The task of the data link layer is to convert the raw bit stream offered by the
physical layer into a stream of frames for use by the network layer. The link layer
can present this stream with varying levels of reliability, ranging from connectionless,
unacknowledged service to reliable, connection-oriented service.
Various framing methods are used, including byte count, byte stuffing, and bit stuffing. Data link protocols can provide error control to detect or correct damaged frames and to retransmit lost frames. To prevent a fast sender from overrunning a slow receiver, the data link protocol can also provide flow control. The sliding window mechanism is widely used to integrate error control and flow control in a simple way. When the window size is 1 packet, the protocol is stop-and-wait.
Codes for error correction and detection add redundant information to messages by using a variety of mathematical techniques. Convolutional codes and Reed-Solomon codes are widely deployed for error correction, with low-density parity check codes increasing in popularity. The codes for error detection that are used in practice include cyclic redundancy checks and checksums. All these codes can be applied at the link layer, as well as at the physical layer and higher layers.
We examined a series of protocols that provide a reliable link layer using acknowledgements and retransmissions, or ARQ (Automatic Repeat reQuest), under more realistic assumptions. Starting from an error-free environment in which the receiver can handle any frame sent to it, we introduced flow control, followed by error control with sequence numbers and the stop-and-wait algorithm. Then we used the sliding window algorithm to allow bidirectional communication and introduce the concept of piggybacking. The last two protocols pipeline the transmission of multiple frames to prevent the sender from blocking on a link with a long propagation delay. The receiver can either discard all frames other than the next one in sequence, or buffer out-of-order frames and send negative acknowledgements for greater bandwidth efficiency. The former strategy is a go-back-n protocol, and the latter strategy is a selective repeat protocol.
The Internet uses PPP as the main data link protocol over point-to-point lines. It provides a connectionless unacknowledged service, using flag bytes to delimit frames and a CRC for error detection. It is used to carry packets across a range of links, including SONET links in wide-area networks and ADSL links for the home.
Various framing methods are used, including byte count, byte stuffing, and bit stuffing. Data link protocols can provide error control to detect or correct damaged frames and to retransmit lost frames. To prevent a fast sender from overrunning a slow receiver, the data link protocol can also provide flow control. The sliding window mechanism is widely used to integrate error control and flow control in a simple way. When the window size is 1 packet, the protocol is stop-and-wait.
Codes for error correction and detection add redundant information to messages by using a variety of mathematical techniques. Convolutional codes and Reed-Solomon codes are widely deployed for error correction, with low-density parity check codes increasing in popularity. The codes for error detection that are used in practice include cyclic redundancy checks and checksums. All these codes can be applied at the link layer, as well as at the physical layer and higher layers.
We examined a series of protocols that provide a reliable link layer using acknowledgements and retransmissions, or ARQ (Automatic Repeat reQuest), under more realistic assumptions. Starting from an error-free environment in which the receiver can handle any frame sent to it, we introduced flow control, followed by error control with sequence numbers and the stop-and-wait algorithm. Then we used the sliding window algorithm to allow bidirectional communication and introduce the concept of piggybacking. The last two protocols pipeline the transmission of multiple frames to prevent the sender from blocking on a link with a long propagation delay. The receiver can either discard all frames other than the next one in sequence, or buffer out-of-order frames and send negative acknowledgements for greater bandwidth efficiency. The former strategy is a go-back-n protocol, and the latter strategy is a selective repeat protocol.
The Internet uses PPP as the main data link protocol over point-to-point lines. It provides a connectionless unacknowledged service, using flag bytes to delimit frames and a CRC for error detection. It is used to carry packets across a range of links, including SONET links in wide-area networks and ADSL links for the home.
No comments:
Post a Comment