ap deployment guide - spiceworks

AP Deployment Guide Best Practices Design Guide

©Ruckus Wireless 2

Best Practices Design Guide AP Deployment Guide

January 2017

Table of Contents Intended Audience....................................................................................................................................... 4 Overview ...................................................................................................................................................... 5 Capacity Planning ....................................................................................................................................... 6 Application Requirements ........................................................................................................................... 6 AP Type and Configuration ....................................................................................................................... 6 Airtime Availability ..................................................................................................................................... 8 Access Point Count (Estimation Method) ................................................................................................. 10 AP Placement Guidelines ......................................................................................................................... 13 AP Cell Sizing ............................................................................................................................................. 13 AP Placement Examples ........................................................................................................................... 14 WLAN Coverage and Physical AP Upgrades ...................................................................................... 16 AP Features and Configuration ............................................................................................................... 17 Power over Ethernet (PoE)........................................................................................................................ 17 Wireless Configuration Optimizations ................................................................................................... 17 Channel Allocation in FCC Regulatory Domain .................................................................................... 18 Channel Allocation in ETSI Regulatory Domain .................................................................................... 20 Channel and Power Options .................................................................................................................... 20 Load Optimizations ................................................................................................................................... 22 Interactions with Band Balancing: ........................................................................................................... 24 SSID Rate Limiting and Airtime Fairness (ATF) ...................................................................................... 24 Overhead Reduction for High Density Deployments ........................................................................... 26 Advanced Optimizations .......................................................................................................................... 28 Appendix A: Understanding Inter-client RF Interference .................................................................... 29 Introduction ................................................................................................................................................. 29 Investigations .............................................................................................................................................. 29 Close Distance Tests .................................................................................................................................. 30 The Impact of Channel Frequency Separation ..................................................................................... 31 Lessons Learned ......................................................................................................................................... 33 About Ruckus............................................................................................................................................... 34

©Ruckus Wireless 3


January 2017

Copyright Notice and Proprietary Information

Copyright 2017 Ruckus Wireless, Inc. All rights reserved.

No part of this documentation may be reproduced, transmitted, or translated, in any form or by any means, electronic, mechanical, manual, optical, or otherwise,

without prior written permission of Ruckus Wireless, Inc. (“Ruckus”), or as expressly provided by under license from Ruckus.

Destination Control Statement

Technical data contained in this publication may be subject to the export control laws of the United States of America. Disclosure to nationals of other countries

contrary to United States law is prohibited. It is the reader’s responsibility to determine the applicable regu lations and to comply with them.

Disclaimer

THIS DOCUMENTATION AND ALL INFORMATION CONTAINED HEREIN (“MATERIAL”) IS PROVIDED FOR GENERAL INFORMATION PURPOSES ONLY. RUCKUS

AND ITS LICENSORS MAKE NO WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, WITH REGARD TO THE MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE

IMPLIED WARRANTIES OF MERCHANTABILITY, NON-INFRINGEMENT AND FITNESS FOR A PARTICULAR PURPOSE, OR THAT THE MATERIAL IS ERROR-FREE,

ACCURATE OR RELIABLE. RUCKUS RESERVES THE RIGHT TO MAKE CHANGES OR UPDATES TO THE MATERIAL AT ANY TIME.

Limitation of Liability

IN NO EVENT, SHALL RUCKUS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL OR CONSEQUENTIAL DAMAGES, OR DAMAGES FOR LOSS OF

PROFITS, REVENUE, DATA OR USE, INCURRED BY YOU OR ANY THIRD PARTY, WHETHER IN AN ACTION IN CONTRACT OR TORT, ARISING FROM YOUR

ACCESS TO, OR USE OF, THE MATERIAL.

Trademarks

Ruckus Wireless is a trademark of Ruckus Wireless, Inc. in the United States and other countries. All other product or company names may be trademarks of their

respective owners.

©Ruckus Wireless 4


January 2017

Intended Audience

This document addresses factors and concerns related to wireless Access Point deployment. Many factors can affect both the initial design and final

performance. These are considered here along with recommendations for an optimal design.

This document is written for and intended for use by technical engineers with some background in Wi-Fi design and 802.11/wireless engineering principles.

©Ruckus Wireless 5


January 2017

Overview

Wireless LAN networks have become increasingly complex and sophisticated. The scale of typical deployments has changed dramatically and Enterprise or

Service provider WLAN deployments consisting of tens of thousands of wireless access points have become common. These networks must be able to provide

ubiquitous access for customers, employees, guests and contractors. Wireless network access rates have become comparable with wired networks and client

demands have risen to match.

To support these demands every WLAN deployment should be completed in accordance with recommended best practices. Many WLAN equipment vendors

have introduced features to facilitate network performance, improve optimization and provide higher reliability for client connections. These practices include

multiple areas where technical and planning decisions have to be made correctly. This document will provide some guidelines on topics such as network capacity

estimates, AP placement, and RF channel selection. It will also describe which product features and configuration settings should be used to achieve optimal

performance.

©Ruckus Wireless 6


January 2017

Capacity Planning

This chapter introduces general principles to estimate the required number of wireless access points for a deployment. There are different ways such planning

can be accomplished. One method is to use projections based on the network load generated by wireless endpoints (clients) and the applications they run.

Another is to estimate based on airtime. Several factors can affect these calculations regardless of which one you use. These include the application type, device

type, AP capabilities, interference, etc.

Application Requirements

Different applications have different requirements for bandwidth, jitter and latency. The following is an example of common applications and some values. These

should be verified, whenever possible, as part of the planning exercise.

TABLE 1 EXAMPLE OF CAPACITY REQUIREMENT CALCULATION

Client Type Application (SLA) # Associated Devices % Concurrently Active Min Bandwidth Required Tablet On-line Testing (100 Kbps) 100 50% 5 Mbps

Laptop Google Doc (500 Kbps) 200 50% 50 Mbps

Smartphone Web/E-mail (500 Kbps) 200 20% 20 Mbps

SmartTV Video Streaming (10 Mbps) 40 50% 200 Mbps

Total Capacity (xput) Needed 275 Mbps

AP Type and Configuration

There are several factors that can affect the capacity of an AP. Some important capacity-related limitations of an AP include:

• Which 802.11 standard(s) can an AP support? Typical options include – 802.11a/b/g/n/ac.

• For the 802.11n and 802.11ac APs, how many spatial streams (SS) can be supported? – 2/3/4

Factors that Affect Available AP Bandwidth Capacity

Higher capacity can be reached with wider channel sizes, but this limits the number of APs that can be installed close to each other. This is because there are only

so many available channels. You can have many smaller channels or fewer wide channels. The key point is channel reuse, which is when you have used all

available channels and new APs must reuse a channel occupied by another AP. When channel reuse occurs, any performance improvement may be diminished in

a real network due to noise, co-channel and adjacent channel interference.

Network deployments with multiple APs that use fewer non-overlapping channels with wider frequency bandwidth have a higher probability of significant levels

of co-channel interference. This should be taken into account and mitigated if possible. If the client applications do not require very high bandwidth levels,

consider using narrower channel sizes. This will increase the number of channels available and help reduce channel reuse and interference.

©Ruckus Wireless 7


January 2017

The following graphic gives an example of this tradeoff. For 5 GHz, there are a maximum of 24 non-overlapping 20 MHz-wide channels, plus channel 144

which can be used in some cases i.e. if AP and client devices support it. Theoretically, you could have a deployment with 24 APs and no overlap. As the channel

width increases, the number of non-overlapping channels goes down. For the 80 MHz-wide channels, which for the higher 802.11ac data rates, the number of

non-overlapping channels is drastically reduced from 24 to 5.

It is important to note that the number of channels available varies by country worldwide. Some available channels may have additional restrictions, such as DFS.

For more information on available channel widths and DFS, please see this Wikipedia page.

FIGURE 1 - 5 GHZ AVAILABLE CHANNEL USAGE BY CHANNEL WIDTH

Client Type

Client capacity is also important. An 802.11ac AP will not improve the performance of a legacy 802.11n client as much as it will for an 802.11ac device.

Important client features that can affect capacity include:

• Which 802.11 standard(s) do the clients support? – 802.11a/b/g/n/ac.

• How many spatial streams (SS) can the clients use? - 1/2/3.

Note: There are no generally available 4SS mobile clients at the time of writing.

https://en.wikipedia.org/wiki/List_of_WLAN_channels

©Ruckus Wireless 8


January 2017

The following chart gives an example of possible data rates by client device standard and spatial stream support.

Why is the client data rate important? Wi-Fi is a shared medium, the longer a client spends on air transmitting, the longer the other devices must wait before

they can transmit. If fewer devices are given an opportunity to transmit in any given timeslot, overall capacity will be reduced.

Let’s go through each of these step-by-step.

Airtime Availability

In order for an 802.11 device to communicate, it must be able to transmit, i.e. occupy airtime. We should be aware that 100% of all possible airtime is not

available for data transfer for multiple reasons.

The IEEE 802.11 protocol is designed to accommodate multiple wireless devices sharing the same media. It uses carrier sense multiple access with collision

avoidance scheme (CSMA/CA). To avoid collisions, each device has to listen first and check if medium is not busy and then use inter frame spacing interval(s) and

back-off processes prior to transmitting a frame(s).

To determine if the wireless media is busy, a 802.11 wireless device will utilize Clear Channel Assessment. If the device can detect and start decoding a 802.11

signal on a given channel, it will mark that media as busy and will defer its own transmission for short time. Other transmissions may come from other 802.11

devices or access points operating on the same channel. Logical reservation is also possible using the Network Allocation Vector (NAV timer) value in 802.11

packets. When multiple wireless access points operate on the same channel they can create Co-Channel Interference (CCI) but will follow the protocol to share

that channel with each other and with other 802.11 devices.

There are multiple channels available for Wi-Fi in both the 2.4 and 5GHz frequency bands. Wireless access points are generally deployed on different

channels to avoid CCI and to provide more available bandwidth. 802.11 radios transmit certain amount of energy outside the channel while still staying in

compliance with the transmit spectral mask mandated by regulations. Invariably, some energy is emitted to either side of the channel and this will produce

interference with devices on adjacent channel(s). This is called Adjacent Channel Interference (ACI). In this case other 802.11 devices will detect it as RF energy

but will not be able to demodulate that traffic. When the amount of energy exceeds the Energy Detect (ED) threshold, 802.11 devices will defer transmission,

see Figure 2 for details.

©Ruckus Wireless 9


January 2017

FIGURE 2 MEDIUM SHARING

Since the same frequency (channel) is used to transmit and receive 802.11 frames, airtime is shared between both the AP and all of the wireless clients. There is

also a considerable amount of airtime required for protocol overhead, which allows wireless networks to operate reliably and share the medium. Typical 802.11

traffic is made of a mixture of data, management and control frames. Wireless stations and APs also have a mandatory wait time between packets. These Inter

Frame Spacing (IFS) and back-off intervals help to avoid “in air” collisions that occur if multiple devices attempt to transmit at the same time.

To summarize:

• Wireless clients and APs must share the medium with other APs/clients within range on same channel.

• Approximately 70-90% of airtime is available depending on the band / DFS / environment.

• There is less spectrum (fewer channels) available in 2.4G than in 5 GHz.

• To fully utilize 5GHz bandwidth, DFS channels should be used if possible.

For QoS-sensitive applications it is desirable to keep total airtime utilization to <50%. This increases the chance that a wireless will be able to transmit quickly

and avoid latency and/or jitter.

©Ruckus Wireless 10


January 2017

Access Point Count (Estimation Method)

This simple step-by-step method will provide a starting point for determining the number of APs needed based on WLAN capacity requirements.

Step 1: Estimate the total capacity (throughput) needed

Step 2: Estimate the total capacity (throughput) available per AP

Step 3: Estimate the number of APs, including any reserve capacity needed

There are several factors, which will help determine the total capacity needed.

Start with the expected maximum number of associated clients. This number can be different from the number of WLAN users as people may carry more than

one device. Let’s use an example in which we have a total of 200 smartphones to support. Next, we need to determine the percentage of devices that are

accessing the wireless simultaneously. The number of devices concurrently active may vary from time to time, however network capacity planning should be

based on the peak load on the network. In our example, we have determined 20% of the devices might be accessing the Wi-Fi at the same time. Finally, we

need to consider which applications are expected to be running on the devices, especially if there are SLA requirements for any of them. In this example, our

smartphone clients will be doing simple web browsing and email, which has low bandwidth requirements. For this example, we are estimating that usage is 500

Kbps per client. For 200 smartphones with 20% concurrent usage for web browsing and email, we can estimate a total minimum bandwidth requirement of 20

Mbps at any given time.

Consider which applications are expected to be running on the devices, especially if there are SLA requirements for any of them.

Refer to Table 2 Example of capacity requirement calculation.

TABLE 2 EXAMPLE OF CAPACITY REQUIREMENT CALCULATION

Client Type Application (SLA) # Associated Devices % Concurrently Active Data Rate Required Tablet On-line Testing (100 Kbps) 100 50% 5 Mbps

Laptop Google Doc (500 Kbps) 200 50% 50 Mbps

Smartphone Web/E-mail (500 Kbps) 200 20% 20 Mbps

SmartTV Video Streaming (10 Mbps) 40 50% 200 Mbps

Total Capacity (xput) Needed 275 Mbps

Now that we have an estimate of what bandwidth the clients require, we need to determine how much of that a single AP can provide. With the total network

capacity now known we may proceed to the next planning step: how much capacity is available per AP?

AP Type and Configuration

Important capacity-related limitations of an AP include:

• Which 802.11 standard(s) can an AP support? Typical options include – 802.11a/b/g/n/ac.

• For the 802.11n and 802.11ac APs, how many spatial streams (SS) can be supported? – 2/3/4



January 2017

Available Bandwidth

Higher capacity can be reached with wider channel sizes, but this limits the number of APs that can be installed close to each other. This is because there are only

so many available channels. When channel reuse occurs, any performance improvement may be diminished in a real network due to noise, co-channel and

adjacent channel interference.

Network deployments with multiple APs that use overlapping channels with wider frequency bandwidth have a higher probability of significant levels of co-

channel interference. This should be considered and mitigated, if possible. If the client applications do not require very high bandwidth levels, consider using

narrower channel sizes. This will increase the number of channels available and help reduce channel reuse and interference.

Coverage and RF Planning

For a wireless network to operate within specifications and deliver reliable connectivity for client devices, it must provide a minimal RSSI level within the area

specified. This goes hand-in-hand with SLA requirements. Wireless coverage and RSSI level also have a direct impact on the PHY rates that 802.11 wireless

devices may use to connect to the AP. Generally, a higher RSSI will allow clients to connect with higher PHY rates. Some RSSI examples:

• -80 dBm RSSI is sufficient for basic connectivity at low PHY rates and is generally considered marginal coverage. It is not recommended as

sufficient for reliable communication

• -60 dBm RSSI or better is desirable for applications requiring Quality of Service (QoS) guarantees

• -68 dBm RSSI is the recommended cell boundary for data/voice communication

These RSSI values must be increased if the site has a high noise floor. A 25dB SNR is good target value to achieve.

The table below provides examples of what can be achieved with a 4 spatial stream AP at -65 dBm RSSI for clients of varying capabilities.

TABLE 3 DATA THROUGHPUT ACHIEVABLE AT -65 DBM RSSI

AP Client MCS PHY Rate TCP/Application Xput* 11ac, 4SS, 80 MHz 11ac, 3SS, 80 MHz 5 780 Mbps 540 Mbps

11ac, 2SS, 80 MHz 5 520 Mbps 360 Mbps

11ac, 1SS, 80 MHz 5 260 Mbps 180 Mbps

11n, 2SS, 40 MHz 6 270 Mbps 160 Mbps

11n, 1SS, 40 MHz 6 135 Mbps 80 Mbps



11a/g, 1SS, 20 MHz 54 Mbps 54 Mbps 25 Mbps

11b, 1SS, 20 MHz 11 Mbps 11 Mbps 5 Mbps

* TCP Throughput is 70%, 60%, and 50% of PHY Rate for 11ac, 11n, and 11a/b/g, respectively

The Modulation Coding Schemes (MCS) index is a good way to compare different performance values for wireless devices. The lower the MCS rate, the lower

the performance and PHY rates will be.



January 2017

Airtime-based Capacity

Using the information in this chapter, a simple AP load-based calculation can be performed. All input variables and final results of the calculation are

summarized below.

Values for “Max Available Throughput (Mbps)” are taken from the last column Error! Reference source not found.with the assumptions made there. These should

correspond to each client type and the average expected MCS rate clients for a given environment.

The value of “Airtime Required” is calculated as follows:

• Airtime Required (%) = ((“Throughput Need per Client” * “# Active Devices”)/ “Max Available Throughput”)*100%

An estimate for “Airtime available per AP” has to be reduced if this network experiences a lot of interference from Wi-Fi or non-Wi-Fi sources.

To calculate “APs needed to meet Airtime Requirement” = “Total Airtime Requirement”/”Airtime Available per AP” and round the result to the next integer

number.

TABLE 4 EXAMPLE OF A LOAD-BASED AP COUNT CALCULATION

Client Type Config Max Available Xput (Mbps) Xput Need per Client (Mbps)

# Active Devices (Band balancing @ 25%)

Airtime Required

2.4G 5G Total 2.4G 5G 2.4G 5G

Laptop 11ac, 2x2 85 360 0.1 50 12 38 2% 1%

Tablet 11ac, 2x2 85 360 0.5 100 25 75 15% 10%

Smartphone 11ac, 1x1 45 180 0.5 100 25 75 28% 21%

SmartTV 11n, 2x2 85 160 10 20 5 15 59% 94%

Total Airtime Requirement

104% 127%

Airtime Available per AP

70% 80%

APs needed to meet Airtime Requirement 2 2

Capacity for Future Expansion (33% Reserve Capacity Requirement 1 1

Total Number of APs Needed 3 3

Note: In this example 3 Access Points are considered for providing sufficient capacity for the total number of 270 clients. This equates to the distribution of 90

clients per one AP. It may be excessive for most common types of Enterprise WLAN deployments. Additionally, coverage criteria will still apply and more APs

may have to be added. These numbers can be used as only an example.



January 2017

AP Placement Guidelines

AP Cell Sizing

There are two types of AP deployment types generally used: coverage and high density. A coverage-based AP deployment emphasizes coverage of a

maximum area versus a maximum number of clients. The tradeoff is that as client-to-AP distance increases, the wireless devices may have to operate with lower

RSSI levels. Weaker RSSI levels will push MCS and PHY data rates down. Packets will be sent slower and the network performance overall may be reduced.

Another impact of such a design is that the larger coverage area will likely generate a higher client-per-AP ratio, unless this is a sparsely populated area. In

either case, all bandwidth is shared between these clients. Therefore, the number of devices per AP becomes even more critical in a coverage-based model. If

there are many client devices, you will have the worst of both worlds: low throughput rates combined with low signal strength.

A wireless LAN with many devices and many APs solves the previous problem. It reduces the number of clients per AP ratio (more APs than a coverage model)

and increases the signal strength (clients are closer to the APs). However, high density deployments have their own tradeoffs. One of the biggest considerations is

the negative impact of co-channel and adjacent-channel interference. Co-channel interference will occur if channels are reused at close distances and with very

little attenuation of the signal strength. The 802.11 protocol is designed to handle such a scenario, but the tradeoff is reduced performance since airtime will be

shared between these cells. To prevent two devices from transmitting at the same time, any Wi-Fi device that hears another device talking on its channel must

wait until the medium is available. This means that, effectively, two APs transmitting on the same channel will act like one AP. You will have two APs, but only one

Wi-Fi device can transmit at a time. All others will be forced to wait.

Adjacent Channel Interference (ACI) can also impact the performance of close-spaced (several feet) clients transmitting on adjacent channels. Even though the

channels are considered non-overlapping, some impact can still occur if the devices are close enough. Access Points mounted too closely and operating on

different channels on same band may also have a measurable performance penalty during concurrent operation. One way to visualize this is to think of a

conversation between two people (person A talking to person B). They are each using a megaphone to talk to the other. They can hear each other fine both at

close range or at a distance. But consider what happens if we add a third individual (person C) standing nearby to A and also using a megaphone to transmit to

a fourth (person D). Even though person C is speaking to someone else entirely, they are so close to person A that A at one cannot understand what it is hearing

from its client, B. Effectively, the APs are so loud they are making each other deaf when they transmit regardless that they are using “non-overlapping” channels.

Deployments that require smooth roaming between APs will need an overlap between cells. Generally, 20% coverage overlap between adjacent cells is

sufficient. Clients should always see at least two APs at an RSSI better than -80 dBm at any location for good roaming to occur. When we refer to cell overlap,

in this case, we are not referring APs on the same channel. Preferably, they are on different channels. They do need to be close enough that a client can hear

both, so the overlap is in physical distance rather than spectrum overlap. In cases where the APs overlap both in distance and channel (CCI), a target of 15-20dB

separation is recommended between each AP.

Deployments requiring location services are highly dependent on the need to have multiple APs hear a client at the same time. This allows the APs to gather

sufficient information to perform triangulation. From an AP deployment point of view, this means there is a requirement for a significant overlap of AP cells. In

this case, clients should see at least three APs at an RSSI better than -80 dBm at any location. Many applications recommend that these APs be on the same

channel as well to achieve highest accuracy.



January 2017

Factors That Affect AP Cell Size

Anything that affects how well the RF signal propagates will affect the cell size. These include:

• AP placement

• Obstructions/absorption of signal due to walls, windows, etc.

• Non-Wi-Fi interference

• Distance between client and AP

• AP transmit power

• Directional vs. omnidirectional antennas

AP Placement Examples

This section describes examples of deployment locations and how it may impact performance.

AP to Client Distance

For best coverage, install APs close to users with unobstructed line of sight. For example, when coverage must be provided inside a room, install the AP in the

room rather than in a hallway next to it.

Where possible, avoid mounting APs too high to minimize AP to client path loss due to greater distance. Free Space Path Loss is a common term used in RF

communications to calculate path loss between two antennas (Transmitter and Receiver) with non-obstructed line of sight between them. This loss increases with

distance. The greater the distance the higher the transmit power and/or antenna gain that may be required to support a reliable connection.

Non-Wi-Fi Interference

Keep APs away from non-WiFi interferers, like microwave ovens, cordless phones etc. that will add unnecessary interference and reduce performance.



January 2017

Directional vs. Omnidirectional Antennas

Where limiting cell size (for example many APs need to be close together and on the same channels) narrow beam or sectored antennas are helpful, especially

if there is no physical object or barrier attenuating the signal. Reducing the RF coverage via a directional antenna will help prevent too much cell overlap.

FIGURE 3 LOCATIONS WHERE USING SECTOR ANTENNAS MAY BE BENEFICIAL

AP Installation Obstructions

Installing access points or antennas next to metal objects can cause attenuation and/or reflections which change how RF signals propagate from an AP. Metal

obstruction may create significant signal path loss or block radio waves completely. Additionally, an environment that has many metal surfaces or walls, will be

very reflective for the radio waves, causing a lot of multi-path.

See an example of what to avoid on Figure 4

FIGURE 4 AVOID AP INSTALLATION NEXT TO METAL OBJECTS

RF does not propagate well through sheet metal or a fine metal mesh screen. In the example pictured above, signals will be reflected or attenuated

considerably. This means the signal coverage may not be what is required or planned for and cause performance issues where clients are either unable to hear

the signal or the signal is so weak they cannot maintain a reliable connection. Metal beams, pipes, pillars, doors, conduits will also impact performance. Signal

propagation characteristics can vary with time of day and circumstances as well, e.g. path loss could be lower when the venue is empty compared to when it is

occupied.

In very high-density designs, where a large number of APs must be very close together, obstruction and reduction of RF propagation can be desirable. For

example, it is better to confine the “cell size” (coverage) through AP placement i.e. place AP inside the room rather than in hallway. In multi-story buildings,

consider the impact of RF leakage through the floor. Where inter-floor isolation is limited, channel assignment has to be coordinated in three dimensions.



January 2017

WLAN Coverage and Physical AP Upgrades

Changes That Can Affect In-place 1:1 AP Swaps

This section discusses issues that arise when APs are upgraded with new units. Often, a simple upgrade is not simple at all. Different AP capabilities and network

requirements may mean changes beyond a simple one-for-one hardware replacement. A hardware upgrade should be designed and planned as if it were a

new deployment. In many cases, it is, for all practical purposes.

AP Capabilities Assessment

First, compare the EIRP power levels and receive sensitivities of the old and new APs. It is common for newer AP hardware to have a greater receive sensitivity

and possibly a higher EIRP as well. This can increase the cell overlap beyond the original design. A higher transmit power (EIRP) means that transmitted signals

will propagate farther. An increased receive sensitivity means the new AP will be able to hear weaker signals from a greater distance. Too much overlap

between AP cells will result in increased interference causing the available airtime per AP to drop. On the other hand, new APs with lower EIRP and/or lower

receive sensitivity than the old APs will leave coverage holes as they will create smaller cell sizes. AP transmit power levels can be adjusted up or down within

the limits of AP specifications to mitigate this.

If replacing the current APs with a different number of APs or APs with significant differences in capabilities (3-stream vs. 4-stream, etc.) the number and

placement of APs will likely be different. A new site survey is highly recommended.

The table below compares the RF capabilities of in-place and new APs. Significant differences in direction or width of radiation lobes can cause problems. With

Ruckus AP upgrades, a transmit power reduction may be needed to compensate for disparate AP types if a site survey is not performed, see

Table 5 for recommended values.

TABLE 5 POWER BACKOFF FOR ONE-TO-ONE AP SWAP

# Chain Configuration Suggested power backoff (dB)

Replacement AP Installed AP

1 4x4 2x2 6

2 4x4 3x3 3

3 4x4 4x4 None

4 3x3 2x2 4

5 3x3 3x3 None

6 2x2 2x2 None



January 2017

AP Features and Configuration

Power over Ethernet (PoE)

PoE is a popular option to provide power to access points over existing CAT5/CAT6 cabling infrastructure. The type of AP can affect the amount of power

required. Thanks to MIMO technology, many modern APs use multiple radio chains, which generally means more power is necessary. As consumed power varies

between different AP models it is best to consult with the datasheets for exact requirements. Most APs that require 802.3at power limits may still operate with

less power provided by the switch port but with some functionality disabled.

Let’s consider a generic 802.11ac 4x4:4 AP as an example of what can be expected with a reduced PoE budget.

• Power consumption of a 4x4 Access Point exceeds PoE/802.3af power limit and it requires a PoE+/802.3at compliant source to function in a 4x4

(5G) + 4x4 (2G) config.

• If powered through a PoE/802.3af source, the chain configuration drops to 4x4 (5G) + 2x2 (2G)

• AP may have a GUI or CLI provisioning option to specify PoE/PoE+ power

o Configurable options may include: Auto, 802.3af PoE, 802.3at PoE+

• Need to properly specify whether the AP is powered by a PoE/PoE+ switch

Note: Since the PoE switch generally provides inline power to multiple devices connected to it, pay attention to total power budget the switch can accommodate.

This information is typically available in the datasheet. Additionally, many switches allow the configuration of a specified amount of inline power on a per-port

basis.

Wireless Configuration Optimizations

Multiple SSIDs

Most wireless access points can support multiple SSIDs. This feature adds flexibility to the WLAN deployments; unique SSIDs can be used to offer different

connection options for wireless devices. For example, an enterprise wireless network may include one SSID for secure connection to an internal network and offer

another SSID for guest access with device onboarding for BYOD devices.

WLAN Beacon Frames

Unfortunately, support for multiple SSIDs does not come without some degree of network performance impact. Remember that each SSID requires a certain

amount of management frame overhead. Too many SSIDs can result in a significant amount of airtime taken up by management traffic rather than client traffic.

Let’s look at beacon frames. Every SSID transmits a beacon once every ~100 ms. This transmission is typically done at a low PHY rate and can cause 1-2%

management overhead per SSID, even if the SSID is hidden (hidden SSIDs are still advertised in beacon frames, the name is just set to null). The more SSIDs are

active on a given radio the more beacons it must send and the amount of overhead will grow.



January 2017

WLAN Probe Request and Probe Response Frames

Clients transmit probe requests to identify APs in their vicinity, not just at the time of entering the network, but periodically to help with roaming. There are two

types of probe requests a client can send: a directed probe and a null probe. For directed probes, the client station sends a probe request for specific SSID. The

AP then sends a unicast probe response to the client. When a client sends a probe with no specific SSID name (null), the AP sends a list of all configured BSSIDs1.

Thus, if there are multiple APs on a channel and multiple SSIDs configured, the APs must send a response for all of them. Probe responses can flood the medium

and take an unacceptable amount of airtime away from clients.

Channel Allocation in FCC Regulatory Domain

As mentioned previously, channel overlap and reuse is an important part of any WLAN design. It is critical to understand the limitations of the regulatory domain

in which the network will be deployed. Regulatory domains vary by country, so be sure to check the rules for what is available first and take it into consideration

when planning. This section uses the FCC regulatory domain as an example, which allows spectrum usage in both 2.4 GHz and 5 GHz. Some of the 5 GHz

spectrum is required to support DFS preemption.

In the FCC regulatory domain ~60% of the 5GHz spectrum is shared with Radar and DFS requirement as shown in Figure 5.

There are fewer channels, which do not have to support DFS, than there are that do:

• Number of available non-DFS 80 MHz channels is limited to 2 vs 6 when DFS channels are also enabled.

• Number of available non-DFS 40 MHz channels is limited to 4 vs 12 when DFS channels are also enabled.

Number of available non-DFS 20 MHz channels is limited to 9 vs 25 when DFS channels are also enabled.

FIGURE 5 FCC 5GHZ CHANNEL PLAN

Since so much of the available 5 GHz spectrum requires DFS preemption support, the question is if it is worth it to use these channels or to avoid them?

Pros of using DFS channels:

• More channels, less overlap and reuse

Cons of using DFS channels:

Enabling DFS channels could result in interoperability issues with client devices

• Some legacy and “low-end” client devices are not certified to operate in DFS channels (to save certification cost!)

1 WLANs with hidden SSIDs do not respond to wildcard probe request.



January 2017

Increased probability of channel changes

• The AP will stop using the channel when radar is detected and switch to a different channel (dynamically selected)

A Word about Channel 144

Channel 144 straddles the U-NII-2 and U-NII-3 bands and was approved by the FCC in 2013. Because of this, older legacy clients (802.11a/n) are unlikely to

be certified to operate on this channel. This channel can still be used, even when legacy clients are expected, provided there are enough other APs on supported

channels as to avoid a coverage gap.



January 2017

Channel Allocation in ETSI Regulatory Domain

ETSI rules do not allow operation in the UNII-3 band. This results in about 20% less available spectrum than in the FCC domain, (see Figure 6).

There are additional conditions as well: U-NII-1 and U-NII-2A bands are restricted to indoor operation only. This means, in general, there is very limited

spectrum available for outdoor use (~50% less spectrum than FCC). One exception is the UK, which allows the use of U-NII-3 for fixed wireless but requires a

license from Ofcom.

Note: Ofcom recently released a proposal to allow outdoor operation in U-NII-1 and U-NII-2A, to remove the licensing requirements for U-NII-3 and possibly to

allow more 5 GHz spectrum.

FIGURE 6 ETSI 5GHZ CHANNEL PLAN

Channel and Power Options

Channel and power level assignment for wireless access points can be done either manually or automatically. In general, the automatic method is preferred as it

initiates adjustments dynamically without the constant involvement of network engineers and administrators. Automatic assignment allows additional flexibility in

selection of algorithms, configurable parameters or grouping of access points. There are certain scenarios, however, where manual channel and/or power

configuration may still be considered a good option. For example, if a wireless network is deployed in a controlled environment where a channel plan is created

and maintained by IT personnel and there is minimal impact from other Wi-Fi networks at the site. This is rarely available in dense, urban areas however. Having

a higher degree of manual control comes with more overhead for the network operations. It also requires IT staff with a good level of RF expertise.

Let’s look at the features available in Ruckus equipment.

AP Power Adjustments

Ruckus access points provide an option to back off transmit power (Tx Power) from the maximum power allowed. The following adjustments are supported:

• In one dB steps from 1 to 10 dB (-1 dB, -2 dB, -3 dB,…, -10 dB)



January 2017

• A “Min” option, which backs off power by 24 dB2

It should be noted that different UNII bands have different power limits. For example, the maximum allowed power is 6 dB lower in U-NII-2. This will affect the

signal strength and coverage of an AP and should be kept in mind when designing a deployment that uses UNII band channels with different maximum power

settings.

Ruckus strongly recommends that any manual power reduction be considered with care.

Automatic Channel Selection Options

RUCKUS ACCESS POINTS SUPPORT TWO MECHANISMS TO AUTOMATICALLY MONITOR AND OPTIMIZE CHANNEL SELECTION: CHANNELFLY AND BACKGROUND SCAN. A

SUMMARY OF THESE OPTIONS IS SHOW IN

Table 6 below.

Background Scanning

Background scanning determines the best channel by checking the amount of interference on all allowed channels. It does this by switching to each channel

briefly before coming back to its original channel. This off-channel scan occurs on a regular interval that is configurable by the administrator. It is important to

understand that when an AP is off-channel it is not servicing clients, therefore the more frequent the scan interval, the less time the AP will be available to service

clients waiting to transmit.

Background scanning allows a Ruckus AP to gather information, which is used in many ways:

• Gather information to determine optimal channel selection

• Discover neighboring AP candidates for load balancing

• Discover neighboring APs for Opportunistic Key Caching (OKC)

• Discover rogue APs

These are important functions and in many cases, are required for high-density optimizations such as Channelfly and client load balancing. The downside is that

whenever the AP is off-channel it is not available to service clients. Background scanning happens fairly quickly, but it can impact overall performance in a busy

network.

To take advantage of background scanning and maintain performance, the scanning interval can be tuned for specific environments. Background scanning can

be set on a global basis as well as a per-SSID basis. Background scanning is enabled by default every 20s. Unless there is highly variable interference, a good

default value to use for background scanning is 3600 seconds. For faster reaction times in very dense deployments with high interference, ChannelFly is

recommended.

2 There is no provision to specify absolute transmit power.



January 2017

ChannelFly

Channelfly is a sophisticated and Ruckus proprietary method of determining the optimal channel selection for an AP. As discussed earlier, RF interference is a

major cause of performance problems. Channelfly is unique in that it not only takes the current noise into consideration (both 802.11 and non-802.11) it also

looks at the potential capacity available on a channel as well. Unlike many other vendor solutions, Channelfly might choose any possible channel rather than

restricting itself to the traditional non-overlapping channels. This is perfectly fine. Theoretically, this can cause co-channel interference with other devices on

nearby channels but in reality it may not be a problem. In particular, when APs and devices are subject to attenuation, the signal strength drops off more quickly.

This is particularly true for the energy transmitted outside the center channel.

For example, an AP might be said to transmit on channel 6, but it is actually transmitting across the “overlapping” nearby channels 4 and 5 and 7 and 8. But the

main power and highest signal strength will be on the center frequency, channel 6. When the signal strength drops, the RF energy across all occupied sub-

channels drops as well. With enough attenuation, the energy on the non-center channels can drop below an acceptable noise floor and be used by another AP.

This means Channelfly might move APs to occupy an “overlapping” channel but not see as much of the other device’s transmissions due to the signal drop-off

beyond the center frequency.

Channelfly uses the 802.11h channel announcement method of notifying clients that it is about to change channels on an AP. Support for 802.11h is mandatory in

5 GHz clients but not 2.4 GHz. Because of this, some 2.4 GHz clients might not deal as well with channel changes. If this is a problem, please turn Channelfly off

for the 2.4 GHz radios.

TABLE 6 AUTO CHANNEL SELECTION ALGORITHMS

Load Optimizations

There are many types of loads that can occur in a wireless system, e.g. how many clients connect to the 2.4 GHz radio vs. the 5 GHz radio of a dual-band AP,

how well clients are spread across multiple nearby APs, etc. Any one of these can impact client performance. Unfortunately, many key decisions regarding client

behavior are determined by the client rather than the AP. This can be a problem since the client rarely knows “the big picture” of system-wide capacity

availability as well as the WLAN infrastructure. Many load optimizations in wireless systems are designed to mitigate poor client decision behavior.

Channelfly Background Scan

Metric Used Channel Capacity RSSI of “rogue” (co-channel) APs Optimality Optimal since capacity estimate factors in both WiFi and

non-WiFi interference, traffic in overlapping BSS, variation of channel characteristics over time, etc.

Sub-optimal since BG scan relies only on the loudness of the neighboring co-channel APs

Off-channel Scan Optional Requirement – Background Scan improves performance

Mandatory Requirement



January 2017

Band Balancing/Band Steering

Client side algorithms tend to favor 2.4GHz over 5GHz even though the data rates and capacity available on the 5 GHz radio tends to be higher. This is

because clients typically use the RSSI as the decision metric. 2.4 GHz tends to propagate further without as much loss, which means it will appear stronger to a

client even when the radio is in a dual-radio AP with a 5 GHz radio. Therefore, a higher Tx Power combined with lower path loss in 2.4GHz can make it appear

more favorable for client devices. Since optimal performance is typically achieved in the 5 GHz band, it is usually better to have dual-band clients connect to

that radio if possible. However, this doesn’t mean that there should be no clients on the 2.4 GHz radio. If there are no clients on the 2.4 GHz radio and the RF

environment is favorable, having some clients on the 2.4 GHz radio can improve overall performance by using this otherwise unused capacity.

Band balancing (also called band steering) maintains a user specified percentage of clients in the 2.4 GHz band with all remaining clients encouraged to connect

in 5 GHz band. Note the word “encouraged”. Since a client makes the decision of which radio to connect rather than the AP, all the WLAN infrastructure can do

is attempt to encourage the correct behavior. The band balancing feature facilitates dual-band clients to choose 5GHz over 2.4 GHz band by withholding

probe responses and/or authentication response from the 5 GHz radio on the access points.

The band balancing algorithm works as follows:

• Recognize client is dual-band e.g. is the AP getting probe requests on both 2.4 and 5GHz?

• RSSI of the probe request exceeds specified threshold

• Don’t do Bandsteering to “weak” clients

• If both satisfied, AP doesn’t send Probe Response and/or Authentication Response in 2.4GHz

Client Load Balancing

Client load balancing addresses another client behavior that can be sub-optimal: which AP the client will attempt to connect to. Since a client will prefer the AP

with the strongest signal, this can result in clients all attempting to connect to the closest AP rather than a less utilized AP with a slightly lower signal strength. This

results in clients overloading the radio of a few APs when there are others that are equally good being underutilized.

Client load balancing distributes the client load uniformly across adjacent APs on a given band. It ensures adjacent APs do not have vastly differing number of

associated clients. This assumes that both APs are able to service a particular client equally well.

The client load balancing algorithm works as follows:

• AP relies on background scan to identify adjacent APs and their client count. When enabled, background scanning will run at a user configurable

interval; 20 sec is the default value. Background scanning can be enabled selectively for 2.4 and 5 GHz bands.

• An AP is deemed to be adjacent if its RSSI exceeds a user specified threshold, which has the following default values

o 2.4 GHz 50 dB

o 5 GHz 43 dB

• If the client limit is reached, the AP withholds probe and/or authentication responses.

It is important to note that this logic is not used to balance clients that are very close or very far. In this case, the client connection is decided based on RSSI

thresholds.



January 2017

Interactions with Band Balancing:

Band balancing will be temporarily disabled if the 5GHz radio has reached its client load client limit. The intent here is that it is preferable that no client is

ultimately denied the ability to connect.

Client Admission Control

Client Admission Control (CAC) is another useful feature, which helps maintain an optimal user experience for the clients already connected to the AP. An access

point will start preventing associations from new clients if it finds itself overloaded. To prevent new client associations, the AP withholds probe and/or

authentication responses. This situation only occurs under a specific set of conditions. Let’s look at the algorithm in more detail.

The conditions when Client Admission Control will be initiated are as follows:

The number of clients associated to the AP exceeds specified threshold - Min Client Count

AND

Airtime utilization as estimated by the AP exceeds specified threshold - Max Radio Load

AND

The estimated available throughput per client falls below specified threshold - Min Client Throughput

Available Throughput Estimate

Where available, throughput per client is calculated as follows:

Estimated capacity / Number of associated clients

CAC vs. Client Load Balancing

How does CAC differ from Client Load Balancing? CAC doesn’t check whether there is an adjacent AP available for the client to associate before withholding

probe/authentication responses. No background scanning is necessary for CAC to operate.

SSID Rate Limiting and Airtime Fairness (ATF)

In real-world environments, many different types of devices may connect to the wireless network. There could be a mix of older 802.11 clients, which can only

support legacy standards (b/g/a), 802.11n clients, as well as the latest devices with 802.11ac support. Although all of these devices can coexist on the same

network and even on the same BSSID, wireless access points offer additional enhancements that help use available resources in most efficient way.



January 2017

Rate Limiting

Ruckus WLAN controllers offer several configuration options to limit a device bandwidth in downlink and/or uplink direction. This setting is specified in Mbps and

can be applied on a per SSID basis. This option is often used to offer different tiers of Wi-Fi access with different rate limits applied based on the tier of

service, e.g. “free” vs. “premium”.

It is important to understand that rate limiting works by dropping packets at the wired interface of the AP. If many devices are rate limited and attempt to

exceed their limit, the number of retries due to the dropped packets could severely impact overall network performance. Rate limiting via the AP should be

approached with caution. A traffic shaping mechanism elsewhere in the network is typically more suitable and will result in fewer dropped packets and

retransmissions in a busy network.

Technically, rate limiting does not directly control the airtime occupied by various clients/SSID even though that is the overall intent. Legacy clients and/or far-

away clients can still occupy too much or too little airtime. Therefore, another layer of traffic management is needed. Airtime Fairness (ATF) helps to address this.

Airtime Fairness and Packet Aggregation

Let’s have a look in more detail at what happens when a network has a mix of legacy 802.11a/b/g and VT/VHT 802.11n/an clients.

FIGURE 7 MIXED CLIENT WIRELESS NETWORK EXAMPLE

802.11n (HT) and 802.11ac (VHT) stations support MSDU and MPDU aggregation. This is an optimization introduced as part of the 802.11n standard that

allows the aggregation of smaller packets into a larger frame with much higher data payload. See “STA2 Data” in Figure 8. Legacy stations don’t support

aggregation, which is far less efficient. As a result, legacy stations often get a smaller amount of overall airtime.

Note: In 802.11ac data frames are sent as an aggregated MPDU. Maximum transmission length of a frame is limited by time and is about 5ms.

AP11ac

STA111a

STA211ac



January 2017

As 802.11ac can offer very high PHY rates, the aggregate frame may contain over 4MB of data. Although 802.11n also offers MSDU and MPDU aggregation,

the maximum level of MPDU aggregation is much smaller making it less efficient than 802.11ac although still better than legacy devices.

FIGURE 8 DATA PACKETS WITH AND WITHOUT MSDU / MPDU AGGERGATION

Near/Far Clients

Another example where unequal distribution of airtime between clients can occur is when there is a mixture of stations close to (Near) and far away (Far) from

an AP. Because it is closer, the Near STA operates at a higher PHY date rate than the Far STA. Because the Far STA has a lower PHY rate, it will require more

airtime to transmit the same amount of data as a faster client. This can significantly slow down the network. See Figure 9 example for Near/Far scenario with

Legacy 802.11a/b/g clients.

FIGURE 9 NEAR /FAR 802.11A CLIENTS

In the Near/Far scenario, airtime utilization (Figure 10) will look similar to the mixed client network as shown in Figure 8. The larger amount of air time required

for STA2 to transmit its data packet is the result of a lower PHY rate and the limitation of being unable to send a packet larger than the standard MSDU /

MPDU size. Unlike the HT and VHT clients, it must also follow each packet with an ACK frame, further decreasing the amount of airtime available for data

packets.

FIGURE 10 NEAR AND FAR STATIONS AIRTIME RELATIONSHIP

Airtime Fairness, ensures stations do not end up with too little or too much downlink airtime. ATF is always in use by Ruckus APs to maintain equitable airtime for

disparate clients and cannot be disabled. If more airtime is desired for certain clients, there is an option to allocate 4 times more airtime to “premium” users by

specifying an SSID as high priority.

Overhead Reduction for High Density Deployments

AP11ac

STA111a

STA211a



January 2017

Efficiencies can be achieved in many ways. This section discusses how to minimize the impact of overhead—frames that do not have data payloads - on a very

dense deployment with many APs and clients.

Management Frames

Although much of 802.11 traffic is data, a significant amount is transmitted that is overhead. Management frames, including beacons, make up a significant

percentage of traffic. This is particularly true in dense deployments. Part of the problem is the data rate: management frames are usually transmitted at the

lowest and slowest rate. This works well for range-limited deployments but it creates unnecessary overhead for high-density deployments.

To optimize for this situation, you can adjust the following:

• Minimum rate for management frames (Mgmt Tx Rate) sent by an AP. This will prevent it from transmitting any management frame at rates below the

specified rate. Keep in mind, this will influence the cell size, as management packets sent at higher PHY rate will require shorter distances between the

client and AP

• The minimum rate for data frames (BSS Min Rate). This prevents an AP from transmitting any data frame at rates below the specified rate

802.11b Client Support

The presence of 802.11b clients will also slow down a network to make sure traffic is backwards compatible. Configuring a radio for OFDM-only support will

force the AP to only transmit OFDM frames, which is more efficient than DSSS/CCK, which is used by 802.11b clients. Doing this, however, will prevent 11b-only

clients from joining the network. If this optimization feature is used, ensure that there are no legacy clients that require 802.11b support.

Proxy ARP

Another feature that may reduce overhead considerably is Proxy ARP. When Proxy ARP is enabled, the AP responds to ARP requests on behalf of clients

associated to it. Enabling Proxy ARP mitigates flooding the wireless medium with excessive ARP requests. More airtime will be left to for data frame transmission.



January 2017

Advanced Optimizations

Management Frame Protection 802.11w

It is also known as Protected Management Frame (PMF). In most 802.11 networks, only data frames get encrypted by default. 802.11 management frames

typically get transmitted without any encryption, making them vulnerable to rogue attack. For example, a rogue device can spoof legitimate MAC address and

send Disassociation/Deauthentication to clients, disconnecting them from an AP with a denial of service attack. The 802.11w standard addresses this

vulnerability.

SmartZone controllers have three options available– Disabled, Capable, Required. The default setting is Disabled. This is because not all devices support

802.11w. Those that do not may have trouble connecting to an 802.11w-enabled SSID.

Wireless Intrusion Detection / Prevention System (WIDS/WIPS)

Wireless Intrusion Detection and Prevention can have a significant impact on wireless network performance. A performance tradeoff is made for the enhanced

security. Before enabling these features, it is helpful to understand exactly what these tradeoffs will be.

Rogue AP Detection and Containment

Rogue APs are typically APs that do not belong to the network but are operating on it. As such, they are viewed as a class of wireless security vulnerability even

though their placement is usually not malicious. APs are classified into one of three types:

• Infrastructure/Authorized – the APs owned and operated by an organization to provide wireless service

• Interfering – APs that are not owned by the organization but are close enough they can be heard by the organization’s client devices

• Rogue – APs that are not owned by the organization but are on the corporate network infrastructure

The types of rogue APs that WIDS can identify includes:

• Unauthorized APs that are spoofing the SSID and BSSID of an authorized AP

• Unauthorized APs that can access the wired internal network

There are external APs that should not get classified as a rogue AP. These are typically APs which do not belong to the network but are within close proximity

e.g. neighbor APs. They are referred to as interferers.

One feature of rogue AP WIPS is known as rogue AP containment. Rogue AP containment provides protection from malicious rogue APs by preventing clients

from associating to the rogue AP. This is done by having an authorized AP broadcast a de-authentication frame to the client. The AP will spoof the BSSID of the

rogue AP and transmit the frame to the client on the rogue AP’s channel. This is true even if the infrastructure AP is on a different channel. Background scanning

must be configured in order to collect information on rogue APs and spoof their associated clients.



January 2017

Appendix A: Understanding Inter-client RF Interference

Introduction

802.11 networks use CSMA/CA to handle contention when there are multiple access points and clients on the same channel. Contention between wireless devices

can be managed by the protocol rules and ensure each 802.11 endpoint gets some share of airtime. Most wireless networks use multiple APs. Ideally, when

multiple APs are used, they are placed on different channels. This provides more bandwidth for the clients and less interference. Clients will connect and use

different access points in this scenario. They will also communicate on different channels. Assuming there is no more than one AP on the same channel, Co-Channel

Interference (CCI) does not exist or is well mitigated with such a design. Adjacent Channel Interference (ACI) or interference between non-adjacent channels may

still exist when different channels are used. In this case, is there any impact from one client to another? Is it measurable? What happens when multiple wireless

clients operate on different channels and they are within a reasonable proximity to each other, i.e. within same classroom?

FIGURE 11 INTER-CLIENT INTERFERENCE IMPACT

Investigations

The Ruckus TME team staged several lab tests to investigate inter-client interference in real-world environments.

Several test cases were performed that were designed to show how the TCP throughput measured between IxChariot endpoints varies depending on the

distance between two 802.11 wireless clients, the 5GHz channels used and the channel bandwidth assigned. Clients (STA1 and STA2) are connected to two

different APs (AP1 and AP2) in these test cases and use two different IxChariot endpoints on the wired side of network to eliminate any dependency of wired

endpoint performance.

All tests were conducted in 5 GHz band and within FCC regulatory domain.



January 2017

Close Distance Tests

For this test two R310 APs were configured to operate on two different channels, 149 and 165, with 20 MHz channel bandwidth. Each of the APs had one client

connected - STA1 or STA2. The distance from the AP to the client was set at 50ft. The two clients were then placed at 2ft from each other for the first TCP

throughput test. Spacing was increased to 18ft and TCP throughput was measured again.

To generate a load, STA1 and STA2 had IxChariot endpoints installed. Each of them communicated with a dedicated IxChariot endpoint on the wired side to

avoid any potential bottlenecks. Both testes were run and measured using the downlink direction where packets with larger payloads would be send from the AP

to the client, while STA1 and STA2 endpoints would send small TCP ACK packets back to the wired server.

Graphs in Figure 12 and Figure 13 show 3 minute TCP throughput tests for one client (STA1). The second client (STA2) initiates its own TCP data transfer 40

seconds into the timeline with a 1 min test duration.

The test results look very different when we change the distance between these clients. With a 2ft client-to-client separation (Figure 12) there is a significant

impact from STA2, with a dip when the tests begins. Starting at that point, STA1 loses about 50% throughput while STA2 is passing data.

With an 18 ft client-to-client separation this impact is not observed (Figure 13) and the first client delivers consistent performance regardless of the second

client running.

FIGURE 12 IMPACT ON CLIENT PERFROMANCE THROUGHPUT WITH 2FT SPACING

FIGURE 13 W ITH18 FT SPACING BETWEEN CLIENTS NO IMPACT OBSERVED



January 2017

The Impact of Channel Frequency Separation

A best practice is to avoid RF interference by using separate channels. This doesn’t help when the two clients are on the same AP, but it should theoretically

remove inter-client interference if they are connected to two APs on separate channels. But is this really the case? What happens if the clients happen to be very

close to each other?

Two tests are performed with different amounts of channel separation. When we talk about channel separation in this case, we are referring to how much

frequency space is between the two devices (in MHz). You can have devices that are on separate channels that are adjacent to each other or you can have

channels that are on the extreme sides of the allowed spectrum.

In these tests, the clients were spaced 2ft from each other but the frequency separation between operating channels was different. In one case, the channels

were separated by 200 MHz and 655 MHz for the second test. Test results are shown on Figure 14 below. We kept the basic configuration the same as shown

on Figure 11 but as a metric we use an aggregate TCP downlink throughput of STA1 and STA2.

FIGURE 14 TCP AGGREGATE THROUGHPUT MEASUREMENTS WITH VARIOUS CHANNEL SPACING

As you can see, closely spaced clients have a higher combined throughput when they connect to APs on different 5 GHz channels with greater frequency

separation. Yet technically, both clients were on non-overlapping channels. Common wisdom is this should be enough to avoid interference but in this case, it is

clear that it is not equally true for all cases.

Considering that there is an improvement with 655MHz separation, how does this compare if both clients connect and pass data through the same AP with even

wider channel bandwidth? Since 20 MHz channels were used for the test above, let’s allocate a single 40 MHz channel and re-run the test with same 2ft

separation between clients.



January 2017

Test result is shown on Figure 15 below.

FIGURE 15 TWO 20 MHZ CHANNELS VS. SINGLE 40 MHZ CHANNEL

Conclusion: Two close spaced clients operating on a single AP with a 40 MHz channel produce higher combined throughput than when they are connected to two

APs on different 5GHz channels with 20 MHz bandwidth. This discovery is significant as it may influence some AP deployment and channel planning scenarios.



January 2017

Lessons Learned

This experiment raises several points that are worth considering for any dense AP deployment design.

• There is a measurable performance impact between clients at close range, even when they operate on different channels

• This impact diminishes as distance between clients increases

• Network aggregate throughput may not grow linearly when a second AP is added on different channel

As always, if the utility of certain scenarios is in question, it is always good to validate them with testing where possible before deployment.



January 2017

About Ruckus

Headquartered in Sunnyvale, CA, Ruckus Wireless, Inc. is a global supplier of advanced wireless systems for the rapidly expanding mobile Internet infrastructure

market. The company offers a wide range of indoor and outdoor “Smart Wi-Fi” products to mobile carriers, broadband service providers, and corporate

enterprises, and has over 36,000 end-customers worldwide. Ruckus technology addresses Wi-Fi capacity and coverage challenges caused by the ever-

increasing amount of traffic on wireless networks due to accelerated adoption of mobile devices such as smartphones and tablets. Ruckus invented and has

patented state-of-the-art wireless voice, video, and data technology innovations, such as adaptive antenna arrays that extend signal range, increase client data

rates, and avoid interference, providing consistent and reliable distribution of delay-sensitive multimedia content and services over standard 802.11 Wi-Fi. For

more information, visit http://www.ruckuswireless.com.

Ruckus and Ruckus Wireless are trademarks of Ruckus Wireless, Inc. in the United States and other countries.

Copyright 2017 Ruckus Wireless, Inc. All Rights Reserved. Copyright Notice and Proprietary Information No part of this documentation may be reproduced, transmitted, or

translated, in any form or by any means without prior written permission of Ruckus Wireless, Inc. (“Ruckus”), or as

expressly provided by under license from Ruckus

Destination Control Statement

Technical data contained in this publication may be subject to the export control laws of States law is prohibited. It is

the reader’s responsibility to determine the applicable regulations and to comply with them.

Disclaimer

THIS DOCUMENTATION AND ALL INFORMATION CONTAINED HEREIN (“MATERIAL”) IS PROVIDED FOR GENERAL

INFORMATION PURPOSES ONLY. RUCKUS AND ITS LICENSORS MAKE NO WARRANTY OF ANY KIND, EXPRESS OR

IMPLIED, WITH REGARD TO THE MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF

MERCHANTABILITY, NON-INFRINGEMENT AND FITNESS FOR A PARTICULAR PURPOSE, OR THAT THE MATERIAL IS

ERROR-FREE, ACCURATE OR RELIABLE. RUCKUS RESERVES THE RIGHT TO MAKE CHANGES OR UPDATES TO THE

MATERIAL AT ANY TIME.

Limitation of Liability

IN NO EVENT SHALL RUCKUS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL OR CONSEQUENTIAL

DAMAGES, OR DAMAGES FOR LOSS OF PROFITS, REVENUE, DATA OR USE, INCURRED BY YOU OR ANY THIRD

PARTY, WHETHER IN AN ACTION IN CONTRACT OR TORT, ARISING FROM YOUR ACCESS TO, OR USE OF, THE

MATERIAL

ap deployment guide - spiceworks

Documents