Application Note
High-Density Wi-Fi
Application Note – High-Density Wi-Fi
Table of Contents
Background .............................................................................................................................. 3 Description................................................................................................................................ 3 Theory of Operation ................................................................................................................. 3 Application Examples ............................................................................................................ 11 Tips and Recommendations.................................................................................................. 17 2
Application Note – High-Density Wi-Fi
Background
One of the biggest challenges faced by 802.11 networks is dealing with high-density user
deployments. Wi-Fi was initially intended to provide LAN access for a moderate number of users.
The evolution and overwhelming success of this technology has brought 802.11 deployments to
environments that go well beyond a few users to a point where it can now be the primary access
to the LAN. It is not difficult to find Wi-Fi networks deployed across campuses, industrial areas,
and even municipalities offering a wide variety of services. In these networks, it is often seen that
the number of users connected to the network surpasses the initial design considerations and as
a result, performance no longer meets expectations.
Description
This Application Note presents Xirrus’ solution for providing high quality network access for high
user densities over a Wi-Fi infrastructure. Traditional 1 or 2 radio APs have proven inadequate in
handling such environments. Truly successful high-density deployment requires Wi-Fi equipment
that is designed with scalability and performance in mind.
There are a number of design elements in Xirrus Wi-Fi Arrays that make them uniquely powerful
for high-density networks. These innovations include, but are not limited to:
•
•
•
•
•
•
•
Multi-radio System (IEEE 802.11a, IEEE802.11b, IEEE 802.11g and 802.11n): Xirrus
Wi-Fi Arrays incorporate 4, 8, 12, or 16 radios into a single device. Each radio can be
assigned to a unique channel providing dedicated bandwidth.
Antenna Sectorization: Directional, high gain antennas in a sectored Array design
provide a key capability for channel re-use in confined environments.
Auto Cell Sizing: Automatic control of power and sensitivity per radio allow control of the
size and performance of the coverage area.
Station Load Balancing and Association Limits: Appropriate distribution of users
among radios is key for high-density without requiring modifications to the Wi-Fi client and
avoid radio overloading.
Traffic Shaping: Controlling user traffic prevents any one station from clogging the
network.
Broadcast /Multicast Control and Station Privacy: Broadcast/multicast traffic can
extract a large toll on any network, so minimizing its effect improves network performance.
Radio Monitoring: Spectrum Analysis is an important troubleshooting aid.
Theory of Operation
Among the many challenges found in high-density Wi-Fi environments, the one that can be most
difficult is channel reutilization. The best way to provide bandwidth to a high number of
simultaneous users is to leverage as much of the RF spectrum available to Wi-Fi as possible and
as many times as possible. This means a multi-radio with intelligent antenna design to use as
many separate channels as possible while avoiding co-channel interference.
The following sections explain how the Xirrus Wi-Fi Arrays can create high-density user networks.
3
Application Note – High-Density Wi-Fi
Multi-Radio System
Perhaps the most intuitive assumption is that in order to provide the best throughput to Wi-Fi
clients, the connection between the client and the access point (AP) must be established at the
maximum possible data rate (also known as link rate). In the case of 802.11a and 802.11g, the
data rates will be 54Mbps, and with 802.11b, it is 11Mbps. The new 802.11n standard takes data
rates up to 600Mbps.
The table below summarizes the current 802.11 technologies.
Band
Channels
802.11b/g
2.4GHz
1,2,3,4,5,6,7,8,9,
10,11,12,13,14
3
802.11a
802.11n
5 GHz
2.4 GHz, 5GHz
Same channels as
36,40,44,48,52,56
802.11g and
60,64,100,104,108,
802.11a.
112,116,120,124
128,132,136,140
149,153,157,161,165
24
27 non bonding
13 with bonding
Number of nonoverlapping
channels
Channel Width
20MHz
20MHz
Max Data Rate
11Mbps / 54Mbps
54Mbps
20MHz and
40MHz
288/600Mbps
Table 1
The data rate is a function of the signal quality that is affected by distance and the noise levels
generated by nearby Wi-Fi or other interference sources. A high error rate will force clients and
APs to negotiate lower data rate connections even if the signal level is strong enough to support
higher data rates.
The actual throughput a user can achieve is a function of Free Air Time, which is the time the
media is available for the client to transmit or receive. The access to the media is controlled by
the CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance) algorithm. When the client
detects energy in the media due to other transmissions, certain levels of noise and interference or
adjacent channel emissions, the client must wait until the media becomes free.
If the media is only available half the time for a particular client, the maximum throughput that
client can transmit or receive will be half as well. This brings the concept of Available Capacity,
which is the product of the Free Air Time and the data rate between the AP and client.
The best method to increase the Available Capacity is to provide the highest data rate and
maximum Free Air Time.
The use of more spectrum and more radio channels increases airtime availability. When clients
are able to associate to multiple APs operating on separate non-overlapping channels, then
simultaneous transmission can occur, thereby increasing overall throughput and system capacity.
The Free Air Time increases proportionally to the number of channels that are being used. The
higher the number of non-overlapping channels used in a particular area, the more Available
Capacity for that area.
4
Application Note – High-Density Wi-Fi
Free Air Time is also related to the data rate of the links between APs and clients. The higher the
data rate, the less time the media will be used to transmit a given amount of information. A user
transmitting a 500byte packet at 1Mbps will be using the media (airtime) 54 times longer than a
user transmitting the same 500byte packet at 54Mbps.
The Free Air Time also depends on the number of clients in a channel. For a particular traffic
pattern and data rate per client, Free Air Time linearly decreases with the number of clients
present in a particular channel.
In summary, in order to increase the highest data rate and maximum Free Air Time, one should
simultaneously use multiple channels (radios) per system and reduce the number of users per
channel. The Xirrus Wi-Fi Array does this and goes even further with antenna sectorization
allowing for smaller coverage areas (small cells), guaranteeing higher data rates and channel reuse to provide the best combination for better overall Wi-Fi Available Capacity.
Antenna Sectorization
Appropriate planning of coverage areas is another critical task when deploying high-density
wireless networks. The coverage area correlates to the maximum distance clients can be from a
given AP with enough signal to associate and operate at a usable data rate. Data rates within the
cell will vary depending on the actual distance clients are from the AP.
In the previous section, it was mentioned how the use of multiple channels and radios allows a
greater number of stations in a given area and thus provides greater capacity. In high-density
environments, the number of stations per channel might be too large for the AP to provide
adequate throughput. In these cases, the Wi-Fi network must re-use the same channels several
times within a given area by creating smaller or sectorized coverage areas. This approach is
similar to the one cellular carriers use for mobile phones. By creating a larger number of smaller
cells, it is possible to achieve greater density, resulting in increased capacity. Because these cells
are smaller, the number of users per channel can be limited and provide additional re-use of
channels at much closer distance.
The choice of omni-directional or directional antenna design will also have a significant impact in
the re-use of channels. The use of sectorized directional antennas presents several advantages
over omni-directional antennas:
• Allows the use of several channels in the same AP system, minimizing co-channel
interference.
• Allows better-defined cells by concentrating energy in a sector.
• Limits the amount of interference received from other directions thus reducing packet
errors.
• Improves receive sensitivity in the direction of the cell since antenna gains work in both
directions.
• Reduces multipath issues because RF is not blindly transmitted in all directions.
• Helps in hidden node problems.
In Figure 1, there is an area where three non-overlapping channels are used with omni-directional
antennas. The black dots represent 30 wireless stations inside a cell. The closest available
channel for those stations will be channel 6. The distance required to reuse channel 6 again
within the coverage area is shown.
5
Application Note – High-Density Wi-Fi
1 1 1
6
6 6 11 1 1
11 1 6 11
6
11
6 11
6 11
1
6
11
1 6 11
11
1 1 6 Figure 1
11
1 1
6
1
6 11 6
1 6 11
11
Figure 2
Figure 2 shows the coverage provided using sectorized antennas using additional radios per
system. For the same station, there are now three channels covering the same area resulting in
greater distribution of stations – only 10 stations share a channel. By tripling the density of radios
per system and using sectorized antennas, the available capacity of the area has been increased
by a factor of three.
This scenario can be further improved by adding more radios per system. Since the number of
non-overlapping channels in the 2.4GHz band is only three, the recommendation is to add radios
in the 5GHz band. Figure 3 shows an example of a Xirrus Array sectorizing 6 channels in one
cell and Figure 4 shows 12 channels in one cell.
Figure 3
Figure 4
Auto Cell Sizing
The size of the cell or coverage areas is determined by the transmit power and receive sensitivity
of both the AP and the stations. By tuning those values, the cell size can be adjusted to
accommodate the dimensions and client density requirements. Adequate power control is also
important to mitigate the interference between radios operating in the same channel. Xirrus
Arrays have the option to either set the cell sizes manually or automatically via the Auto Cell
feature.
Auto Cell is an automatic, self-tuning mechanism that balances cell size between Arrays to
guarantee coverage while limiting the RF energy that could extend beyond the organizational
boundary. This is accomplished by setting radio power dynamically so that complete coverage is
6
Application Note – High-Density Wi-Fi
provided to all areas, yet at the minimum power level required to diminish potential interference
with neighboring networks. Additionally, Arrays running Auto Cell can detect and compensate for
coverage gaps caused by system interruptions.
Several Auto Cell parameters can be defined including minimum cell size, scheduled RF
assessment/adjustment, and the ability to define a coverage overlap percentage for roaming. An
overlap of 15% to 20% between cells is recommended for seamless roaming.
Without Auto Cell
Figure 5. Without Auto Cell
Figure 5 shows that when two radios are operating, stations B and C could potentially experience
interference from neighboring Arrays 1 and 2 resulting in corrupted packets. Strong co-channel
interference could cause stations in adjacent cells (cells 1 and 3) to defer communication while
stations in cell 2 are transmitting, resulting in the reduction of overall throughput. In the event of
Array 2 failure, stations A and D could lose all connectivity.
With Auto Cell
Client A
Client A
Array 2
Array 1
Array 2
Client C
Client B
Array 3
Array 1
Client D
Client C Array 3
Client B
Client D
Figure 6. Without Auto Cell
In Figure 6 above, Array radios balance power levels between themselves to guarantee client
coverage, but without the potential of interfering with other cells. In the event of Array 2 failure,
Arrays 1 and 3 automatically detect the loss of energy from Array 2 and raise their own radio
power levels to compensate for the loss. Stations who were on Array 2 then re-associate to the
other Arrays.
7
Application Note – High-Density Wi-Fi
In the case of manually configuring the cell sizes, the Xirrus Arrays have predetermined cell sizes
for different applications. In the case of high-density environments where channel re-use is
definitely required, medium or small cell sizes should be used.
Table 2 below shows the values of TX power and RX sensitivity for different cell sizes.
Cell Size
Transmit Power
(dBm)
5
12
19
20
AutoCell
1 to 20
Small
Medium
Large
Max (Default)
Auto
Manual
Receive Sensitivity
(dBm)
-75
-81
-87
-90
AutoCell
-95 dBM to -65 dBm
Table 2. Power and Receive Sensitivity settings on Array
Other combinations of values with more granularity can be manually configured to accommodate
other deployment requirements.
Station Load Balancing
Load Balancing allows the Array to distribute stations among all available radios in an area with
the goal of providing maximum bandwidth to all stations. By moving a station from one
congested radio to a less congested radio, load balancing allows that station to have more
available bandwidth. Additionally, it also decongests the radio where the station was initially
connected allowing the remaining stations on that radio greater overall bandwidth. This is very
important as described in the first point of this section where the available capacity linearly
decreases with the number of stations on a channel.
The Xirrus Wi-Fi Array supports automatic Load Balancing to distribute Wi-Fi stations across
multiple radios. In 802.11, it is the station that decides to which radio it will associate. The Array
cannot actually force station association to a specific radio, however the Array can “encourage”
stations to associate in a more uniform fashion across all of the radios of the Array.
In high-density environments where the stations are uniformly distributed within the coverage
area, it would be expected that the load distribution among radios would be also uniform. This is
not always the case. Typically the station will connect to the AP that has the strongest signal
strength, but when the station moves to another location, it may remain connected to the previous
radio until the signal drops below a certain threshold. Stations that are constantly moving can
create inconsistent load distribution conditions.
Another case of inconsistent distribution occurs when many stations move into a concentrated
space such as meeting rooms, conferences or auditoriums. In this case, the strongest signal
level for those stations may come from one particular radio and all those users will associate to
that radio leaving other radios in the vicinity lightly loaded.
The Array decides if a particular radio is over utilized and should not allow any more associations.
This decision process is based on a load-balancing algorithm that takes 3 key factors into affect:
• Fewer stations on a radio is preferable
• The strength of the signal
8
Application Note – High-Density Wi-Fi
•
5GHz channels preferred over 2.4GHz channels
Station Limits
To Load Balancing stations, the Arrays have the option to limit the number of associations per
radio. When the number of associations reaches the configured limit, the radio sends the station
a message indicating that the radio has reached its limit, encouraging the stations to associate to
another radio.
Traffic Shaping
802.11 is a shared medium. In a high-density Wi-Fi environment, traffic shaping is recommended
to optimize or guarantee performance levels. It is commonly applied at the network edges to
control traffic entering the network. Traffic shaping causes additional delays by serving queues
more slowly than if traffic shaping was not applied. Xirrus has incorporated different traffic
shaping methods such as:
• Traffic Classification: Support for WMM /IEE802.11e (Quality of Service)
• Rate Limiting: To control the maximum rate at which traffic is sent
Traffic classification can be configured on Xirrus Wi-Fi Arrays to provide traffic prioritization for
delay sensitive applications such as voice and video.
Rate limiting can be configured on the Array per SSID or User Group in the following ways:
• Traffic overall
• Traffic per station
Broadcast and Multicast Control
Unnecessary broadcast and multicast packets are a type of traffic that is undesirable in Wi-Fi
networks. The reason broadcast and multicast traffic packets are so detrimental is because
those packets are sent at the lowest basic data rates. Because these packets are intended to
reach multiple stations that might be located at different distances, only the lowest data rate can
guarantee reception to all of them. As mentioned, low data rate transmissions require more time
to send the same amount of information.
To optimize performance in high-density environments, it is important to minimize the amount of
unnecessary broadcast and multicast traffic. Both types of traffic are typically required and
cannot be eliminated, but the amount of this type of traffic transmitted over the air needs to be
controlled.
Xirrus Wi-Fi Arrays incorporate several features to control broadcast/multicast traffic in the air:
• ARP Filtering: Address Resolution Protocol finds the MAC address of a device with a
given IP address by sending out a broadcast message requesting this information. ARP
filtering allows the proliferation of ARP messages to be reduced by restricting how they
are forwarded across the network.
The following are options for handling ARP requests:
o Off: ARP filtering is disabled and requests are broadcasted to stations.
This is the default value.
o Pass-thru: The Array forwards the ARP request. It passes along only ARP
messages that are associated to the target the stations.
o Proxy: The Array replies on behalf of the stations to which it is associated.
The ARP request is not broadcasted to the stations.
9
Application Note – High-Density Wi-Fi
•
Broadcast/Multicast Optimization:
o Broadcast/Multicast Optimization: Restricts all broadcast/multicast
packets to only those radios that need to forward them. For instance, if a
broadcast comes in from VLAN 10, and there are no VLAN 10 users on a
radio, then that radio will not send out that broadcast. This increases
available airtime for other traffic.
o Rate Optimization: Changes the rates of broadcast traffic sent by the
Array (including beacons). When set to Optimized, each broadcast or
multicast packet that is transmitted on each radio is sent at the lowest
transmit rate used by any station associated to that radio at that time. This
results in each IAP broadcasting at the highest Array TX data rate that can
be heard by all associated stations, thus improving system performance.
The rate is determined dynamically to ensure the best broadcast/multicast
performance possible. The benefit is dramatic; consider a properly
designed network (one that has -70db or better everywhere), where virtually
every station should have a 54Mbps connection. In this case, broadcasts
and multicasts will all go out at 54Mbps vs. the standard rate. This means
that with the broadcast rate optimization on, broadcasts and multicasts use
between 2% and 10% of the bandwidth that they would in Standard mode.
When set to standard (default), broadcasts are sent out at the lowest basic
rate only—6 Mbps for 5GHz stations, or 1 Mbps for 2.4GHz stations.
In addition to those features, traffic filters can be created to limit or block any other type of traffic
that is not necessary or is slowing down the air portion of the network.
Station-to-Station Blocking
Station-to-station blocking prevents stations connected to the same Array from sending traffic
directly to each other. As a result, only traffic to/from the wired network is allowed.
This feature provides multiple benefits. First, it improves security since it prevents users from
accessing other devices on the network without the need to implement firewalling or filtering.
Second, it helps optimize broadcast and multicast traffic in the air. Broadcast traffic sent by a
station will not be sent back to the air by the other radios in that Array. Since broadcasts are sent
at the lowest data rates, this helps increase the overall available air time.
Station-to-station blocking may also be enabled in the network switches where the Arrays are
connected; this will prevent stations from communicating with each other when associated to
different Arrays.
Radio Monitoring
Spectrum Analyzers are used for troubleshooting Wi-Fi networks as well as for spectrum
analysis. There are numerous devices including microwave ovens, cordless phones, and
Bluetooth devices that can cause RF interference and degrade the performance of an 802.11 WiFi network. Often the sources of interference are signals from nearby Wi-Fi networks and can be
located with the use of a spectrum analyzer.
There is an integrated Spectrum Analyzer function (Figure 8) in every Xirrus Wi-Fi Array that
allows network administrators to monitor and troubleshoot their wireless networks in both 2.4GHz
and 5GHz spectrum in a distributed manner. They are mostly used to detect existing sources of
interference.
10
Application Note – High-Density Wi-Fi
Figure 8. Radio Monitoring
Using the Spectrum Analyzer on the Array provides users with the following real-time statistics in
each and every channel in the 2.4GHz and 5 GHz band.
• Packets/ sec: Total number of packet per second
• Bytes/sec: Total number of Bytes per second
• 802.11 Busy: % of time that 802.11 traffic is seen on that channel
• Other Busy: % of time that non-802.11 traffic is seen on that channel
• Signal to Noise: Average Signal to Noise Ratio
• Noise Floor: Average noise floor seen on that channel
• Error Rate: % of 802.11 packet with CRC errors
• Average RSSI: Average RSSI level seen on 802.11 packets
• Average Data Rate: Average Data Rate over time
Application Examples
This section shows two real-world examples of high-density environments that have been
deployed using Xirrus Wi-Fi Arrays. In these examples the configuration guidelines are outlined
for the important Xirrus features discussed previously. For additional details and how to configure
these features, please refer to the Xirrus Array User Guide as well as documents that are
available on the Xirrus website.
College Classroom/Auditorium
A large college classroom or auditorium is a typical case of a high-density environment. In such
applications, laptop-to-student ratio can reach 1:1 when all students are required to have a laptop
to access online or school resources during class.
In this example, the design goal was to provide wireless access to 140 simultaneous students in
a lecture hall. Only one Xirrus Wi-Fi Array, model XS8, was deployed. On this 8 radio Array, 3
were deployed at 2.4GHz, 4 at 5GHz, and one as a dual band monitor radio. It is important to
note that the monitor radio can also be used as a regular access radio in case additional capacity
is required.
11
Application Note – High-Density Wi-Fi
A ratio of 20 wireless stations per radio is used in this deployment, which represents a good
balance for bandwidth and coverage. Within the coverage area, the expected signal level is
sufficient to guarantee the highest possible data rates. The most important Array features for this
scenario are multi-radio, sectorization and load balancing.
Network Topology
The topology in this example is very simple. Just one Array was deployed and powered through
PoE and installed hanging from the ceiling in the center of the facility as shown in Figure 9.
Figure 9. Lecture hall
Configuration
• Channel Plan: The Array is configured for Auto Channel selection. After scanning the
area, the Array will select the most optimal channels in which to operate. Since three of
the 2.4GHz radios are enabled, the Array will pick channels 1, 6, and 11. For the 5GHz,
the Array will select four channels taking into consideration outside RF interference
conditions. The eighth radio is typically used in a threat sensor mode.
•
Cell Size: Although Auto Cell is not relevant when deploying only one Array, it should be
considered in case other Arrays are deployed within the vicinity, such as in other
classrooms.
The cell sizes of each radio in the Wi-Fi Array can be customized. In Figure 10 below,
each radio is set to a different cell size for illustration purposes. However, in most
implementations, all the radios are set to similar cell size. When configuring cell size
manually, the cell size can be set as small, medium, large, max, or manual.
12
Application Note – High-Density Wi-Fi
Figure 10. Auto Cell
Please refer to the AutoCell Application Note for additional configuration details.
Load Balancing
Having the stations homogenously distributed on the Array is very important to offer maximum
bandwidth.
There are three possible settings for Load Balancing:
• Off: Load balancing is disabled.
• On: The overloaded radio will ignore Probe Requests for a few seconds expecting that the
station will go to another radio. During that time, the radio will still respond to Association
and Authentication Requests (i.e. if the station already knows the radio is there). After a
few seconds if the station keeps sending Probe Requests, the radio will start responding
admit access.
• Aggressive: In this case, if a radio is overloaded it will ignore all Probe Requests,
Association Requests, and Authorization Requests from additional stations. It will do this
indefinitely until its load decreases. This way the Array makes sure the users have
associated to other radios less congested.
Please refer to the Station Load Balancing Application Note for configuration details.
Station Limits
If Load Balancing is not used, another possibility is to set a station limit. When a radio reaches
the established limit it will not accept connection from additional stations.
Traffic Shaping
The type of applications that will be accessed during class will determine the amount of allowed
traffic per user. Traffic shaping should be enabled to prevent students from utilizing more
bandwidth than necessary for the purpose of the class or to from transferring large amounts of
information between them.
The traffic limitation can be enabled on per SSID or per user group.
• Per SSID
o At SSIDs >SSID Management, select the SSID that rate limiting is to be
configured. User can choose to configure either one or both of the following:
- Uncheck “Unlimited” and enter a value for the overall traffic for the
selected SSID.
- Uncheck “Unlimited” and enter a value for maximum rate allow for
each station associated to the selected SSID.
13
Application Note – High-Density Wi-Fi
Figure 11. Traffic Shaping Per SSID
•
Per User Group
o At Groups >Group Management, select the Group that rate limiting is to be
configured. User can choose to configure either one or both of the following:
- Uncheck “Unlimited” and enter a value for the overall traffic for the
selected Group
- Uncheck “Unlimited” and enter a value for maximum rate allow for
each station in that Group.
Figure 12. Traffic Shaping Per User Group
Broadcast and Multicast Control
ARP filtering should be set to Pass-thru or Proxy.
•
To control of ARP traffic with Proxy and Pass-through mode
Figure 13. ARP Filtering
Station-to-Station Blocking
Go to IAPs > Global Settings to enable/disable intra-station traffic
Figure 14. Station-to-Station Blocking
This feature was not enabled at the above example.
Radio Monitoring
As mentioned before, the monitor radio can be also used as a regular access radio. In this
example, the number of radios available in the Arrays is sufficient to accommodate the number of
students, so the monitor radio is used for spectrum analysis and intrusion detection.
In order for the spectrum analyzer to function, there are 2 things that needs to be enabled
on the Array:
14
Application Note – High-Density Wi-Fi
•
•
The abg2 (XS series) or abgn2 (XN series) IAP must be configured as a monitor
using the omni-directional antenna.
Intrusion Detection Mode must be set to Standard.
Go to IAPs > IAP Settings, configure abg2/abgn2 as monitor
Figure 15. Radio Monitoring
At IAPs> Advanced RF Settings, select “Standard” mode for Intrusion Detection
Figure 16. Intrusion Detection
In RF Monitor >Spectrum Analyzer, an overview of the 802.11 and non-802.11 activity in both
2.4GHz and 5GHz channels can be monitored.
Convention Center
A convention center is a perfect example of a high user density wireless environment. In this
case, high-speed wireless Internet access is provided to 7000 attendees with approximately 3000
simultaneous users in a 150,000 square-foot hall. Various types of applications are supported
including web access, email, corporate VPN, text messaging, and VoIP stations connecting in
both the 2.4GHz and 5GHz bands.
Network Topology
The area is covered using twelve Xirrus Wi-Fi XS16 Arrays. The XS16 Array includes twelve
802.11a, three 802.11bg and one monitor spectrum/analyzer radio.
The physical distribution of Arrays within the area is mostly uniform with half of the Arrays
standing on tripods along the perimeter of the hall and the other half hanging from the ceiling.
In this configuration (Figure 17), a total of 180 radios are available – 36 operating in 2.4GHz
proving 802.11bg access and 144 operating in 5GHz providing 802.11a access.
15
Application Note – High-Density Wi-Fi
Figure 17. Array Deployment with Channel Allocation
All Arrays are wired to GigE switch ports and powered via PoE.
Configuration
• Channel Plan: In such an environment with a high density of radios, close proximity
between Arrays, and no considerable obstructions between them, the channels can be
easily assigned manually. Twelve 5GHz channels and three 2.4GHz channels are used
for a total of 15 per Array. For the 2.4GHz band, only three non-overlapping channels in
the US regulatory domain are used: 1, 6 and 11. In the 5 GHz band, if the UN-II band
channels are not considered, there are a total of 13 non-overlapping channels to allocate
to 12 radios. The channels are set manually, replicating the same channel plan on all the
Arrays and placing each Array in the same orientation (rotation).
•
Cell Size: Considering the distance between Arrays and a static and controlled
environment, the cells size are set to medium because the average distance from station
to Array is relatively short. Channel interference between Arrays will determine the cell
size rather than the size of the coverage area or required data rate.
•
Load Balancing: This feature is enabled in all the Arrays and set as normal. The
behavior of the users should be monitored at peak time and if the distribution does not
look uniform, Load Balancing may be set to aggressive. Consider that in this
environment, users are free to move and they could be creating areas of higher density
than others.
•
Traffic Shaping: In this particular deployment, traffic shaping was set to 2000 packets
per second per user. Other traffic limitations can be set depending on the network
requirements, for example applying filters to limit application support.
16
Application Note – High-Density Wi-Fi
•
Broadcast Control: ARP filtering is set to proxy and station-to-station blocking is
enabled. In this example, these two features are more critical since the amount of
stations in the same broadcast domain is much higher than in the previous example.
Tips and Recommendations
High user density environments can push Wi-Fi networks to their limits, largely based on the
contention created by many stations accessing a shared communication medium. The most
important capabilities of the Xirrus Wi-Fi solution to enable successful deployment in such
environments are its multi–radio/multi-channel design, antenna sectorization, load balancing, and
broadcast control. Beyond these facilities, the following recommendations may be used to
optimize these types of deployments:
•
Monitor the number of users per radio and radio utilization during peak times using the
XMS to understand the user distribution. Enable aggressive load balancing if necessary.
•
Monitor that the radios in all Arrays are functional and reporting information for future data
collection and statistics generation.
•
In cases where access for legacy 802.11b stations is not required, configure the Arrays for
802.11g only. This will eliminate the lowest data rates from the air and improve the overall
performance on those radios for 802.11g stations.
•
Analyze the type of traffic that is being sent by the stations using a packet sniffer. If
certain traffic types are seen that are not desired and can be eliminated (e.g. video
streaming, P2P, etc.), create filters in the Arrays to block them.
•
Implement station-to-station blocking on the switches to which the Arrays are connected
to prevent users on different Arrays from communicating with each other.
•
Monitor the backhaul and WAN utilization of the network to understand the traffic flow and
magnitude to ensure these components of the network are not becoming a bottleneck.
•
Use the XMS to monitor the traffic utilization of each Array to see if some are carrying
more traffic than others or have many more users associated to them. Use the results to
better distribute usage accordingly. For example, a lower user limit can be set in the
Arrays that carry more traffic.
•
Look at the traffic statistics and check the number of retransmissions. If a high
percentage of retransmissions is detected, there may be too much inter-Array
interference. In this case, lower the TX power of the Array radios
•
Check the Array IAP and station statistics to understand the number of users connected at
low data rates. A lot of traffic at low data rates may indicate users are roaming and not
selecting the closest radio. In this case, aggressive load balancing may be enabled and
additionally increase the sensitivity threshold.
17