Ch26 Coding Systems

Ch26 Coding Systems

1. Character Set

   • A table which maps each character with a unique binary number.
   • ASCII
     • American Standard Code for Information Interchange
     • 7 bits, so holds 128 characters.
     • More recently extended ASCII was released, 8 bits, 256 characters.
     • Only contains basic English characters and punctuation.
     • Not enough: advent of World Wide Web created need for universal system.
   • Unicode
     • Was brought in to replace Unicode and include most characters from across the world.
       • Note: this includes Emojis!
       • Includes the characters for over 20 countries.
     • 16 bit so can code for around 6.5 thousand characters.
     • Constantly updated and maintained by the Unicode Consortium.

2. Error Checking

   • Parity Bit
     • Where an additional bit is added to the end of a binary number to show whether there are an even or an odd number of 1 digits.
     • Unreliable as if an even number of bits flip or the parity bit flips, it doesn't detect the error.
   • Majority Voting
     • Data is sent multiple (typically 3) times and if there is a discrepancy, the computer goes with the majority.
     • Very reliable and can correct data without retransmitting, however, uses a lot of bandwidth.
   • Check Digit
     • An additional digit is added to the end of a number which is derived from the number.
     • A simple method is by simply adding up all the digits, and adding up the digits on the sum, repeat this until you are left with a single number.
       • This is unreliable as if the order changes, the check digit remains the same.
     • A more advanced method would multiply each number by a weight, so as to ensure that order is needed. An example of this is the modulo-11 method.

3. Graphics

   • Bit Maps
     • A bit map graphic is a 2D array of pixels.
     • Each pixel holds a colour value, stored as a binary number, many thousands of these pixels typically make up an image.
     • A problem with bit maps is that quality degrades when you zoom in, additionally they take up a large amount of space.
     • Each pixel will have a colour depth, which is the number of bits allocated to represent the colour of a pixel, the higher the colour depth, the more colours that can be expressed.
   • Vector Graphics
     • A vector graphic is an image generated from a set of instructions and mathematical formulae. This generates the image consisting of geometric shapes, relative to a point of origin.
     • These are not suitable for photos as they simplify things significantly, however, they are good for diagrams and CAD/CAM images.
     • A vector graphic typically takes up little space and can be easily scaled up without losing quality.

4. Audio

   • Audio is converted to digital data by sampling the sound wave many times per second.
     • The higher the number of samples per second (measured in Hz), the higher the quality of the sound and the truer it is to the original.
       • The computer then synthesises a wave by extrapolating the data sampled, this is usually indistinguishable from the original for a human ear.
   • Nyquist's Theorem states that to faithfully recreate sound, you must record at least twice the highest frequency.
   • Humans hear 20Hz to 20kHz.
   • The resolution of sound is the number of bits allocated to each sample, hence, the more bits allocated, the more pressure levels each sample can represent.

5. Compression

   • Lossy compression
     • Some data is discard, to reduce file size, however, the new file is usually indistinguishable from the original.
     • An example would be replacing the hundreds of different shades of blue in an image of the sky, with just a few shades. A human would probably not notice the difference.
     • JPEG files use this.
     • It can be useful when transmitting images across the internet, as it reduces the bandwidth significantly. This is especially important for people on slow connections.
   • Lossless compression
     • Compression of data in such a way that the original data can be reacquired in it's totality, after decompressing. i.e. no data is discarded.
     • Examples would be run length encoding, where for example, if in an image, there are many pixels of the same colour, the computer just stores this as x*"blue" pixels. Rather than storing "blue pixel", "blue pixel"... This doesn't lose accuracy, it just encodes repeated data more simply. Another example is dictionary-based encoding, where commonly occurring strings in a text file are coded for in a simpler way, perhaps by assigning some token. A simple way to understand this could be replacing every "and" with an "&" this takes up 67% less space than writing "and". Again, no loss of accuracy as the original data can be easily reacquired.
       • Dictionary based encoding can be used on binary data also, if it is considered as a string of 1s and 0s.