Wi-Fi Network: Single-Channel Architecture vs Multi-Channel Architecture
The purpose of this article is to conduct a short, high-level comparison of the two main architectures of today, to which the 802.11 standards were interpreted. Based on these architectures, the absolute majority of Wi-Fi networks of corporate and carrier class are built and built.
We will use the following common terms:
- Single Channel Architecture: Single Channel Architecture (SCA)
- Multichannel Architecture: Multi Channel Architecture (MCA)
In simple words, we can say that the single-channel architecture uses one frequency channel at all access points in the coverage area, while the multi-channel architecture uses a cellular structure with different non-overlapping frequency channels. When designing an MCA network, it is often necessary to avoid adjacent placement of cells with the same or adjacent frequency channels, the same applies to the vertical placement of cells in a three-dimensional model (on different floors).
')
Multichannel architecture has a lot of obvious advantages. The main vendors of Wi-Fi solutions at the corporate level use this approach and most of the existing Wi-Fi networks are built on this architecture. Statistics show that ISA significantly prevails as the basis for most Wi-Fi networks in the world. Much has been written about the MCA networks, but for those who want to quickly brush up on the basic principles of planning and deploying multi-channel architecture, you can refer here .
The main players in the area of single-channel architecture are companies Meru and Extricom, between which there are quite a few disagreements regarding patent primacy. Similarly, larger players, such as Aruba, also did similar research. Nevertheless, this technology is interesting and has its advantages and disadvantages, which we will discuss in more detail.
So, MCA uses different frequency channels at geographically separated access points to minimize the effects of interference and associated client device problems with decoding signals from different access points. The radio engineers manually (and / or the WLAN solution itself automatically) control the size of the cells, the frequency plan, the levels of interference in their network. This happens by changing the radiated power of access points, using different antennas and using a static, dynamic or semi-automatic method of changing frequency channels at specific points.
Once high-quality coverage has been achieved, the next most important task of the Wi-Fi network is to provide mobility. In the case of MCA, the client always chooses which access point to associate with, and the algorithms of mobile device manufacturers are different (in general, the available signal levels from access points on different channels, frame loss levels, etc., are taken into account, but how does the - this is the secret sauce of manufacturers and this sauce is not always tasty, remember how iPhones “stuck” on access points before recent releases of iOS). Therefore, we often observe completely different results when roaming different smartphones on the same Wi-Fi network. Hence, there are much more serious requirements for designing WLAN and conducting radio surveys for roaming tasks.
In contrast to the MCA network of a single-channel SCA architecture, in general, they use access points tuned to one frequency channel. Like in MCA, SCA uses network cell overlap in scheduling. Given the practical impossibility of performing territorial separation of transmitters on a single frequency, in order to avoid interference, the main approach in SCA is time transmission diversity. The entire network must be synchronized very precisely and coordinated from the central device in order to avoid sending frames at the same time on the same frequency. This is one of the main disadvantages of the SCA network, since The growth in the number of clients on the network who want to actively send or receive traffic quickly reduces the overall network performance. Especially if there is a mix of client devices, including outdated and slow ones. At the same time, the usual MCA network quite easily allows traffic to be transmitted in parallel and independently in each cell (it is assumed that we designed the network with cells on non-overlapping channels), fully realizing the initially asymmetric nature of the 802.11 standards.
In SCA networks, the client device cannot distinguish between single-channel access points. In this case, the Wi-Fi network (controller) decides when each specific access point should transmit and receive data from each specific client. Thus, the client device itself is not involved in the process of making a decision on the handover, for it the Wi-Fi network performs it. The user moves to the coverage area, and the network sends traffic to this user through the access point closest to him, taking into account the availability of available resources on the access points. In fact, here the main advantage of SCA networks is manifested - providing a handover without the need to involve client equipment and software, which makes the solution to the problem of practical mobility much more predictable compared to MCA networks. From this, we can conclude that SCA networks are well adaptable to tasks like VoIP over Wi-Fi, where it is important to provide predictable and very hard parameters for transferring the session without breaking from the access point to the access point.
SCA networks use a modified IEEE 802.11 model that shapes the vision of client devices so that they “think” that they always interact with only one access point. In reality, there can be many access points in such a network. Such a client "vision of the world" is achieved due to the fact that all points transmit in Beacon identical values of their BSSID and MAC. Such a model has become possible due to the "non-analog" nature of the 802.11 standards. All information is transmitted using frames. Because of this, with very precise control of the transmission and reception of frames, problems of interference between access points operating on the same frequency channel can be avoided. The control task usually lies on the central brain of a WLAN - Wi-Fi network controller. In some cases, this task is performed by LAN switches with integrated WLAN controller functionality.
The problem of lack of frequency channels in SCA has led to the emergence of solutions such as Channel Blanket, where more than one non-intersecting frequency channel can be used in parallel at access points (extension of the SCA model). This forms layers, where each is represented by one frequency channel (blanket / blanket) at all access points. But each of these parallel layers will have the same problems with capacity, and plus to this, problems of inter-channel interference will begin to grow sharply due to the minimum separation of emitters and receivers, which automatically leads to an increase in frame losses and an additional decrease in performance. And even sophisticated techniques of temporal coordination of sending frames, taking into account all the available layers, will quickly approach the limit of their efficiency with an increase in the number of users. Of course, the more layers there are, the greater the overall network capacity, but, obviously, the increase will be far from linear with the addition of each new layer. Moreover, there are real limitations, expressed in the fact that in 2.4GHz there are only 3 non-overlapping channels, which does not allow using more than 3 layers effectively in this band. The transition to 5GHz is unlikely to become an output in the foreseeable future, given that most mobile devices, even with 11n, do not yet support 5GHz. At the same time, in the MCA network, we simply add additional cells and decrease the radius of existing cells to quickly increase the capacity both on the entire network and in any part of it, without significantly affecting the entire infrastructure.
Interference affects the SCA network much more than the MCA network due to the use of one frequency channel throughout the coverage area. The 2.4GHz spectral band is the most suitable for Wi-Fi, but it is also the “dirtiest” in terms of the presence of interference sources . In MCA, a frequency plan can be dynamically and quickly rebuilt in the entire network or in any local part of it, but in SCA any significant interference will immediately affect the entire coverage area.
Thus, in tasks with the main focus on providing real mobility with support for a media session, first of all, such applications as VoIP in Wi-Fi networks or streaming video in motion (but in this case with small requirements for capacity for video), it can be expected SCA-based networks can be a good choice. Although if you think about it, how often do we build networks specifically for voice? At the same time, if it is required to ensure a high network capacity with a high density of users, with serious requirements in the lane, then it is worth first of all thinking about a solution based on MCA. Perhaps you should stay at the MCA in the case when you intend to use the Wi-Fi network for different tasks simultaneously.
We will use the following common terms:
- Single Channel Architecture: Single Channel Architecture (SCA)
- Multichannel Architecture: Multi Channel Architecture (MCA)
In simple words, we can say that the single-channel architecture uses one frequency channel at all access points in the coverage area, while the multi-channel architecture uses a cellular structure with different non-overlapping frequency channels. When designing an MCA network, it is often necessary to avoid adjacent placement of cells with the same or adjacent frequency channels, the same applies to the vertical placement of cells in a three-dimensional model (on different floors).
')
Multichannel architecture has a lot of obvious advantages. The main vendors of Wi-Fi solutions at the corporate level use this approach and most of the existing Wi-Fi networks are built on this architecture. Statistics show that ISA significantly prevails as the basis for most Wi-Fi networks in the world. Much has been written about the MCA networks, but for those who want to quickly brush up on the basic principles of planning and deploying multi-channel architecture, you can refer here .
The main players in the area of single-channel architecture are companies Meru and Extricom, between which there are quite a few disagreements regarding patent primacy. Similarly, larger players, such as Aruba, also did similar research. Nevertheless, this technology is interesting and has its advantages and disadvantages, which we will discuss in more detail.
So, MCA uses different frequency channels at geographically separated access points to minimize the effects of interference and associated client device problems with decoding signals from different access points. The radio engineers manually (and / or the WLAN solution itself automatically) control the size of the cells, the frequency plan, the levels of interference in their network. This happens by changing the radiated power of access points, using different antennas and using a static, dynamic or semi-automatic method of changing frequency channels at specific points.
Once high-quality coverage has been achieved, the next most important task of the Wi-Fi network is to provide mobility. In the case of MCA, the client always chooses which access point to associate with, and the algorithms of mobile device manufacturers are different (in general, the available signal levels from access points on different channels, frame loss levels, etc., are taken into account, but how does the - this is the secret sauce of manufacturers and this sauce is not always tasty, remember how iPhones “stuck” on access points before recent releases of iOS). Therefore, we often observe completely different results when roaming different smartphones on the same Wi-Fi network. Hence, there are much more serious requirements for designing WLAN and conducting radio surveys for roaming tasks.
In contrast to the MCA network of a single-channel SCA architecture, in general, they use access points tuned to one frequency channel. Like in MCA, SCA uses network cell overlap in scheduling. Given the practical impossibility of performing territorial separation of transmitters on a single frequency, in order to avoid interference, the main approach in SCA is time transmission diversity. The entire network must be synchronized very precisely and coordinated from the central device in order to avoid sending frames at the same time on the same frequency. This is one of the main disadvantages of the SCA network, since The growth in the number of clients on the network who want to actively send or receive traffic quickly reduces the overall network performance. Especially if there is a mix of client devices, including outdated and slow ones. At the same time, the usual MCA network quite easily allows traffic to be transmitted in parallel and independently in each cell (it is assumed that we designed the network with cells on non-overlapping channels), fully realizing the initially asymmetric nature of the 802.11 standards.
In SCA networks, the client device cannot distinguish between single-channel access points. In this case, the Wi-Fi network (controller) decides when each specific access point should transmit and receive data from each specific client. Thus, the client device itself is not involved in the process of making a decision on the handover, for it the Wi-Fi network performs it. The user moves to the coverage area, and the network sends traffic to this user through the access point closest to him, taking into account the availability of available resources on the access points. In fact, here the main advantage of SCA networks is manifested - providing a handover without the need to involve client equipment and software, which makes the solution to the problem of practical mobility much more predictable compared to MCA networks. From this, we can conclude that SCA networks are well adaptable to tasks like VoIP over Wi-Fi, where it is important to provide predictable and very hard parameters for transferring the session without breaking from the access point to the access point.
SCA networks use a modified IEEE 802.11 model that shapes the vision of client devices so that they “think” that they always interact with only one access point. In reality, there can be many access points in such a network. Such a client "vision of the world" is achieved due to the fact that all points transmit in Beacon identical values of their BSSID and MAC. Such a model has become possible due to the "non-analog" nature of the 802.11 standards. All information is transmitted using frames. Because of this, with very precise control of the transmission and reception of frames, problems of interference between access points operating on the same frequency channel can be avoided. The control task usually lies on the central brain of a WLAN - Wi-Fi network controller. In some cases, this task is performed by LAN switches with integrated WLAN controller functionality.
The problem of lack of frequency channels in SCA has led to the emergence of solutions such as Channel Blanket, where more than one non-intersecting frequency channel can be used in parallel at access points (extension of the SCA model). This forms layers, where each is represented by one frequency channel (blanket / blanket) at all access points. But each of these parallel layers will have the same problems with capacity, and plus to this, problems of inter-channel interference will begin to grow sharply due to the minimum separation of emitters and receivers, which automatically leads to an increase in frame losses and an additional decrease in performance. And even sophisticated techniques of temporal coordination of sending frames, taking into account all the available layers, will quickly approach the limit of their efficiency with an increase in the number of users. Of course, the more layers there are, the greater the overall network capacity, but, obviously, the increase will be far from linear with the addition of each new layer. Moreover, there are real limitations, expressed in the fact that in 2.4GHz there are only 3 non-overlapping channels, which does not allow using more than 3 layers effectively in this band. The transition to 5GHz is unlikely to become an output in the foreseeable future, given that most mobile devices, even with 11n, do not yet support 5GHz. At the same time, in the MCA network, we simply add additional cells and decrease the radius of existing cells to quickly increase the capacity both on the entire network and in any part of it, without significantly affecting the entire infrastructure.
Interference affects the SCA network much more than the MCA network due to the use of one frequency channel throughout the coverage area. The 2.4GHz spectral band is the most suitable for Wi-Fi, but it is also the “dirtiest” in terms of the presence of interference sources . In MCA, a frequency plan can be dynamically and quickly rebuilt in the entire network or in any local part of it, but in SCA any significant interference will immediately affect the entire coverage area.
Thus, in tasks with the main focus on providing real mobility with support for a media session, first of all, such applications as VoIP in Wi-Fi networks or streaming video in motion (but in this case with small requirements for capacity for video), it can be expected SCA-based networks can be a good choice. Although if you think about it, how often do we build networks specifically for voice? At the same time, if it is required to ensure a high network capacity with a high density of users, with serious requirements in the lane, then it is worth first of all thinking about a solution based on MCA. Perhaps you should stay at the MCA in the case when you intend to use the Wi-Fi network for different tasks simultaneously.
Source: https://habr.com/ru/post/215299/
All Articles