Data link layer

OSI model blade.

The data link level, or data link layer, is the second layer of the OSI model, it is responsible for the reliable transfer of information through a data transmission circuit. It receives requests from the network layer and uses the services of the physical layer.

The objective of the link layer is to ensure that information flows, error-free, between two machines that are directly connected (connection-oriented service). To achieve this goal it has to assemble blocks of information (called frames at this layer), provide them with a link layer address (MAC Address), manage error detection or correction, and deal with "flow control" between computers (to prevent a faster computer from overwhelming a slower one).

When the means of communication is shared between more than two teams, it is necessary to arbitrate its use. This task is performed in the medium access control sublayer.

Within the IEEE 802 group of standards, the logical link sublayer is included in the IEEE 802.2 standard and is common for all types of networks (Ethernet or IEEE 802.3, IEEE 802.11 or Wi-Fi, IEEE 802.16 or WiMAX, etc..); they all specify a medium access sublayer as well as a separate physical layer.

Other types of link layer protocols are: Point-to-Point Protocol (Point-to-Point Protocol, PPP); high-level link protocol (High-level Data Link Control, HDLC), among others.

In practice, the medium access sublayer is usually part of the communications card itself, while the logical link sublayer would be in the card's adapter program (driver).

Plots

The link layer is the wide area facility over which systems can communicate using a data link layer protocol.

Functions

The data link layer is responsible for the reliable transfer of information over an electrical data transmission circuit. Data transmission is carried out through frames, which are the units of information with logical sense for data exchange at the link layer. It must also be taken into account that in the TCP/IP model it corresponds to the first layer.

Its main functions are:

1. Initiation, termination and identification.
2. Segmentation and blockage.
3. Synchronization of octeto and character.
4. Delimitation of plot and transparency.
5. Bug control.
6. Flow control.
7. Bug recovery.
8. Communication management and coordination.

Initiation, termination and identification

The initiation function comprises the processes necessary to activate the link and involves the exchange of control frames in order to establish the availability of the stations to transmit and receive information.

The termination functions are to release the resources occupied until the reception/sending of the last frame. Also to use control frames. The identification is to know to which terminal a frame should be sent or to know who sent the frame. It is carried out using the link layer address.

Segmentation and Grouping

Segmentation arises from the length of the frames, since if it is very long, smaller frames must be made with the information from that excessively long frame.

If these frames are excessively short, efficiency-enhancing blocking techniques must be implemented by concatenating several higher-level short messages into a single, longer link-layer frame.

Octet and character synchronization

In information transfers in the link layer, it is necessary to identify the bits and know what position corresponds to them in each character or octet within a series of received bits.

This synchronization function comprises the processes necessary to acquire, maintain and recover character or octet synchronization. That is to say, put in phase the encoding mechanisms of the sender with the decoding mechanisms of the receiver.

Frame Delimitation

The link layer should take care of frame delineation and synchronization. Three methods can be used for delimitation:

1. The first is "principle and end" (specific characteristics to identify the beginning or end of each plot).
2. You can also use "principle and account" (use a character to indicate start and followed by a counter that indicates its length).
3. Finally you can use the "guion" (a specific grouping of bits is used to identify the beginning and end by flags or flags).

“Transparency” is achieved by ‘bit insertion’. It consists of counting the consecutive ones and when it finds five characters "1" followed and consecutive (11111) introduces the bit "0" after the fifth 1. Example: consider the frame 0101111110, when applying transparency it becomes 01011111010.

Error handling

Provides detection and correction of errors in the forwarding of frames between computers, and provides control of the physical layer. Its functions, in general, are:

• Identify data plot;
• Detector codes and error fixers;
• Flow control;
• Communication management and coordination.

Error correctors: it is optional in this layer, the one in charge of carrying out this function is the transport layer, in a WAN it is very probable that the verification is carried out by the link layer.

For the identification of frames you can use different techniques such as:

• Character Counter
• Start and end features with filler characters
• Sequence of bits start and end indicator, with fill bits

Flow control is necessary to avoid 'overwhelming' to the receiver. It is usually done at the transport layer, sometimes also at the link layer. Use feedback mechanisms. It is usually coupled with error correction and should not limit the efficiency of the channel.

Error control methods are basically two:

• Correction of errors forward (Forward Error Correction, FEC) or error correction in advance, has no flow control.
• Automatic repeat request (Automatic Repeat-reQuest, ARQ): has flow control through stop and wait, and/or slide window.

Possible implementations are:

• Stop and wait simple: the transmitter sends the plot and expects a signal from the receiver to send the next plot or the one that has just sent in case of error.
• Continuous shipment and simple rejection: the transmitter continuously sends plots and the receiver is validating them. If you find a wrong plot, remove all subsequent ones and ask the transmitter to send from the wrong plot.
• Continuous shipment and selective rejection: Continuous transmission except that it only relays the defective plot.

Error detection is carried out through various types of codes that must be highlighted:

• Control of cyclical redundancy (CRC)
• Simple parity
• Cross parity (Horizontal and Vertical Parity)
• Verification amount

Error corrections are based on Hamming Code, by repetition, cross-parity checking, Reed-Solomon and Goyle.

Flow control

The flow control is necessary to avoid saturating the receiver from one or more senders. It is usually done at the transport layer, sometimes also at the link layer. Use feedback mechanisms. It is usually coupled with error correction and should not limit the efficiency of the channel. Flow control involves two very important actions that are error detection and error correction.

The error detection is used to detect errors when sending frames to the receiver and try to solve them. It is done through various types of codes of which CRC must be highlighted, simple parity (can be even, "1" even, or odd numbers), cross parity (horizontal and vertical parity) and checksum.

The correction of errors arises from the detection to correct detected errors and they need to add to the useful information a number of redundant bits much higher than what is necessary to detect and retransmit. His techniques are varied. The Hamming Code, Repetition, that each bit is repeated three times and in case of failure the bit that is repeated the most is taken; It can also be done by cross parity checking, Reed-Solomon, and de goyle.

It is also worth noting the HDLC protocols, which is a high-level data link control, bit-oriented and obeys a sliding or continuous window ARQ. There are also character-oriented protocols.

Failover

Refers to the procedures to detect situations and recover from abnormal situations such as no response, reception of invalid frames, etc. The most typical situations are the loss of frames, the appearance of duplicate frames and the arrival of frames out of sequence.

If these events are not handled correctly, information will be lost and erroneous data will be accepted as if they were correct. Generally, counters are used to limit the number of errors or retries of processes and procedures. You can also use timers to set timeouts for events.

Management and coordination of communication

Management and coordination.

Management attends to two types:

1. The first is a centralized system where there is a master machine and several slaves. These connections can be made from point to point or multipoint.
2. The second is the distributed, where there is no master machine and all compete for the control of the communication system.

Coordination can be done by selection or contention:

• The selection can be implemented by polling/selection, where the teacher collects a message from a secondary school and is delivered to whom he selects. It is also possible to assign a witness to a machine that can issue messages/tramas. The Token Ring and Token Bus configurations are typical.
• The contest is based on that each computer emits its plot/message when you want it. All components of the network are both emitters and receivers. ALOHA and CSMA/CD systems are typical. You have to be careful with collisions.