Duplex mode. Duplex radio communication
WiFi connections operates in half-duplex mode, and the wired part of the local network is in full duplex. Find out more by reading this article.
Duplex vs simplex
In networking, the term "duplex" refers to the ability for two points or devices to communicate with each other in both directions, as opposed to "simplex", which refers to unidirectional communication. In a full-duplex communication system, both points (devices) can send and receive information. Examples of duplex systems are telephones and walkie-talkies.
On the other hand, in a simplex system, one device transmits information and the other receives. Remote controller remote control is an example of a simplex system where the remote control transmits signals but does not receive them in response.
Full and half duplex
Full duplex communication between the two components means that both can send and receive information to each other at the same time. Telephones are full duplex systems because both parties can talk and listen at the same time.
In half-duplex systems, transmission and reception of information must occur alternately. While one point is transmitting, the others must only receive. Walkie-talkies are half-duplex systems, at the end of the transmission the participant must say “Receive”, this means that he is ready to receive information.
WiFi routers are devices that modulate and schedule information flows to and from any WiFi-enabled electronic device(such as a laptop or smartphone) to the Internet using a specific standard or protocol called IEEE 802.11, which operates in half-duplex mode. WiFi is only trademark for a specific IEEE standard.
WiFi devices connect to the router using 2.4 GHz or 5 GHz radio waves. The router guarantees correct distribution information flows between the connected device and the Internet; using a Time Division Calling (TDD) process that operates in full duplex mode.
TDD emulates full duplex communication by creating or dividing periods of time that alternate between transmitting and receiving. Data packets flow in both directions as dictated by the schedule. By precisely staggering these time periods, connected devices can transmit and receive simultaneously.
Most big problem To achieve full-duplex control over radio communications is intra-system interference. This is interference or noise more intense than the signal itself. Simply put, interference in a full duplex system occurs when one point transmits and receives at the same time, and also receives its own transmission, hence self-interference occurs.
Near full-duplex wireless communications are possible in research fields and scientific communities. This is largely achieved by eliminating self-interference at two levels. The first method is to invert the noise signal itself, and then the noise reduction process is further enhanced digitally.
What about a wired network?
The wired part of the local network exchanges data in full duplex mode using two pairs of twisted wires forming cable connection Ethernet. Each pair is designed to transmit and receive packets of information simultaneously, so there is no data collision and transmission occurs without interference.
Progress in WiFi communications
As part of the IEEE 802.11 protocol, changes have been made to achieve better range or better throughput, or both. From its inception in 1997 to 2016, wireless standards have evolved from 802.11, 802.11b/a, 802.11g, 802.11n, 802.11ac, and finally the latest 802.22. No matter how advanced they have become, they still belong to the 802 family, which will always operate in half-duplex mode. Although many improvements have been made, especially with the inclusion of MIMO technology, operating in half-duplex mode reduces the overall spectral efficiency by a factor of two.
It's interesting to note that MIMO supported by routers (with many inputs and many outputs) advertises much more high speeds data transmission. These routers use multiple antennas to transmit and receive multiple data streams simultaneously, which can increase the overall transmission speed. This is also common in 802.11N routers, which advertise speeds of 600 megabits per second and higher. However, since they operate in half duplex mode, 50 percent (300 megabits per second) throughput is reserved for transmission while the other 50 percent is used for receiving.
Full duplex WiFi in the future
To full duplex wireless communication commercial interest is growing. The main reason is that the progress in half-duplex FDD and TDD is not saturated. Improvements software,modulation advances and improvements in MIMO technology are becoming more and more complex. As more and more new devices have wireless connection, the need to improve spectrum efficiency is ultimately paramount. The advent of full-duplex wireless communications will instantly double spectral efficiency.
In the previous article, I briefly mentioned what .
Now we will get acquainted with the coordination of parameters between devices, as well as speed and operating mode ( full-duplex or half-duplex).
By default, each Cisco port is configured in such a way that the device itself determines what settings to use on this port, what speed to choose, what data transfer mode. This technology is called Auto-negotiation(Auto detection). You can also set these parameters “manually” on each port of the device.
Cisco automatically detects the speed between network devices (for example, between a switch port and network card computer) using some methods. Cisco switches used to determine speed Fast Link Pulse (FLP), this is some electrical impulse by which devices can understand which optimal speeds a connection can be established between these network devices.
If the speeds are set manually and they match, then the devices will be able to establish a connection using electrical signals.
If the speeds are manually set on the switch and on the network device of the computer (for example), and they do not match, then the connection will not be established.
Determining the connection operating mode is approximately the same: half-duplex or full-duplex.
If both devices operate in auto-detection mode, and the devices can operate in duplex mode, then this mode will be installed.
If auto-detection is turned off on devices, the mode will be assigned according to some “default” rules. For 10 and 100 megabit interfaces the half-duplex mode will be set, for 1000 megabit interfaces the Full-Duplex mode will be set.
To disable auto-duplex detection, you must manually specify the mode settings.
Ethernet devices can operate in Full-Duplex mode ( FDX), only when there are no collisions in the transmission medium.
Modern technologies say that collisions do not occur. Collisions occur only where there is a shared data transmission medium, for example, with a bus topology, or when using a device such as a hub (although now it is quite difficult to see such a “dinosaur” :)).
Still, it is necessary to imagine what technologies exist and how they deal with such shared resources.
The algorithm for dealing with collisions is called CSMA/CD (Carrier Sense Multiple Access Collision Detection), which means multiple access with carrier sensing and collision detection.
What is a collision anyway?
Collision this is signal superposition, that is, when several network devices begin transmitting data over a shared medium, these two signals meet, overlap each other, and a collision occurs (that is, the data is distorted and does not carry any payload.
Now let's look at how it works.
- The device that wishes to send first listens to see if the communication line is clear.
- When the link is idle, this device starts sending frames over Ethernet.
- The device “hears” that there is no collision, which means everything is fine.
- If a collision did occur (what about the first step? where did the device make sure that the line was not busy? The fact is that another device could also listen to the line, and these two devices sent frames at almost the same time, which is why a collision occurred ). Now, when the sending devices “realize” that a collision has occurred, they send a so-called jam signal, which “tells” other network participants that transmission is now impossible because a collision has occurred and they will have to wait a little.
- After the jam signal, each sending device is randomly assigned some time, which can be called “idle time,” when the device cannot send any data on the network.
- After this timer expires, the algorithm moves to step 1.
Switching technology itself has no direct bearing on the media access method used by switch ports. When you connect a segment that is a shared medium to a switch port, this port, like all other nodes on such a segment, must support half-duplex mode.
However, when not a segment, but only one computer is connected to each switch port, and over two physically separate channels, as happens in almost all Ethernet standards except coaxial versions of Ethernet, the situation becomes less clear. The port can operate in both normal half-duplex mode and full-duplex mode.
In half-duplex operation, the switch port still detects collisions. The collision domain in this case is a section of the network that includes a switch transmitter, a switch receiver, a computer network adapter transmitter, a computer network adapter receiver, and two twisted pairs, connecting transmitters to receivers. A collision occurs when the switch port and network adapter transmitters begin transmitting their frames at or near the same time.
In duplex mode, simultaneous data transmission by the switch port transmitter and the network adapter is not considered a collision. In principle, this is a fairly natural mode of operation for individual duplex data transmission channels, and it has always been used in protocols global networks. With full-duplex communication, 10 Mbps Ethernet ports can transmit data at a speed of 20 Mbps - 10 Mbps in each direction.
Already the first Kalpana switches supported both modes of operation of their ports, allowing the switches to be used to connect segments of a shared medium, as their bridge predecessors did, and at the same time allowing the data exchange rate on the ports intended for communication between the switches to double due to the operation of these ports in duplex mode.
For a long time Ethernet switches coexisted in local networks with Ethernet hubs: the lower levels of the building network, such as networks of work groups and departments, were built on the hubs, and switches served to combine these segments into a common network.
Gradually, switches began to be used on the lower floors, displacing hubs, as the prices of switches were constantly falling and their performance was increasing (due to support not only Ethernet technology with a speed of 10 Mbit/s, but also all subsequent more high-speed versions this technology, that is Fast Ethernet at a speed of 100 Mbit/s, Gigabit Ethernet at 1 Gbps and 10G Ethernet at 10 Gbps). This process ended with the displacement Ethernet hubs and the transition to fully switched networks, an example of such a network is shown in Fig. 1
Rice. 1 Fully switched Ethernet network.
In a fully switched Ethernet network, all ports operate in full duplex mode, and frame forwarding is based on MAC addresses. With the development of Fast Ethernet and Gigabit Ethernet technologies, duplex mode became one of the two full-fledged standard modes of operation of network nodes. However, the practice of using the first switches with Gigabit Ethernet ports has shown that they are almost always used in duplex mode to interact with other switches or high-speed network adapters. Therefore, when developing the 10G Ethernet standard, its developers did not create a version to operate in half-duplex mode, finally cementing the departure of the shared medium from Ethernet technology.
Answer:
IN Everyday life we communicate with each other in duplex mode, i.e. we can speak and hear the other person at the same time. Wood grouse, for example, during the current, singing a mating song, do not hear anything around, i.e., speaking scientific language, communicate in simplex mode (take turns exchanging information with each other). IN technically An intermediate option is possible, the so-called dual-frequency simplex, or half-duplex, but from the point of view of the end user it is equivalent to a simplex.
Thus, duplex is more familiar and natural for communication. Regular telephone communications, including in cellular networks, is carried out in duplex mode. However, duplex is not without its disadvantages. The simplex mode, despite some inconveniences during radio communication, has a number of technical advantages.
The simplex quite simply implements one of the main radio exchange modes in PMR networks - group calling and its various variations. In modern duplex networks, it is possible to organize so-called conference calls, but for operational communications it is of little use, since turning on the mode requires a certain time.
Duplex mode is less economical. This is because, in order to maintain the radio channel in both directions, the transmitter mobile station operates continuously, while conversation usually occurs in the form of dialogue or monologue, so on average 50% of the signal transmission time in one direction is not required, and the energy of the power supply is not optimally consumed. In simplex radios, the energy from the power source is used more efficiently.
In conditions of unstable communication, duplex is less reliable, since it requires maintaining a reliable communication channel in both directions.
In technical terms, the implementation of the duplex mode is much more complicated, since it requires the use of additional technical solutions to ensure simultaneous operation of the receiver and transmitter, which is why duplex radios are usually more expensive than simplex ones.
When organizing a communication network whose radios operate in simplex mode, as a rule, significantly fewer communication channels are required. Thereby simplex mode helps save radio frequency spectrum resources.
It should be noted that in some cases the decisive factor in completing the task may be the ability to transmit a message from the dispatcher of a stationary radio station mobile subscribers, even if for some reason the reverse communication channel is impossible. In simplex mode this will not cause difficulties; in duplex this is impossible.
Many professional mobile radio networks allow the simultaneous use of subscriber radios in both full-duplex and simplex modes. In this case base station operates in duplex mode, and a simplex subscriber radio station operates in half-duplex, i.e., with a separation of reception and transmission frequencies and alternate activation of these modes. Considering the above, we can give the following general recommendations: for communication systems that have access to telephone network, the use of a duplex mode of operation of subscriber terminals may be advisable for operational radio communications - the best option is the simplex operating mode of the stations.
The IEEE 802.3-2012 standard defines two modes of operation of the MAC sublayer:
● half duplex (half-duple x) – uses the CSMA/CD method for node access to the shared medium. A node can only receive or transmit data at one time, subject to gaining access to the transmission medium;
● full duplex (full-duplex) – allows a pair of nodes having a point-to-point connection to simultaneously receive and transmit data. To do this, each node must be connected to a dedicated switch port.
Access method CSMA/CD
The basic idea of Ethernet was to use a bus topology based on coaxial cable. The cable was used as a shared transmission medium over which workstations connected to the network broadcast bidirectional (in all directions) transmission. Terminators (plugs) were installed at both ends of the cable.
Rice. 5.21 Ethernet network
Since a common transmission medium was used, control over nodes' access to the physical medium was required. To organize access of nodes to the shared transmission medium, it was used multiple access method with carrier sense and collision detection(Carrier Sense Multiple Access With Collision Detection, CSMA/CD).
The CSMA/CD method is based on competition(contention) of nodes for the right to access the network and includes the following procedures:
● carrier control;
● collision detection.
Before transmitting, the network device must ensure that the transmission medium is clear. This is achieved by listening to the carrier. If the medium is free, the device begins to transmit data. During frame transmission, the device continues to listen to the transmission medium. This is done to ensure that no other device has started transmitting data at the same time. After the end of frame transmission, all network devices must withstand a technological pause (Inter Packet Gap) equal to 9.6 μs. This pause is called interframe interval and is needed to restore network adapters to their original state and to prevent exclusive seizure of the environment by one network device. After the end of the technological pause, devices have the right to begin transmitting their frames, because Wednesday is free.
Network devices can begin transmitting data any time they determine that the channel is free. If a device tries to start transmitting a frame but finds that the network is busy, it is forced to wait until the sending node finishes transmitting.
Rice. 5.22 Frame transmission on an Ethernet network
Ethernet is a broadcast medium, so all stations receive all frames transmitted over the network. However, not all devices will process these frames. Only the device whose MAC address matches the destination MAC address specified in the frame header copies the contents of the frame to the internal buffer. The device then checks the frame for errors, and if there are none, it transmits the received data to the higher-lying protocol. Otherwise, the frame will be discarded. The sending device is not notified whether the frame was successfully delivered or not.
In Ethernet networks, conflicts are inevitable ( collisions), because the possibility of their occurrence is inherent in the CSMA/CD algorithm itself. This is because there is some time between the time of transmission, when the network device checks whether the network is free, and the moment the actual transmission begins. It is possible that some other device on the network will begin transmitting during this time.
If multiple devices on a network started transmitting at approximately the same time, the bit streams coming from different devices, collide with each other and are distorted, i.e. a collision occurs. In this case, each of the transmitting devices must be able to detect a collision before it finishes transmitting its frame. Having detected a collision, the device stops transmitting the frame and reinforces the collision by sending a special sequence of 32 bits to the network, called jam-consistency. This is done so that all network devices can recognize the collision. After all devices have recognized the collision, each device is turned off for a certain randomly selected time interval (different for each network station). When the time has expired, the device can begin transmitting data again. When transmission resumes, the devices involved in the collision do not have priority for data transmission over other devices on the network.
If 16 attempts to transmit a frame cause a collision, then the transmitter must stop trying and discard the frame.
Rice. 5.23 Ethernet Collision Detection
Collision domain
In half-duplex Ethernet technology, regardless of the physical layer standard, there is a concept collision domain.
Collision domain(collision domain) is a part of the Ethernet network, all nodes of which recognize a collision regardless of in which part of the network it occurs.
An Ethernet network built on repeaters and hubs forms one collision domain.
Recall that a repeater was a physical layer device of the OSI model used to connect segments of a data transmission medium in order to increase the overall length of the network.
Ethernet networks (10BASE2 and 10BASE5 specifications) based on coaxial cable used two-port repeaters connecting two physical segments. The repeater worked as follows: it received signals from one network segment, amplified them, restored synchronization and transmitted them to another. Repeaters did not perform complex filtering and other traffic processing, because were not smart devices. Also, the total number of repeaters and the segments they connected was limited due to time delays and other reasons.
Later, multiport repeaters appeared, to which workstations were connected with a separate cable. Such multiport repeaters are called “hubs”. The reason for the appearance of multiport repeaters was as follows. Since the original Ethernet technology used coaxial cable and bus topology, it was difficult to install the cabling system of the building. Later, the international standard for structured building cabling specified the use of a star topology, in which all devices were connected to a single concentration point using twisted pair cabling. The technology perfectly suited these requirements Token Ring and therefore, to survive the competition, Ethernet technology had to adapt to new requirements. This is how the 10BASE-T Ethernet specification emerged, which used twisted-pair cables and a star topology as the transmission medium.
The concentrators worked at physical level OSI models. They repeated signals received from one of the ports to all other active ports, pre-restoring them, and did not perform any traffic filtering or other data processing. Therefore, the logical topology of networks built using hubs has always remained a bus.
At one point in time, in networks built on repeaters and hubs, only one node could transmit data. In the case of simultaneous arrival of signals into a common transmission medium, collision, which led to damage to transmitted frames. Thus, all devices connected to such networks were in the same collision domain.
Rice. 5.24 Collision domain
As the number of network segments and computers in them increased, the number of collisions increased and network throughput decreased. In addition, the segment's bandwidth was divided among all devices connected to it. For example, when ten workstations were connected to a 10 Mbps segment, each device could transmit at an average speed of no more than 1 Mbps. The task arose network segmentation, i.e. dividing users into groups (segments) according to their physical location, in order to reduce the number of clients competing for bandwidth.
Switched Ethernet
The problem of network segmentation and increasing its performance was solved using a device called bridge(bridge). The bridge was developed by Digital Equipment Corporation (DEC) engineer Radia Perlman in the early 1980s and was an OSI data link layer device designed to connect network segments. The bridge was invented a little later than routers, but since it was cheaper and transparent to the protocols network layer(worked for link level), it became widely used in local networks. Bridge connections ( bridging) are a fundamental part of the IEEE local area network standards.
The bridge worked according to an algorithm transparent bridge(transparent bridge), which is defined by the IEEE 802.1D standard. Before sending frames from one segment to another, it analyzed them and transmitted only if such transmission was really necessary, that is, the MAC address workstation destination belonged to a different segment. In this way, the bridge isolated the traffic of one segment from the traffic of another and divided one large collision domain into several small ones, which increased the overall performance of the network. However, the bridge transmitted broadcast frames (for example, necessary for the operation of ARP protocol) from one segment to another, so all network devices were in one broadcast domain (Broadcast domain).
The transparent bridge algorithm will be discussed in more detail in Chapter 6.
Switched Ethernet(Ethernet switched network) – an Ethernet network whose segments are connected by bridges or switches
Rice. 5.25 Connecting two network segments using a bridge
Since bridges were usually two-port devices, their effectiveness remained only as long as the number of workstations in the segment remained relatively small. As soon as it increased, congestion occurred in the networks, which led to the loss of data packets.
An increase in the number of devices connected in networks, an increase in the power of workstation processors, and the emergence of multimedia applications and client-server applications required more bandwidth. In response to these growing demands, Kalpana launched the first switch (switch), called EtherSwitch.
The switch is a multiport bridge and also operates at the data link layer of the OSI model. The main difference between a switch and a bridge is that it is more productive, can simultaneously establish several connections between different pairs of ports and supports advanced functionality.
Rice. 5.26 The local network built on switches
In 1993, Kalpana introduced full duplex Ethernet technology(Full Duplex Ethernet Switch, FDES) to your switches. After some time, during development Fast technologies Ethernet full duplex operation has become part of the IEEE 802.3 standard.
Operation in full duplex mode provides the ability to simultaneously receive and transmit information, because Only two devices are connected to the transmission medium. Reception and transmission are carried out over two different physical point-to-point channels. For example, over different pairs of twisted pair cable or different fibers of an optical cable.
This eliminates the occurrence of collisions in the transmission medium (the CSMA/CD method is no longer required, since there is no contention for access to the transmission medium), increases the time available for data transmission, and doubles the useful bandwidth of the channel. Each channel provides full speed transmission. For example, for the 10BASE-T specification, each link transmits data at 10 Mbps. For the 100BASE-TX specification - at a speed of 100 Mbit/s. At the ends of a duplex connection, the connection speed doubles because Data can be transmitted and received simultaneously. For example, in the 1000BASE-T specification, in which data is transmitted over channels at a speed of 1000 Mbit/s, the total throughput will be equal to 2000 Mbit/s.
Rice. 5.27 Data transmission in full duplex mode
Also, thanks to the full duplex mode, the limitation on the total length of the network and the number of devices in it has disappeared. The only thing left is the limitation on the length of cables connecting adjacent devices.
Operation in full duplex mode is only possible when connecting network devices whose ports support it. If a segment representing a shared medium is connected to a device port, the port will operate in half-duplex mode and recognize collisions. The ports of modern network devices support the function of auto-detection of half-duplex or full-duplex operating modes.
When the port is operating in full duplex mode, the sending interval between successive frames should not be less than a technological pause equal to 9.6 μs. In order to prevent overflow of device receive buffers when operating in full duplex mode, it is necessary to use a frame flow control mechanism.
It should be noted that the 10, 40, and 100 Gigabit Ethernet specifications only support full-duplex operation. This is due to the fact that modern networks have become fully switched, and switches almost always use full duplex when interacting with other switches or high-speed network adapters.