CAPsMAN WiFi Layer1 / Layer2

Optimisation

Optimising CAPsMAN settings to improve performance and why

by Ron Touw

EU MUM 2018
 Ron Touw

• 1970’s Trained to be a Naval Radio and Electronics

Officer and learnt a lot of theory about RF and
Marine Electronics
• 1980’s Entered into UK Government Service and
learnt a lot about the practical application of RF
• Next 30+ years ‘playing’ with RF from 10kHz to
100GHz

EU MUM 2018
 Ron Touw

• 1997 - 1999 Saw the birth of point to point bridging

by the likes of Breezom (now Alvarion)
• Early 2000’s started to be more involved with WiFi
being used by WISPs and troubleshooting
interference problems mainly on 2.4GHz
• 20 years later, still mostly troubleshooting WiFi
problems on… 2.4GHz 😛

EU MUM 2018
 Ron Touw

• Became MikroTik Consultant in 2008

• Became MikroTik Trainer #57 in 2009
• Head of Technical Training at LinITX
• Founding Member in 2017 of the MikroTik Training
Bootcamp hosted at MikroTik’s RouterOS Training
Centre in Riga

EU MUM 2018
 MikroTik BootCamp 2018

EU MUM 2018
 MikroTik BootCamp 2018

*Correct as of March 2018

EU MUM 2018
 Optimising CAPsMAN settings to
improve performance

But, more importantly - “Why is this even important?”

EU MUM 2018
 It’s only WiFi - it works!

• View of WiFi industry looking in from ‘outside’?

• ‘Anyone can install WiFi - just stick a box in the
house / office / garage / wherever you like’
• ‘It just works, my one AP feeds the entire house!’
• It’s all automatic, let it choose the best channel

EU MUM 2018
 It’s only WiFi - it works!

• Therefore the perception is that it must be really

easy to cover the office / warehouse / school
• Just throw loads more APs in!
• Leave everything to default settings
• Reality is that the WiFi protocol was designed to go
as slow as required to make it still work

EU MUM 2018
 Example of poor install, yet still works!

EU MUM 2018
 Common Mistakes

• Too few APs - massive gaps in coverage

• Too many APs - too much noise / interference
• APs set to maximum power - noise / interference
• 40MHz on 2.4GHz - no known clients support it
• 80MHz on 5GHz - poor coverage and more
interference

EU MUM 2018
 Common Mistakes

• Running tests to speedtest.net in each area to test

speed to internet
– does not test JUST the internal network / AP only!
– Also a risk you are not connected to the right AP
being tested

EU MUM 2018
 Common Mistakes

• Looking at phone/tablet signal level and going for

“4-5 bars” coverage everywhere
• Most common mistake of all. Signal strength of
APs is not the same as ‘quality’ or high throughput
• Higher signal strengths of APs usually generates
poorer performance not better

EU MUM 2018
 If WiFi is not placed under stress - does anyone notice?

• “Low density WiFi” (e.g. home, small office / hotel)

• Usage more light / bursty in nature and limited by
internet connection, not wireless connection
• “High density WiFi” (e.g. sports arena, concerts,
theatres, Conferences, Schools, MikroTik MUM?)
• If the WiFI is sub-par, now everyone complains!

EU MUM 2018
 Why is the WiFi ‘broken’?

• Cabling faults are easy to find

• WiFi faults are more hidden
• RF is ‘invisible’
• Far more difficult to
diagnose without specialist
tools
© omegaanimalremoval.com

EU MUM 2018
 Interference

• Typical sources of 2.4GHz problems?

• Interference - some non-802.11 devices
– Bluetooth, Microwave Ovens, CCTV, ‘Video
Senders’, RF based Motion Sensors, USB3.0,
Zigbee, electrical switches, commutator motors.
– Can be highly disruptive especially if ‘100%
occupancy’

EU MUM 2018
 Interference

• Besides non-802.11 signals, what is the biggest

‘interferer’?
– Plain ordinary ‘Congestion’
– i.e. ’Co-Channel-Interference’ (CCI) from other
802.11a/b/g/n/ac devices!

EU MUM 2018
 Interference - CSMA/CA - A little theory

• WiFi devices use a system called Carrier Sense

Multiple Access / Collision Avoidance (CSMA/CA) to
determine if the channel is busy
• Carrier Sense uses a method of determining if the
channel is available called Clear Channel
Assessment (CCA)

EU MUM 2018
 Interference - CCA

• 802.11 uses two methods to assess the Channel

occupancy and will wait until it has a clear channel
before transmitting based on :-
• Energy Detection (for non-802.11 signals)
• Carrier Sense (where the signal is 802.11)

EU MUM 2018
 Interference - CCA Energy Detection

• Any non-decodable signal that is above the “ED

threshold level” makes the channel marked as being
“busy” / “in use”
• ED threshold is set to be 20dB more than the level
required to successfully decode an 802.11 header
• If interference is 100% duty cycle - device will
never transmit

EU MUM 2018
 Interference from motion alarm sensor - 100% occupancy

(WiFi Explorer Pro Screenshot)

EU MUM 2018
 Interference - CCA Carrier Sense

• CCA Carrier Sense decodes the signal’s physical

header at the start of each transmission and takes
note of the transmitted duration field
• Requires the signal to be 802.11 and for the
physical header to be decoded without errors
• Header contains transmission time
• Device will ‘back off’ and wait for that time

EU MUM 2018
 Interference - CCA Carrier Sense

• Regardless of which system used (Carrier Sense or

Energy Detection) station will wait and not transmit
until the chanel is clear
• Busier the channel, the slower the throughput
• So what consumes ‘airtime’ on an 802.11 network?

EU MUM 2018
 Interference - Contention Types

• Three major types of 802.11 based contention

• Clients all contending for one AP’s airtime
• APs contending with other APs
• Clients connected to different APs but contending
with each other for airtime

EU MUM 2018
 Client to Client and Client to AP CCI

Ch1

EU MUM 2018
 AP to AP CCI
Also when using ‘repeater’ mode

Ch1 Ch1

EU MUM 2018
 Client to Client CCI
(with APs far enough apart)

Ch 1 Ch 1

EU MUM 2018
 Airtime - CSMA/CA

• AP to AP CCI is more common than you think

• APs will usually be able to ‘see’ each other better
than client to clients can (as the APs are often
higher up and mounted on the same horizontal
plane)
• Clients will be shielded more from each other by
people and/or building

EU MUM 2018
EU MUM 2018 © Duncan Soar
 Airtime - CSMA/CA

• Therefore important to keep ‘same channel APs’ as

isolated from each other as possible
– Use the building to shield between them
– Use directional antennas?
• Also bear in mind the potential of Co-channel
interference with clients half way between two APs
on same channel!

EU MUM 2018
 CCI or ACI - which is preferred?

• 802.11 is designed to cope with CCI - better called

‘channel contention’ or ‘congestion’ than noise / non
802.11 interference
• Very important to use non-overlapping channels
• Overlapping channels will be seen as non 802.11
noise / interference

EU MUM 2018
 Airtime - CSMA/CA

• Two very active APs on the same channel, will force

them to both back off and send beacons late (or
even give up altogether and miss a beacon),
• Thus each AP will reduce user throughout until the
other AP is quiet. Especially with ‘repeaters’
• Two APs on the same channel is effectively having
a very large cell on one channel

EU MUM 2018
 Airtime - Beacons

• Access points have to send out beacons

• Beacons contain data such as SSID, RadioName
and advertised features / extensions
• Each SSID requires a separate beacon
• Beacons are sent at the slowest basic rate

EU MUM 2018
 Airtime - Beacons

• Therefore more SSIDs equals more time consumed

• Starving time from clients
• Beacons can be configured to be sent at higher data
rates, therefore less time consumed
• Increasing airtime for clients

EU MUM 2018
 Airtime - Probes

• Client will always send probe requests at their

slowest data rate (e.g. 6Mbps)
• AP will reply with Probe Response at the AP’s lowest
data rate (perhaps 24Mbps)
• Client will ACK with AP’s lowest data rate 24Mbps
• But the initial Client probe will always be at 6Mbps :(

EU MUM 2018
 2.4GHz vs. 5GHz

EU MUM 2018
 2.4GHz vs. 5GHz

• Majority of clients now support both bands

• More channels on 5GHz, less interference
• Therefore need to encourage clients to use 5GHz

EU MUM 2018
 Enough of the Theory!
Time for Solutions :)

Or why it is important - now learn what to do about it :)

EU MUM 2018
 Design Solutions?

• Try to avoid using too many SSIDs

• No more than 4-6 SSIDs per physical interface
• The lower the number of SSIDs the better
• Turn off WPS as the WPS ‘Information Element’
increases the size of all transmitted beacons

EU MUM 2018
 Design Solutions?

• Try to avoid using 802.11b if possible

• Reduce Tx power on 2.4GHz to encourage clients to
connect to the stronger 5GHz signal instead
• consider using two SSIDs to use human psychology
• On 2.4GHz use “Slow Internet” ?
• On 5GHz use “Fast Internet” ?

EU MUM 2018
 Design Solutions?

• If you *have* to use 2.4GHz - Ch’s 1, 6, 11 ONLY

• On 5GHz consider that some clients do not support
DFS channels and also that APs on DFS channels
may go offline
• Therefore ensure there are always some non-DFS
channel APs nearby (Channels 36-48)

EU MUM 2018
 Surveying Tip to measure AP to AP CCI

• Measure signal levels of *other* APs on all APs to

determine if AP to AP co-channel interference is
present
– Look for other AP levels around -80dBm or less
– CAPsMAN now allows background scanning :)

EU MUM 2018
 Measuring other AP signal levels

Background Scan on remote CAP

EU MUM 2018
 Measuring other AP signal levels on same channel

Check signal levels of all other APs on the same

channel as that AP
Ensure other co-channel APs’ signal levels are as low
as possible (>-80dBm)
EU MUM 2018
 Improving Beacons and Probes - Data rates

• Beacons and probes transmitted at lowest basic rate

• So, raise the data rate of the lowest basic rate!
• Not recommended to go above 24Mbps!
• 12Mbps is usually well supported by most clients
• Test connectivity and roaming with customer’s
devices not your own!

EU MUM 2018
 CAPsMAN Configuration Overview

Interface Higher

Configurations Precedence

Lower
Rates Security Cfg Datapaths Channels

EU MUM 2018
 CAPsMAN Configuration Override - Frequency

Lower precedence
Higher precedence
• Changing any setting at a ‘higher’ level (on the left)
overrules the same lower settings (on the right)

EU MUM 2018
 • For example, create a new channel definition with a
frequency of 2412MHz

EU MUM 2018
 • Apply channel setting to new configuration but then
maybe ‘overrule’ the frequency on the configuration
with 2437MHz?

EU MUM 2018
 • Then add the configuration to
an interface and overrule the
frequency to 2462MHz?

EU MUM 2018
 * End Result *

EU MUM 2018
 Removing 802.11b

• First major improvement to speed up the wireless

network is to remove all support 802.11b

EU MUM 2018
 Before changes - 802.11b active (the theory)

• Default
shows
CCK:1-11
meaning
802.11b is
active

EU MUM 2018
 Before changes - 802.11b active (yes - reality!)

(WiFi Explorer Pro Screenshot)

EU MUM 2018
 Removing 802.11b from 2.4GHz

• To remove
802.11b, one
could create and
use a channel
setting the Band
to ‘2ghz-g/n’

EU MUM 2018
 Removing 802.11b from 2.4GHz

• Or, set Band to

‘2ghz-g/n’ on the
Channel tab on
Configuration

EU MUM 2018
 Removing 802.11b from 2.4GHz

• Or, set Band to

‘2ghz-g/n’ on
the Channel
tab on
Interface

EU MUM 2018
 End result? g/n only

• Result is
802.11b
disabled,
(CCK:1-11
has gone)

EU MUM 2018
 End result? g/n only!

(WiFi Explorer Pro Screenshot)

EU MUM 2018
 Further improvements

• Now with 802.11b banished to where it belongs,

what about the slower data rates?
• First we need to create a ‘Rates’ config that contains
all the same rates as the default

EU MUM 2018
 Default starting position - 802.11g/n only (no 802.11b)

EU MUM 2018
 Hopefully - nothing has changed

EU MUM 2018
 Further improvements

• Recommend lowest basic rate is raised to at least

12Mbps or possibly even as high as 24Mbps
• Remove the slower rates of 6Mbps and 9Mbps and
enable 12Mbps (preferred option)
• Or, remove the slower rates of 6Mbps, 9Mbps,
12Mbps and 18Mbps and enable 24Mbps

EU MUM 2018
 Removal of all slow rates below 12Mbps

EU MUM 2018
 End result? g/n only, no slow rates (the theory?)

EU MUM 2018
 End result? Better performance achieved

(WiFi Explorer Pro Screenshot)

EU MUM 2018
 Testing Coverage and interference

• Low cost Solutions (Mac OSX)

– WiFi Signal (~€6) - perfect for testing when laptop
connects, disconnects and roams between APs
– WiFi Explorer (~€16) - In depth analysis of AP
Beacons
– WiFi Explorer Pro (~€81) - Adds Analog Spectrum
Analyser (with suitable hardware) and much more

EU MUM 2018
 Testing Coverage and interference

• Low cost Solutions (Windows)

– Ekahau HeatMapper (Free)
– NetSpot (€0 to €405) (also available for Mac)

EU MUM 2018
 Testing Speed

• Possible Solution
– BTest.exe on laptop with at least same capability
as the AP under test
• However Bandwidth Test tool can use up too much
CPU on slower CPU APs - test to another more
powerful RouterBoard ‘behind’ the AP (CCR?)
• Ensure you are connected to the AP being tested

EU MUM 2018
 CAPsMAN DataPath

• Try to use ‘local forwarding’ of data rather than

tunnelling back to CAPsMAN controller
• ‘local forwarding’ is faster than ‘manager
forwarding’ mode
• At the moment FastPath on the CAP is not
possible

EU MUM 2018
 End Result?

• End result should now be increased throughput

• More clients on 5GHz than on 2.4GHz
• Higher data rates, smaller cells, higher throughput
• Co-Channel Interference reduced
• non-802.11 Interference removed/reduced
• Happier customers? We can hope! :)

EU MUM 2018
 Want a couple of CAPsMAN Easter Eggs?

• Since v6.42rc39 (7 March 2018) …

• In Winbox, Access List can only directly specify a
specific named interface or the default of ‘any’
• However, in CLI, it can also specify an Interface List
• /caps-man access-list set {AccessListName}
interface={all|any|discover|dynamic|none|
{InterfaceListName}|{SpecificInterfaceName}

EU MUM 2018
 Want a couple of CAPsMAN Easter Eggs?

• When a CAP is provisioned and the interface is

added in CAPsMAN, the CAP interface can be also
dynamically added to an interface list named in the
DataPath config (now also in Winbox)
• /caps-man datapath set {DataPathName}
interface-list={all|discover|dynamic|none|
{InterfaceListName}

EU MUM 2018
 Thank You!

EU MUM 2018