are plotted in Fig. 6. Description of the EFMPlus Encoder The aim of our investigations was the design of a new "EFM-like" code having a higher rate than EFM. The most important design issue is that critical parameters such as low-frequency (lf) content and timing should definitely not be worsened. 1\bese parameters are considered to be $tremely critical, as they affect the seivos and the timing recovery, which are the Achilles' heels of the system. The principle of operation of the EFMPlus encoder can be represented by a finite-state machine with an 8-bit input, a 16-bit output, and four states that are functions of the (discrete) time. The states are connected by edges, and the edges, in turn, are labeled with tags called words. A word in this context is a 16-bit sequence that obeys the prescribed ( d = 2, k = 10) constraints. Each of the four states is characterized by the type of words that enter, or leave, the given state. The states and word sets are characterized as follows: Words entering State 1 end with {0, 1 } trailing zeros Words entering State 2 and 3 end with {2,..., 5} trailing zeros Words entering State 4 end with { 6,..., 9} trailing zeros The words leaving the states are chosen in such a way that the concatenation of words entering a state and those leaving that state obey the (d = 2, k = 10) channel constraints. For example, words leaving State 1 start with a runlength of at least two and at most nine zeros. In an analogous manner, we conclude that words leaving State 4 start with at most one zero. Obviously, the sets of words leaving State 1 or 4 have no words in common. Words emerging from State 2 and 3 comply with the above runlength constraints, but they also comply with other conditions. Words leaving State 2 have been selected so that the first most significant bit (MSB), x 1, and the 13th bit, x 13, are both equal zero. Words leaving State 3 have x 1 x 13 :t:. 00. Any walk through the graph, stepping from state to state, produces a (d = 2, k = 10) constrained sequence by reading the words tagged to the edges that connect the states. It can be verified that from each of the states at least 351 words are leaving. An encoder is constructed by assigning a source word to each of the 351 edges that leave each state. Excess edges are removed. As a result, each edge in the graph now has two labels; 487

7 I l '' r; namely, a 16-bit word and a source word numbered from 0 to 350. Given the source word and the encoder state, the encoder will transmit the word tagged to the same edge as the source word at hand. It is immediate which codeword will be sent, and also which state will be next. A sequential input will define a walk in the graph, and as said, this walk generates a ( d = 2, k = 10) constrained sequence. As a result, the above finite-state encoder graph is a (d = 2, k = 10) RLL encoder that freely accommodates 351 source words. The encoder requires accommodation for only 256 source words. The excess, 95 words, will be used for controlling the low-frequency power. This will be discussed in a later section. The encoder graph is defined in terms of three sets: the inputs, the outputs and the states, and two logical functions: the output function and the next-state function. The specific codeword, x 1, transmitted by the encoder at instant t, is a function of the source word b 1 that enters the encoder, but depends further on the particular state, s 1, of the encoder. Similarly, the "next" state at instant t+ 1 is a function of x 1 and s 1 The output function h(.) and the next-state function g(.) can be succinctly written as x 1 :::: h(b 1, s 1 ) st+ 1 = g(b 1, s 1 ) (4) Both the output function h(.) and the next-state function g(.) are described by four lists with 351 entries. A part of the output function and the next-state function is listed in Table 5. This table. has an entry column that describes the source (input) word i by an integer between 0 and 255. The table also shows h(i,s) the 16-bit output to a particular input i when the encoder is in one of the four states s. A "one" means, as in EFM, a transition pitlland or land/pit, while a "zero" means the absence of such a transition. The third, fifth, seventh, and ninth columns show the next state function g(i,s). For example, let the encoder. graph be initialized at State 1, and let the source sequence be 8, 3, 4. The response to input 8, while being in State 1, equals h(8,1) = (Table 5). The new state becomes g (8,1) = 3. As a result, the response to input 3, while now being in State 3, is In the next clock cycle, the encoder state becomes g(3,3) = 2. From State 2 with the input equal to 4 we find fromthe table that the corresponding output is h(4,2) = The decoder translates 16-bit words into 8-bit data words. It does not suffice to look at the 16-bit word only. The decoder must also look at symbols at positions 1 and 13 of the upcoming codeword, namely (5) Thus, decoding of the new code is done by a logic array that translates (16+2) channel bits into 8 bits. In contrast, under EFM rules, it suffices to observe 14 of the 17 channel bits that constitute an EFM codeword. The encoder defined above can freely accommodate 3 51 source words. In order to make it possible to use a unique 26-bit sync word, 7 candidate words were barred, leaving a code size of 344. As only accommodation for 256 source words is needed, the surplus words can be exploited for minimizing the power at low frequencies. The suppression of low-frequency components, or de-control, is done by controlling the running digital sum (RDS). The 88 surplus words are used as an alternative channel representation of the source words 0,...,87. The full encoder is described by two tables called main and substitute table, respectively (Fig. 7). The source words 0,...,87 can b,e represented by the designated entries of the main table or alternatively by the entries of the substitute table. For source words 0,...,87 the encqder opts for that particular representation from the main table or the substitute table that minimizes the absolute value of the RDS. The power spectral density of the new code has been computed by a computer program that simulated the encoder algorithm. The results are plotted in Fig. 8. During the DVD debate, an alternative "EFM-like" code of 8/15 rate was discussed. As there is no such thing as a free lunch, the code will have much more low-frequency (lf) contt:nt than EFMPlus (Fig. 9). Although the extra 6% of density increase this 8/15 rate code offers is very attractive, the increased If-content of the code (16 to 18 db above EFM level) may impose Main Table Substitute Table Figure 7. Block diagram of EFMP/us encode~

8 0 / / 5 10 iii' :!2. 0 "' ~ le t j Freq. 1/fb Freq. 1/lb Figure 8. Spectrum of EFMPius. The upper curve shows the spectral density without look-ahead; lower curve shows the spectral density using 3 bytes look-ahead. Figure 9. Spectrum of 8115 rate "EFM-Iike" code. too great a risk for the servo systems and data recovery. This is particularly so sipce typical tolerances, such as defocusing and mistracking, are approximately two times tighter than with CD. Conclusion The DVD is a new optical recording medium with a storage capacity seven times higher than the conventional CD. The user capacity of the single-layer disc is 4.7 Gbytes, and the dual-layer disc increases the data capacity to 8.5 Gbytes. A significant part of the capacity incr~ase has been achieved by the use of optics, shorter laser wavelength, and larger numerical aperture, which reduces the spot diameter by a factor of The track formed by the recorded pits and lands, as well' as the track pitch, can be reduced by the same factor. The system requirements of the DVD and the related channel coding have been outlined. The conventional GIRC and EFM codes have been replaced by a more powerful error correction, RS-PC, and recording code, EFMPlus. As a result, the format efficiency of the DVD relative to audio CD is improved by 32% (49% with respect to CD-ROM). Even though consuming a significantly lower data overhead, the error correction code can cope with longer bursts and burst sequences, and more random errors. References 1. K. A. S. Immink, "EFMPlus: The Coding Format of the MultiMedia Compact Disc," IEEE Trans. Consumer Electr., CE-41: , Aug , Coding Techniques for Digital Recorders, Prentice-Hall Inti. (U.K.) Ltd., Englewood Cliffs, N.J., J. Watkinson, The Art of Digital Audio, Focal Press, London, S. B. Wicker and V. K. Bhargava, eds., Reed-Solomon Codes and Their Applications, IEEE Press, Kees A. Schouhamer Immink was born in Rotterdam, The Netherlands. He received an M.S. in electrical engineering in 1974 from the Eindhoven University of Technology, The Netherlands, and a Ph.D. from the same university in He has spent his entire professional life at the Pfiilips Research Laboratories in Eindhoven, where he currently holds the position of Research Fellow. In addition, he is adjunct professor at the Institute of Ex"perimental Mathematics, Essen University, Genfany. Dr. Immink ~as contributed to the design and develppment of a variety of digital audio recorders, such as the compact disc, compact disc video, THE AUTHOR DAT, DCC, and, recently, the DVD. He holds 30 U.S. patents, is coauthor of three books, and has written numerous papers in the field of optical and magnetic recorders. Immink has received many awards for his part in the digital audio and video revolution. He was recognized with the AES Silver Medal in 1992, the lee Sir J.J. Thomson Medal in 1993, the SMPTE Poniatoff Gold Medal for Technical Excellence in 1994, and the IEEE Masaru lbuka consumer electronics award in Immink has served as program chairman and conference chairman in various international conferences over the years. A member of the SMPTE, he currently serves on the Board of Editors of the SMPTE Journal. He is a Fellow of the AES, lee, and IEEE, and serves as a governor of the AES and the IEEE Information Theory Society. In 1996 he was elected to life membership in the Royal Netherlands Academy of Arts and Sciences, one of the highest honors that can be accorded a scientist or engineer. 489