Encyclopedia of Computer Science and Technology

186 error correctionSalvendy, Gavriel. Handbook of Human Factors and Ergonomics. 3rded. New York: Wiley, 2006.U.S. Department of Labor. Occupational Safety & Health Administration.“Computer Workstations.” Available online. URL:http://www.osha.gov/SLTC/etools/computerworkstations/index.html. Accessed July 30, 2007.Vredenberg, Karel, Scott Isensee, and Carol Righi. User-CenteredDesign: an Integrated Approach. Upper Saddle River, N.J.:Prentice Hall, 2001.For even parity, if the number of ones in the byte is odd, the paritybit is set to one to make the total number of ones even. Odd paritywould work the same way, except the parity bit would be set whennecessary to ensure an odd number of ones.error correctionTransmitting data involves the sending of bits (ones andzeros) as signaled by some alternation in physical characteristics(such as voltage or frequency). There are a numberof ways in which errors can be introduced into the datastream. For example, electrical “noise” in the line might beinterpreted as spurious bits, or a bit might be “flipped” fromone to zero or vice versa. Generally speaking, the faster therates at which bits are being sent, the more sensitive thetransmission is to effects that can cause errors.While a few wrong characters might be tolerated insome text messages or graphics files, binary files representingexecutable programs must generally be received perfectly,since random changes can make programs fail orproduce incorrect results. Data communications engineershave devised a number of methods for checking the accuracyof data transmissions.The simplest scheme is called parity. A single bit isadded to each eight-bit byte of data. In even parity, theextra (parity) bit is set to one when the number of ones inthe byte is odd. In odd parity, a one is added if the data bytehas an even number of ones. This means that the receiver ofthe data can expect it to be even or odd respectively. Whenthe byte arrives at its destination, the receiving programchecks the parity bit and then counts the number of onesin the rest of the byte. If, for example, the parity is evenbut the data as received has an odd number of ones, thenat least one of the bits must have been changed in error.Parity is a fast, easy way to check for errors, but it has someunreliability. For example, if there were two errors in transmissionsuch that a one became a zero and a different zerobecame a one, the parity would be unchanged and the errorwould not be detected.The checksum method offers greater reliability. Thebinary value of each block of data is added and the sumis sent along with the block. At the destination, the bits inthe block are again added to see if they still match the sum.A variation, the cyclical redundancy check or CRC, breaksthe data into blocks and divides them by a fixed number.The remainder for the division for each block is appendedto the block and the calculation is repeated and checked atthe destination. Today most modem control software implementsparity or CRC checking.A more sophisticated method called the Hamming Codeoffers not only high reliability but also the ability to automaticallycorrect errors. In this scheme the data and checkbits are encrypted together to create a code word. If theword received is not a valid code word, the receiver can usea series of parity checks to find the original error. Increasingthe ratio of redundant check bits to message bits improvesthe reliability of the code, but at the expense of having todo more processing to encrypt that data and requiring moretime to transmit it.Further Reading“Error-Correcting Code.” Wolfram MathWorld. Available online.URL: http://mathworld.wolfram.com/Error-CorrectingCode.html. Accessed July 31, 2007.Fung, Francis Yein Chei. “A Survey of the Theory of Error-CorrectingCodes.” Available online. URL: http://cadigweb.ew.usna.edu/~wdj/teach/ecc/codes.html. Accessed July 31, 2007.Wicker, S. B. Error Control Systems for Digital Communication andStorage. Upper Saddle River, N.J.: Prentice Hall, 1995.error handlingAn important characteristic of quality software is its abilityto handle errors that arise in processing (also calledrun-time errors or “exceptions”). Before it is released forgeneral use, a program should be thoroughly tested witha variety of input (see quality assurance, software).When errors are found, the soundness of the algorithm andits implementation must be checked, as well as the programlogic (see algorithm). Interaction between the programand other programs (including the operating system) aswell as with hardware must also be considered. (See bugsand debugging.)However, even well-tested software is likely to encountererrors. Therefore a program intended for widespread usemust include instructions for dealing with errors, anticipatedor otherwise. The process of error handling can bedivided into four stages: validation, detection, communication,and amelioration.Data validation is the first line of defense. At the “frontend” of the program, data being entered by a user (or readfrom a disk file or communications link) is checked to seewhether it falls within the prescribed parameters. (In thecase of a program such as a data management system, theuser interface plays an important role. Data input fieldscan be designed so that they accept only valid characters.