To store information onto a physical surface, the data must be transformed into a series of marks using a
modulation code. The codes used in most optical disc systems are binary, meaning that resulting surface has only two states: marks and non-marks. The following figure illustrates the
EFM code used in CDs and DVDs: Because the edges are positioned on a grid that is finer than the minimum mark size,
EFM achieves about 1.5 bits per minimum-mark, even though it is a binary code. MultiLevel recording refers to the use of multiple reflectivity values to encode data onto an optical disc. By using more than two levels, more information can be put into the minimum feature size. The following figure illustrates a MultiLevel code (note: colors are used only to represent differences in intensity): The 8-level code used on the prototype systems is a
convolutional code, storing about 2.5 bits per data cell. By using this code in combination with a smaller mark size and a more efficient error-correction code, the capacity of CD media was tripled. When applied to dual-layer DVD, ML-recording can increase capacity by a factor of 1.9. MultiLevel optical recording is an example of
baseband pulse-amplitude modulation. While Non-binary, or M-ary, modulation is common in the
telecom industry, the technique was originally developed and patented for optical disc recording at Optex Corporation in the early 1990s (in conjunction with the University of Rochester) for use with their Electron Trapping Optical Media (ETOM). Although simple in principle, implementation of ML was challenging, in large part because data storage
channels are highly
nonlinear. The overall ML-system response is much more sensitive to variations in its individual components (
operating temperature, media uniformity, read-head fluctuation, etc.) than a conventional CD/DVD system. To compensate, the ML logical-format devotes a substantial portion of bits (as forward-error correction coding) to enhance robustness against media defects and
signal noise. ML-drives used sophisticated power-optimization during writing and adaptive equalization during reading. MultiLevel recording is sometimes confused with
multi-layer storage, in which multiple data surfaces are combined into a single disc. Multi-layer and multiLevel technique can be combined (as in dual-layer ML-DVD ROM), where ML-modulation is applied to each individual layer of the disc. ==References==