Telairity Dives Deep into Digital Video Technology Part 1 In an age when data is increasingly digital, and video is consuming a disproportionate and ever-increasing share of all digitized data, digital video is a broad and important subject. For example, looking at the great worldwide superhighway for digital data, the Internet, it is currently estimated that Netflicks video streaming service, by itself, consumes over onethird of all U.S. Internet traffic during peak hours. If you aggregate Netflicks consumption with the second-ranked video streaming service (YouTube), then these two companies, by themselves, account for over half the peak Internet traffic in North America. Why does digital video require so much data? Can anything be done about it? And what about everything else? In the Digital Era, information of all types text, graphics, images, sounds, in short, anything that can be communicated by eye or ear is recorded, stored, transmitted, and manipulated in digital form. One benefit of understanding digital video, then, is that it opens up an understanding of the wider world of digital information which is to say, today s whole world of information technology. My goal in this series of short articles on digital video technology is to explain the subject of digital video in particular, and digital information more generally, in a way that is both clear and comprehensive. No background in video, digital video, or digital information, in general, is either assumed or required to follow this discussion. Only 10 types of people: those who understand binary, and those who don t For this reason, I will start with the fundamental topic of binary numbering. A clear understanding of the whys and hows of binary numbers is foundational to all discussions of digital information and the so-called digital revolution. 1 Binary is simply a numbering system where there are only two base symbols available for use in counting conventionally, the digits 0 and 1. Of course, this does not mean it is impossible or even difficult to count higher than 1 in binary, any more than the fact that there are just ten base digits in the decimal numbering system, 0 9, makes it impossible or even difficult to count higher than 9. In any numbering system, regardless of the number of base symbols supported (two, ten, or any other number), the secret to unlimited counting is simply to add places. When 1-place numbers (base digits) are exhausted, move to 2-place numbers, then (when two-place combinations are exhausted) to numbers with 3-places, 4- places, 5-places, and so on, to any number of places that may be desired or needed. A digression on digits Digit, as used here, just means a base or 1-place number symbol. The digital era is the era where everything we can characterize as information gets reduced to numbers. The English word digit itself comes from the Latin word for finger or toe. Our familiar decimal numbering system, which uses a set of ten base digits that starts at 0 and runs to 9, is probably related to the fact that numbering doubtless got its start by our remote 1 If you already have a clear understanding of binary numbering, you may want to skip ahead to later and more advanced topics.
ancestors counting on their fingers and, before the invention of shoes, likely on their toes as well. If the usual number of fingers per hand and toes per foot were six rather than five, the prevalent numbering system probably would be duodecimal or base-twelve, rather than decimal or base-ten. Alternatively, if we were octopuses, we probably would have an octal or base-eight numbering system. The main point here is that the number of digits in a numbering system is arbitrary. Our decimal system happens to be base-ten, but with a different number of appendages to count on, it might as easily have been base-eight or base-twelve. There is nothing special or privileged about the base-ten system in common use today. Note that, if we use all ten base digits in the decimal system to count fingers or toes, the sequence begins with 0 and ends with 9, producing the odd consequence that our tenth finger or toe is numbered nine rather than ten. The beginning of nothing Programmers quickly become accustomed to thinking of zero simply as the first counter. Soon, they instinctively understand that a sequence in which 255, for example, is the last value, is 256 steps long, because it starts with an initial value of zero and step 1 is, in fact, its second step. For the rest of us, however, this way of thinking about zero, as simply the first counter in a sequence of counters, takes some adjustment. Our instinct is to think of zero as meaning literally nothing. For us, zero is not the name of the first penny in a dollar, but rather means you are broke and have no pennies at all. In this more mundane world, the first penny in a dollar is penny 1 (not penny 0), and its last penny is penny 100 (not penny 99).
In truth, the idea of seeing zero as a number at all, alongside other numbers, is far from obvious and ranks as one of the more notable accomplishments of human ingenuity. As far as we know, the world s first formal numbering system 2 entirely lacked any symbol for zero, an innovation that had to wait another thousand years or so. 3 However, so slippery is the idea of nothing as something, that zero then vanished again, for yet another thousand years or so, 4 until either rediscovered or independently reinvented by Indian mathematicians, between 400 and 500 AD. 5 The value of two As humans, we learn the decimal numbering system from an early age. Decimal is humanly convenient, not just because we happen to have ten fingers and ten toes, but because scaling up and down by factors of ten is very useful when thinking about things both large and small. However, binary numbering is the basis of the digital era, not the familiar and convenient decimal numbering system. Indeed, the Digital Era could, more precisely, be termed the Binary Era. Hence, the first and most critical prerequisite to understanding digital information is to understand the binary numbering system. Binary (rather than octal or decimal or duodecimal or any other base value) is the foundation of the digital information age for a good reason. Since it uses just two base digits, the minimum possible number, it is not only the simplest but also the most robust and least error-prone system for communicating information. In Goldilocks terms, when it comes to communication, one digit is too few, three digits are too many, but two digits are just right. 2 Developed 4 to 5 thousand years ago in Sumer, perhaps the first of the early fertile crescent civilizations that arose where the Tigris and Euphrates rivers drain into the Persian gulf (a region that today is part of Iraq). 3 Until introduced by the Babylonians, a successor fertile crescent civilization to the Sumerians, that flourished during the second millennia B.C. 4 To name just two later examples, neither Greek nor Roman numbering systems included a zero symbol. 5 Ninth-century Arabian mathematicians (the source of our decimal Arabic numbering system), in turn, seem to have gotten the idea of a zero symbol from their study of these earlier fifth-century Indian sources.
The case of the distant flagman An example may help make this point clear. Suppose we have a message that we need to communicate from point A to point B, and our only means of communication is hand signals, made more visible, say, by a flag held by a signalman stationed at point A. We could, of course, invent an elaborate set of signals based, say, on two distinct flags, a big flag in one hand and a little flag in the other, pointing simultaneously to two different spots on the face of an imaginary clock. Suppose, in a military context, the flagman s job was to communicate both the direction and distance of an approaching enemy. Then, using the above scheme, the big flag might indicate the direction of their approach 12 for straight north, 1 for north-north-east, 2 for east-north-east, 3 for straight east, and so on while the little flag indicated their distance in thousands of yards or meters. Although this sort of elaborate scheme is capable of conveying a lot of information very concisely, it suffers to a similar degree from the flaw of unreliability. Not only does it depend on the flagman maintaining two distinct flags and retaining use of both arms, but also, in dim light or other conditions affecting visibility (smoke, fog), it might be hard to tell the two flags apart, or determine exactly where they are pointing. The solution to these problems of potential miscommunication is to simplify the communication scheme. Rather than two flags, use just one, a flag so big it takes two hands to hold it comfortably. And, rather than twelve positions, use just two, say, up and down. By thus simplifying the flag scheme to its bare minimum (one flag, two positions), we create the widest possible buffer against possible miscommunication. As long as it is possible to communicate by hand signals at all, despite all the potential visual hazards (darkness, fog, mist, smoke, rain, etc.), this simplified up-down binary scheme will continue to work long after systems requiring finer visual distinctions have broken down into hopeless ambiguity.
Binary robustness By definition, then, binary signals do away with all intermediate states. They provide only two alternatives, variously designated as: up/down, high/low, on/off, black/white, true/false, yes/no, 1/0. Restricting signals to one of just two alternative values necessarily minimizes the chances of a receiver mistaking the message sent. In short, as long as it is possible to communicate any distinction at all across an intervening sea of interference and noise, of all the possible messages we might send, a binary signal is the most likely to get through intact. The basic bit The tradeoff for maximizing signal robustness is minimizing the amount of information carried by a signal. A single binary value only provides 1 of 2 states. This minimum 2-state value is called a bit of information and, indeed, is the least amount of information it is possible to convey. If all you can transmit is a single state to go back to our flagman example, if the only flag position is up then there is no need to transmit anything at all. The single value up can be written down without bothering to look at the flag or, indeed, without bothering even to have a flagman. So, to communicate anything meaningful, it is necessary to have a minimum of two states. We need a flagman and we need to look at their flag to see whether it is in the up or down position. Just this much and no more is real information, conveyed across distance. Though, admittedly, it is not a lot of information. In fact, it is the least amount of information that can be transmitted, namely, a single bit of information. The need for length Of course, if all we ever needed to know were a simple yes/no, true/false, if the only two colors we ever used were black/white, if the only two quantities we needed were 0/1, then getting information in minimal binary bits would not be a problem. But what about all the cases where we need to convey more complex information? What if there is a range of possible maybes, rather than a definite yes or no; say, 50 shades of gray, rather than simple black or white; or many different quantities, beyond nothing and everything? The solution to this problem of complexity is not to surrender binary s irreducible simplicity and associated maximal certainty, but rather to make up what binary lacks in its base variety by resorting to length. That is to say, while an individual bit can convey only two states, two bits in combination can convey four states, three bits in combination eight states, and so on. We will go over all this in more detail in the next part of this series, where I will also present what I believe is, by far, the easiest system for translating unfamiliar binary place values into familiar decimal place values. Telairity has made a name for itself as one of the industry s leading providers of video encoding solutions. Please write to us at sales@telairity.com to learn more about our products and to collaborate with our team.