This error detection method is used by the higher-layer protocols (TCP/IP).
It is based on the concept of redundancy.
 

Checksum Generator

In the sender,

• The Checksum generator subdivides the data unit into equal segments of n bits (usually 16).
• These segments are added together using one's complement arithmetic in such a way that the total is also n bits long.
• The total (sum) is then complemented and appended to the end of the original data unit as redundancy bits, called the checksum field.
• The extended data unit is transmitted across the network.
• If the sum of the data segment is T, the checksum will be -T.

Checksum Checker

• The receiver subdivides the data unit
• The receiver then adds all segments together and complements the result.
• If the extended data unit is intact, the total value found by adding the data segments and the checksum field should be zero.
• If the result is not zero, the packet contains an error and the receiver rejects it.

• The checksum detects all errors involving an odd number of bits, as well as most errors involving an even number of bits.
• If one or more bits of a segment are damaged and the corresponding bit or bits of opposite value in second segment are also damaged, the sums of those columns will not change and the receiver will not detect a problem.
• If the last digit of one segment is a 0 and it gets changed to a 1 in transit, then the last 1 in another segment must be changed to a 0 if the error is to go undetected.
• Checksum retains all carries; so, although two 0s becoming 1s would not alter the value of their own column, they would change the value of the next higher column. But anytime a bit inversion is balanced by an opposite bit inversion in the corresponding digit of another data segment, the error is invisible.