Once data is transmitted outside of a computer, there is a much increased probability that bit errors (corruptions) will occur. Therefore, in most applications it is necessary to incorporate not only a means to detect when bit (transmission) errors occur, but also a way to obtain another (hopefully correct) copy of the affected data. The above process encompasses what is known as error control.

Error detection and correction is of fundamental importance in data manipulation and transmission - Garbage In causes Garbage Out. The vast majority of errors introduced into data, occur in the transmission - the main causes being:

• Noise - refers to the addition of extraneous signals to the transmitted information which, if sufficiently strong, can lead to altered bit values. Noise can come from a number of sources including:
  • crosstalk from signals on adjoining lines
  • continuous thermal noise from heat sources nearby
  • impulse noise from lightning, switches and components acting intermittently
  • electromagnetic noise from adjacent sources of electrical energy
• Distortion - refers to the altering of signals caused by the different physical characteristics of individual lines - different levels of resistance in each line can result in skewing in parallel data transmission.
• Others - some errors may be the result of faulty components which function intermittently, short breaks in the transmission path and so on.

Depending on the duration of the interference causing the error it may affect only a single bit or group of bits. In many cases the receiver will be unable to distinguish erroneous data from correct data - the corrupted data may still appear correct. It is therefore essential that data transmission systems incorporate some kind of error detection mechanism. Two main forms of error detection are:

• parity bits and checksums

However, it should be noted that these methods do not always provide a mechanism for correcting the error.