Question

A Network engineer needs to make a decision as of where to do error handling, in the Data Link Layer at frame level or in the TCP Layer at packet level.   In most of the error handling protocols, when there is an error, it has to re-transmit the frame or the packet respectively.   In general there are multiple frames in a packet and it costs more to re-transmit a packet than a frame. On the other hand, the overhead cost for initiating error checking is higher for multiple frames than the cost of a single packet, since there is more than one frame within a packet.

Given a physical link with certain quality measurement, p < 1, as the probability of error for a frame, by common sense concerns, in what case you would like to do error handling at the Data Link Layer or at the TCP Layer? (You need to justify your decision in best (case where almost p=0) and worst case (where almost p=1) scenarios with respect to the quality of physical layer.

Answer 1

Answer :

At the data link layer, the sender transmits data to the receiver in the form of data-frames. The errors can only be detected in the frames and such frames can be discarded without asking for its retransmission.

The receiver can determine if the frame is damaged or not after it accepts it from the sender. In case if the frame is demaged, it is directly discarded. Since receiver does not ask for re-transmission of damaged frame, the probability of error (i.e. p ) is high at the datalink layer ( p = 1 )

At the transport layer, the sender transmits data to the receiver in the form of packets, which are recomposed by the receiver. The errors can be detected as well as it can ask for re-transmission of the entire packet containing multiple data frames. This can be a disadvantage. But also the transport layer, includes checksum for error correction purpose.   Hence, the probability of error is less at the transport layer ( 0 < p < 1 )

Therefore, the error handling has to be done at transport layer.