Errors in transmission and parity checks

Last time

Before we start: last time, in Packet switching, you learned these points. Here they are again on one card. Keep them in mind, because today builds on them.

A message that changed

You send the number five, but your friend's phone receives seven. One bit changed on the way: 0 0 0 0 0 1 0 1 became 00000111. Errors like this happen all the time on networks. So computers need ways to notice them.

Why errors happen

Electrical interference, from motors, power cables or the weather, can flip bits. Data can be lost: some bits, or a whole packet, never arrive. Data can be gained: extra bits appear that were never sent. And on parallel cables, bits can skew and arrive out of step.

What error detection does

Every error-detection method sends some extra data with the real data. The receiver uses that extra data to check whether what arrived matches what was sent. If it does not match, the data is sent again.

The parity check

The simplest method is the parity check. Sender and receiver first agree on even parity or odd parity. Each byte carries seven bits of data, and one extra bit, the parity bit, usually on the left. With even parity, the parity bit is chosen so the whole byte has an even number of ones. With odd parity, it is chosen so the byte has an odd number of ones.

Working out a parity bit

Here are two worked examples. The data 1 0 0 1 1 0 1 has four ones. For even parity, four is already even, so the parity bit is zero: 01001101. For odd parity, we need one more one, so the parity bit is one: 11001101.