1. technology and computing
2. computer certification

Design and Deployment of Wireless
LANs for Mobile Applications
BRKEWN-2000
Jerome Henry, Technical Marketing Engineer
Agenda
• How Much Bandwidth do You Need?
• Determine Cell Size – Shape
• Determine Cell Size – AP Power
• Determine Cell Size – Protocols and Rates
• Determine Cell Size – 20, 40, 80 MHz?
• AP Placement Strategies
• Taking Care of the Roaming Path
• Client and AP 802.11 Optimizations
• Last Words - Troubleshooting
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
3
How Much Bandwidth do you Need?
How Much Bandwidth Is Required?
• Often Less than You May Think
• It is most likely that you won’t be supporting
just one application
• Design for the highest bandwidth demand
that you intend to support
– What you need is the minimum acceptable
throughput that the application will require
– Most users use only ONE high performance
demanding application at a time
• Multiply this number by the number of
devices that you need to support
• This is the aggregate bandwidth you will
require in your space
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
5
Application – By Use
Case
Throughput –
Nominal
Web - Casual
500 Kbps
Web - Instructional
1 Mbps
Audio - Casual
100 Kbps
Audio - instructional
1 Mbps
Video - Casual
1 Mbps
Video - Instructional
2-4 Mbps
Printing
1 Mbps
File Sharing - Casual
1 Mbps
File Sharing - Instructional
2-8 Mbps
Online Testing
2-4 Mbps
Device Backups
10-50 Mbps
How Much Bandwidth do You Need?
• Example, Skype (Up/Down):
Call type
Audio
Video/screen share
Video HD
Group Video (5 people)
Typical
Bandwidth
30Kbps/30kbps
130kbps/130kbps
1.2 Mbps/1.2 Mbps
130 kbps/2 Mbps
• Now that you get the picture, a few other examples:
–
–
–
–
Fring (video): 135 kbps,
Facetime (video, iPhone 4S): 400 Kbps, (audio) 32 kbps
Viber (video) 120 kbps, (audio) 30 kbps
Skype/Viber/other chat: around 850 to 1000 bytes (6.8 to 8 kb) per 500 character
message
– Netflix (video), from 600 kbps (low quality) to 10 Mbps (3D HD), average 2.2 Mbps
– This bandwidth consumption is one way, you need to double for 2-way conversations
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
6
Real Life Example?
Medical Center
• Density studies show 12 users / cell on average
I need 6.65 Mbps throughput
everywhere in the cell
- > therefore I need it here
– Expected 2 HD video calls (Skype type)
– 5 audio calls
– All users may browse
• Let’s do the math:
– 2 HD video calls = 1.2 Mbps x 2 x 2 ways = 4.8 Mbps
– 5 audio calls… mmm what application?
• Skype too? 30 kbps x 5 x 2 ways = 600 kbps
– Others are browsing (5 people)… 250 kbps /people?
– Total = 6.65 Mbps needed
Funny that browsing requires more than voice
Should I design for browsing?
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
7
AP
VoIP Requirements
• VoIP carries voice sound with UDP and Real Time Protocol (RTP), voice
control traffic uses Real Time Control Protocol (RTCP)
– Voice sound is converted to digital packets using codecs
– Resulting packet size ranges from 8 to 64 bytes per packet (+40 bytes L4/L3
headers, +L2 header)
• Voice has very strict requirements as an “application”
–
–
–
–
–
–
Packet Error Rate (PER) <=1%
As low jitter as possible, less than 100ms
Retries should be < 20%
End to end delay 150 – 200 ms, 30 ms in cell
When these values are exceeded, MOS reduces
Your mission is to keep MOS high
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
8
Video Applications
• Video uses video and audio codecs
– Some codecs are built for real time exchange, some for streaming
– Video algorithms refresh entire images when large changes occur
– The changes generate traffic bursts
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
9
Cell Size and Shape
Cell Shape and Cell Size
 Your cell shape depends on the antenna you use:
 Directional
 Omnidirectional
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Cell Shape and Cell Size
 Your cell shape depends on the antenna you use:
 Directional
 Omnidirectional
 The cell size depends on 3 parameters:
 The AP power level
 The protocol you use (802.11a/b/g/n/ac)
 The Data rates you allow
 This is in open space… in real world, you also need to
account for RF obstacles
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Cell Size - AP Power
Power and Channel Plans
 Many BYODs support all UNII bands
 BUT: BYOD power varies with the band
 Client devices cannot actively scan UNII-2e (ch 100 to 140)…
 So those that support UNII-2e passively scan them only once every 6 to 18 cycles
(this means that, when on the Wi-Fi page, where an Iphone 5 scans every 11.4s, you
get UNII-2e channels every 68.4 s; when running to 8 second scans in roaming panic
mode, you get UNII-2e channels every 48 s)
 Avoid UNII-2e in your channel planning if fast roaming is a concern
 This may limit your options if you decide to use 40 MHz channels
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
15
Cell Size – Depends on Protocol and Rates
• Channel Utilization levels should be
kept under 50%.
• Higher power does not always mean
higher SNR…
• You can check it using a spectrum
analyzer like SpectrumExpert or Fluke
AirCheck
• Noise levels should not exceed -92
dBm, which allows for a Signal to
Noise Ratio (SNR) of 25 dB where a 67 dBm signal should be maintained.
Is it better now?
Blah blah blah
You are a bit quiet
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
16
Cell Size – Let’s get rid of the “Power” Checkbox
 RF is symmetrical on paper:
 “If your AP signal is heard by a client,
the AP should hear the client signal
symmetrically” (because the antenna Rx matches its Tx gain)
 In real world, this is true, if the client RF parameters (Tx/Rx sensitivity,
antennas, power level) are the same… otherwise, the client signal may be so
weak that the AP can’t make sense out of it
 2 causes:
 Client rate decision is based on CLIENT perception of AP signal – if AP
signal is strong, client will use high rate
 The client reaches the AP mixed in surrounding noise – SNR too low and AP
cannot demodulate
This is the AP signal (at phone level)
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
This is the phone signal (at AP level)
17
Power -client Side vs AP Side
 BYODs commonly have smaller power ranges than APs
Example: 3700i AP
(+4 dBi antenna on 2.4 GHz,
Example: Iphone 5
+6 dBi antenna on 5 GHz)
Band
Max Tx Power
2.4 GHz ISM
16 dBm
UNII-1
14 dBm
UNII-2
13.5 dBm
UNII-2e
12 dBm
UNII-3
13 dBm
ISM (Ch 165)
13 dBm
Source: FCC
This is the max Tx power
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
18
This is the TX Power
(to the antenna)
Power, Spatial Streams and Data Rates
 Do not think that multiple stream devices are always better
 They may have higher power, but also require higher SNR
e.g.: 4 streams gives you a higher throughput
at same SNR level, than 2 streams
BUT the 2 stream device reaches its max speed
at 24 dB SNR,
while the 4 stream device needs 30+ dB
Conclusion: at same distance from the AP,
multiple stream devices will operate faster than
single stream device, but each individual stream
is slower
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
How can you tell the AP Power level?
 WLC global level gives you the overall resulting power (this is what you care
about):
(Cisco Controller) >show advanced 802.11a txpower
…/…
AP Name
Channel
TxPower
-------------------------------- ---------- ------------AP702W
157
*1/8 (20 dBm)
AP2602
48
1/4 (14 dBm)
AP3702
(52,56)
*2/5 (12 dBm)
AP3602
(40,36)
*2/7 (12 dBm)
Allowed Power Levels
-----------------------[20/17/14/11/8/5/2/-1]
[14/11/8/5/5/5/5/5]
[15/12/9/6/3/3/3/3]
[14/12/10/8/5/-1/-4/-4]
AP is on 40 MHz channel
Power is dynamically assigned by WLC
Current level is 2 (12 dBm),
there are 7 levels
Allowed levels, 7 to 8 are the same,
so AP is configurable down to level 7
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
How can you tell the Client Power level?
 You can check, live the client power levels on the AP (useful to check symmetry
in AP to client and client to AP signal when building your cell edge):
This is on 5GHz radio, d0 is 2.4 GHz radio
2 client signals reported
AP7cad.74ff.36d2#debug dot11 dot11Radio 1 trace print rcv
*Jun
*Jun
1 04:11:43.663: D5B70D90 r 6
1 04:11:43.664: A2CEF918 r m15-2s
49/46/42/48 54- 0803 000 m010B85 477AAF m010B85 33E0 477AA0 l46
53/63/54/61 40- 8841 030 1A096F A36F20 m333300 76B0 q0 l100
Timestamp
Client used MCS 15 (2SS)
Client RSSI on each antenna
BRKEWN-2000
Client SNR
L+length of
rest of the frame
With WMM, shows the queue
without WMM, DCF queue index
Sequence number
Address 3
Frame type (follows 802.11 spec)
Frame duration
Receiver and transmitter
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
addresses (last 3 bytes)
Some Client Max EIRPs
Model
EIRP 2.4 GHz
Worst* EIRP 5 GHz
Iphone 5
14.6 dBm
10 dBm
Ipad 4
15.2 dBm
22.67 dBm
Samsung S3
14.9 dBm
10.18 dBm
Samsung S4
12.05 dBm
11.24 dBm
Samsung S5
13.4 dBm
10.61 dBm
HTC One
14.4 dBm
13.8 dBm
Nokia Lumia 1520
13.1 dBm
11.6 dBm
ASUS PCE-AC66
22 dBm
22.83 dBm
* EIRP varies with sub-band, displaying worst of all sub-bands
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
So, what is the right Power?
 In short: half your worst client max power
– E.g. you design for 5 GHz, worst client max is at 11 dBm, set your AP power to 8 dBm
 Otherwise, you get this:
 Which BYOD is the worst out there? No names, but 11 dBm is a good
assumption
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Power Levels and Client Density
– Reducing cell size by matching client
power and disabling low speed
decreases the user number per cell
– Each user has better throughput and
less risk of encountering interference
– Coverage holes might cause probing
devices to disturb the WLAN clients
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
What if I crank up my AP power a bit?
 Bad design example: Client @ 12 dBm, AP @20 dBm
Based on Rx AP signal, BYOD thinks 54 Mbps rate is okay
But client message is too weak,
and AP does not ACK
Retry @ 54
Again
and Again
Okay try 36
Wow, let’s drop to 12
Now I get an ACK
Start over…
Each message takes 8 times more to be transmitted
(including EIFS and retries)
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Cell Size - Protocols and Rates
Cell Throughput by Protocol
Protocol
Throughput (Mbps)
802.11b
7.2
802.11b/g mix (1 b client)
9.5
802.11g
22.5
802.11a
22.5
802.11n (HT20 1ss MCS7)
35
802.11n (HT20 2ss MCS15)
75*
802.11n (HT20 3ss MCS23)
110
802.11ac (VHT80 3SS MCS 9)
630**
• These are average throughputs, with one client close to the AP (high SNR/RSSI)
* Two spatial streams – note most PDA’s are SISO (MCS 7) 35 Mbps max
** You could have guessed that : 256-QAM max PHY is 1.3 Gbps, max throughput is
typically less than half of max PHY
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Cell Size and Data Rate
 Each cell’s useful radius is determined by the minimum allowed
data rate
1 Mbps DSSS
2 Mbps DSSS
5.5 Mbps DSSS
6 Mbps OFDM
9 Mbps OFDM
11 Mbps DSSS
12 Mbps OFDM
18 Mbps OFDM
24 Mbps OFDM
36 Mbps OFDM
48 Mbps OFDM
54 Mbps OFDM
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Protocol Support and Channel Utilization

Example low rate time waste

ACK frames are sent at the first mandatory speed below the speed at which the
frame was received

802.11a/n: SIFS = 16 microsec

802.11b/g/n: SIFS = 10 microsec

802.11a/g/n: slot = 9 microsec

802.11b: slot = 20 microsec

1 DIFS = 1 SIFS + 2 slots

OFDM physical header = 16 microsec (312 b at 6 Mb/s)

802.11 physical header = 192 microsec (192 b at 1 Mb/s)
 ACK sent at 24 Mb/s uses 4.67 microsec. airtime
 ACK sent at 1 Mb/s uses 304 microsec. Airtime
 Impact on throughput higher for shorter frames
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Protocol Support and Channel Utilization

Before: 8 SSIDs, all
rates allowed

After: 2 SSIDs, 802.11b
rates disabled
60% Before
5% After
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
36
Protocol Support and Channel Utilization

Disabling 802.11b in this network would:

Suppress 27% of frames (slow frames would be sent faster)

Decrease airtime consumption from 62% to 18 % if using 24 Mbps
(slow frames take much longer
to be sent than faster frames)

Reduce cell size:

Clients nearby would benefit from higher speeds

Clients far would not sick to the AP
DSS/CCK
Airtime
consumption
OFDM
Airtime
consumption
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Low rates impact Depends on frame size…
20000
18000
16000
14000
12000
10000
Time/
8000
μS
6000
4000
2000
0
Mb
ps
64 Byte
DSSS
CCK CCK
DSSS
OFDM
OFDM
512 Byte
1024 Byte
2048 Bytes
2
5.5 11
6
Time
consumption
per voice flow
at 1 Mb/s
Time
consumption
per voice flow
at 24 Mb/s
Time
consumption
per voice flow
at 54 Mb/s
G.711
(64 Kb/s)
102.4 ms
9.45 ms
6.49 ms
G.729
(8 Kb/s)
46.4 ms
6.27 ms
5.20 ms
G.726
(32 Kb/s)
70.4 ms
7.27 ms
5.64 ms
G.728
(16 Kb/s)
42.43 ms
4.72 ms
3.74 ms
128 Byte
256 Byte
1
Codec & Bit
Rate
12 24 36 48 54 130 300
Frame
Size/Bytes
Individual theoretical time consumption:
SLOT + DIFS + (voice packet + headers) x speed x (number of packets per second) + SIFS + ACK
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
And most BYODs know that
 Most BYODs take advantage of 802.11 blocks to group small
frames (even if they end up sending one frame at a time):
 Viber on Iphone 5S
 Viber on Samsung S5
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
What should be your Minimum rate?


You want to stop your cell in a place where the following happens:

Signal to your clients is still strong

Clients and overhead traffic still “reasonably fast”

Retries are low
54 Mbp
Beyond that point, clients should be able to get to another AP
if they want to. On the right:
BRKEWN-2000
6 Mbps
STA3

STA1 and STA2 hear each other -> less collisions

STA 1 and STA2 send @ 54 Mb/s -> short delays

STA3 is far from AP -> lower data rate (longer transmission delay),
higher PER and loss risks

STA3 does not hear STA1 and STA2 -> higher collision risk
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
40
STA1
STA2
What should be your Minimum signal level?
 Multiple measurements show a “sweet spot” by -67 dBm:
802.11n client still communicates at 72 Mbps (MCS 7)
Management/control frames still sent fast (24 Mbps)
But you start seeing devices (here the AP) dropping rate because signal starts to degrade
 What minimum configured data rate is that? Depends…
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
41
What should be your Minimum data rate?
 And BER is important, because more retries means more
chances that the frame will be dropped
 Your job is to limit frame drops to
1% or less to maintain 4.1 MOS
 At -67 dBm RSSI, SNR is
typically around 25 dB or more*

You can run any rate of 24 Mbps
and up, and still have good
frame success rate
* well, at least in ideal conditions… see next slides
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
42
Signal Attenuation
 Your hand position may make things worse
Object in Signal Path
Signal Attenuation
Through Object
Plasterboard wall
3 dB
Glass wall with metal frame
6 dB
Cinderblock wall
4 dB
Office window
3 dB
Metal door
6 dB
Metal door in brick wall
12 dB
Phone and body position
3 - 6 dB
Phone near field absorption
Up to 15 dB
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
There can be a 20 dB difference
between these photos
Let’s do the signal level math
-67 dBm
AP
AP
-67 – 20 = -87 dBm
Signal is too weak…
But you can roam to the other AP @ -67 dBm!
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
BTW, where do you put an antenna on a BYOD?
Head – not good
Iphone 5,
Antenna is at
bottom
Hand – not good
Samsung S5, antenna
is at bottom, behind button
HTC One, whole
back cover is metal and
antenna
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
45
Disable 802.11b but not 802.11n low rates!
 Each SSID will advertise at the
minimum mandatory data rate
 Disabled – not available to a client
 Supported – available to an associated client
 Mandatory – Client must support in order to
associate
 Lowest mandatory rate is beacon rate
 Highest mandatory rate is default Mcast rate
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
46
Disable 802.11b but not 802.11n low rates!
 Many BYODs rely on the beacon to validate that the
AP is still there (and sync their clock)
 Many BYODs also ignore AP instructions about
supported 802.11n rates (disable them, and your
client talks at a speed the AP will ignore)

In standard density environment, stop your cell @ -67 dBm.

When power is @ 11 or 14 dBm, this is about 12 Mbps
Everything below 12 Mbps is disabled
(but NOT 802.11n low rates)
First allowed rate (12 Mbps) is mandatory
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
47
Cell size strategies
 In HIGH density environment, also stop your
cell @ -67 dBm.
 Power is usually low, 14 dBm or lower
 Cells are smaller than in standard density
environment
 Roaming occurs faster
 Rate @ -67 dBm is more commonly 24 Mbps
 You want to allow your client to roam at that point
-> 24 Mbps is set to Mandatory (below 24 Mbps,
client does not hear the beacons and typically scans
to find alternate AP)
 You still want the client to communicate with the AP
while getting into panic scan
 Set lower rates (18, 12 Mbps) to Supported
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Disable Slow rates, and maybe fast rates!
 What about the other rates?
 For Voice, rates faster than 24 Mbps do not bring any clear advantages
Codec & Bit
Rate
Time
consumption
per voice flow
at 1 Mb/s
Time
consumption
per voice flow
at 24 Mb/s
Time consumption
per voice flow
at 54 Mb/s
G.711 (64 Kb/s)
102.4 ms
9.45 ms
6.49 ms
G.729 (8 Kb/s)
46.4 ms
6.27 ms
5.20 ms
G.726 (32 Kb/s)
70.4 ms
7.27 ms
5.64 ms
G.728 (16 Kb/s)
42.43 ms
4.72 ms
3.74 ms
Time consumption = SLOT + DIFS + (voice packet + headers) x speed x (number of packets per second) + SIFS + ACK
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Disable Slow rates, and maybe fast rates!
 Faster rates DO have an impact on rate shifting:
 200 byte frame @ 54 Mbps is sent in 3.7 μs
 200 byte frame @ 24 Mbps is sent in 8.3 μs
 Rate shifting from 54 Mbps to 24 Mbps wastes 1100 μs
(65 times longer to send the next frame), in ideal (no congestion) conditions
36 Mbps
24 Mbps
24 Mbps
36 Mbps
48 Mbps
54 Mbps
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
54 Mbps
Cisco Public
Disable Slow rates, but not 802.11n rates!
 Conclusion: disable low rates
 If your real time applications are Voice only,
disable rates higher than 24 Mbps, and set
channels to 20 MHz
 If your real time applications are Voice AND
Video, then you need higher rates
 In 5 GHz, set channels to 40 MHz… if your
clients support 40 MHz
 Leave all 802.11n / ac rates enabled
(if your clients support 802.11n and 802.11ac)
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
51
Cell Size - 20, 40, 80 MHz?
Cell Size – Depends on Power, Protocol and Rates
 With the explosion of 802.11ac, majority of devices are dual-band
 98% of devices observed are 802.11n
 5 GHz 802.11n devices are not always 40/80 MHz-able
120
100
99
80
62
60
38
40
20
0
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
53
5 GHz
0.1
60
40
71
55
45
5050
29
60
40
Cell Size – Depends on Power, Protocol and Rates
 What about YOUR network?
 If there is no network deployed yet, capture
traffic at different times and observe
 Example: large airport on US East Coast,
last month
12 captures of 10 minutes each at different
times / days, with wireshark – display rates
20 MHz rates
 In this network, enabling 40 MHz is a waste
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
54
40 MHz rates
AP Placement Strategy
Real Time AP Placement
 Coverage Areas
 A huge percentage of problems come from
incorrectly defined coverage areas
 Coverage areas: where the real time service
should be offered
 Typical errors: “not needed in the bathrooms”,
“not in the elevators”, “not in the stairs”,
“not in the outdoor smoking area”
 Talk to end-users. Think what they will need
and when
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
56
AP Placement Guidelines
 Mount APs so that antennas are vertical (we use vertical polarization)
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
AP Placement Guidelines
 Avoid metallic objects that can affect the signal to your clients
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
RF Design Matters
 Highly reflective environments
 Multipath distortion/fade is a consideration
 Legacy SISO technologies (802.11a/b/g)
are most prone
 802.11n improvements with MIMO
 Devices are susceptible
 Things that reflect RF
– Irregular metal surfaces
– Large glass enclosures/walls
– Lots of polished stone
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
62
AP Placement – Bad Examples
 Issue is sometimes in the environment
Ceiling is highly reflective
metallic mesh
AP behind ceiling
(yes they did that)
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
AP Placement – Bad Examples
 AP too high:
– Low rate to the ground
– Client signal too weak at the AP level
Nice… but you won’t cover the
jetway as soon as the door closes
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
> 20ft
AP Antenna Guidelines
 Use matching antennas
These are dual band plugs (2.4 and 5 GHz)
They require dual band antennas… they are labelled “a,b,c,d”
These are single band antennas…
are you going to get 2.4 or 5 GHz?
Antenna gain mismatch: AP
won’t know which to believe
Coverage pattern mismatch:
Are you covering through the
wall?
When RF cluelessness becomes art…
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Taking Care of the Roaming Path
Rates and Cell Overlap
 Cell overlap is designed so that when a VoWLAN device gets to the
–67 dBm area, it is already in good range of another access point.
 20-percent overlap between cells is recommended
 How much is that? Use the -75 dBm rule if you are not sure.
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Network Design
 Try to design small cells, with clever overlap…
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
70
Channel Plan and Data Rate
 Each cell’s useful radius is determined by the minimum allowed
data rate
RF
edge
24 Mbps OFDM
36 Mbps OFDM
48 Mbps OFDM
54 Mbps OFDM
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Client Signal Detection Optimization
 AP Rx sensitivity is MUCH higher than that of most BYODs… and improved over the
years beyond the 802.11 requirements
 E.g. 1242 Rx sensitivity at 1 Mbps is -96 dBm, with a typical SNR of 3 dB (so you need -96
dBm signal, with background noise not more than -99 dBm, to recognize and read a frame
sent at 1 Mbps). With the 3700 AP, that threshold is -101 dBm (with the same SNR
requirements), so we are 5 dB better
 We need more signal to read 802.11n than legacy protocols. If we disable legacy
protocols on our new APs, the problem would probably more or less go away by itself, but
the issue is that we keep the old protocols and at the same time build smaller cells to
benefit from the additional throughputs.
Rate
Recommended RSSI (3700AP)
Recommended SNR
1 Mbps
-91 dBm
1
6 Mbps
-87 dBm
2
MCS 0 (6.5 Mbps)
-82 dBm
1
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Sending modulation depends on Received RSSI and
SNR
 802.11 determines “minimum RX performance values” – RSSI based
 Vendors achieve “at least” these values
 Example: Cisco 3700e Rx performances:
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
74
RX-SOP – (Receive - Start of Packet) – What is it?
 Receiver Start of Packet Detection Threshold (RX-SOP) determines the Wi-Fi
signal level in dBm at which an AP radio will demodulate and decode a packet.
 The higher the level, the less sensitive the radio is and the smaller the receiver cell size
will be
 By reducing the cell size we can affect every thing from the distribution of clients
to our perception of channel utilization
 This is for High Density designs – and requires knowledge of the behavior you
want to support
 A client needs to have someplace to go if you ignore it on the current cell
WARNING – This setting is a brick wall – if you set it above where your clients
are being heard – they will no longer be heard. Really.
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
75
RX-SOP configuration
 Settings High, Medium, Low, Auto
 Auto is default behavior, and leaves
RX-SOP function linked to CCA
threshold for automatic adjustment
 Most networks can support a LOW
setting and see improvement
 This affects all packets seen at the
receiver
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
76
Radiation Pattern and Roaming Buffer
 When users are expected to roam while communicating, make sure
their BYOD can detect neighboring APs BEFORE roaming
AP signal drops slowly
AP signal drops fast
User does not have much space/time
to find the next AP
Directional vs omnidirectional antenna
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Floor
VoWiFi Rate Shifting and Transition AP
 At “A” the phone is connected to AP 1
A
1
2
B
C
 At “B” the phone has AP 2 in the neighbor
list, AP 3 has not yet been scanned due
to the RF shadow caused by the elevator
bank
 At “C” the phone needs to roam, but AP 2
is the only AP in the neighbor list
3
 The phone then needs to rescan and
connect to AP 3
– 200 B frame @ 54 Mbps is sent in 3.7 μs
– 200 B frame @ 24 Mbps is sent in 8.3 μs
– Rate shifting from 54 Mbps to 24 Mbps can
waste 1100 μs
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
79
VoWiFi Rate Shifting and Transition AP
 At point A the phone is connected to
AP 1
1
A
B 2
 At point B the phone has AP 2 in the
neighbor list as it was able to scan it
while moving down the hall
C
 At point C the phone needs to roam
and successfully selects AP 2
 The phone has sufficient time to scan
for AP 3 ahead of time
3
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
80
Controller Redundancy and Roaming Paths
 Design expected roaming paths and make sure all APs connect to the same
controller, and overlap allows for next AP discovery
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Avoid Ping Pong Zones
Client stays here
Ping-pong effect occurs when a wireless client is at the edge of two cells
and hops between them.
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Avoid Ping Pong Zones
Ping Pong zone recipe:
Set overlap along pacing path
Let user head force the roam
“Pacing back and forth” zone
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
Cisco ClientLink Technology
• Advanced Beam Forming Technology
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
84
BandSelect – Test Before Full Deployment
 Caveat – Possible Increased Roaming Delay
No Delay
2.4G band
5G band
Some Delay
(1.5s)
Possible Delay
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
90
Client and AP 802.11 Optimizations
Cisco VideoStream - How Does it Work?
5
6
7
2
3
1.
2.
3.
4.
5.
BRKEWN-2000
Client sends IGMP
join
WLC intercepts
IGMP join
WLC sends AP
RRC request
AP sends RRC
response
WLC forwards join
request
4
8
6.
.
© 2015 Cisco and/or its affiliates. All rights reserved.
1
7.
9
8.
9.
Cisco Public
98
.
Multicast source
sends IGMP join
response
Multicast stream
sent
WLC forwards
multicast stream to
AP
AP converts stream
to unicast and
delivers to client
Thinking BANDWIDTH in Terms of Directions
 In a standard cell, 70% of traffic is downstream (from AP to client)
 30% is upstream
 We can definitely control downstream, especially as 802.11n/ac stations are
necessarily WMM
 Can we control the upstream? Not directly, but we may have an indirect way of
controlling it…
I decide, alone,
when to send (thank
you CSMA/CA)
Don’t
send!
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
102
10
2
Bandwidth Control
 You can also control upstream and downstream bandwidth consumption:
 Per QoS Profile (Gold etc.)
 Per SSID
 Per user type (guest etc)
 Per device type
 Per individual user
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
103
If you have Several Traffic Types to Target: Use Application
Visibility and Control
Don’t Allow
Voice
Video
Best-Effort
Background
Client Traffic
Rate Limiting
Identify Applications using NBAR2
Control Application Behavior
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
107
Last Words - Troubleshooting
Troubleshooting Tools
 Wireless Captures, RF Analysis, Configuration Analysis
 Wireless sniffer
 Omnipeek/Wireshark (multichannel, for roaming issues)
 Mac with OS X 10.6 and above, Windows 7 with Netmon 3.4
 AP in Sniffer mode
 L1 analysis: SpectrumExpert
 WLCCA (WLC Configuration Analyzer) – TAC support
 NCS / Cisco Prime Infrastructure for Historical view
and « Client Troubleshooting tool »
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
112
Support Community
https://supportforums.cisco.com/community/5771/wireless-ip-voice-and-video
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
113
Continue Your Education
• Demos in the Cisco Campus
• Walk-in Self-Paced Labs
• Table Topics
• Meet the Engineer 1:1 meetings
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
115
Call to Action
• Visit the World of Solutions for
– Cisco Campus
– Walk in Labs
– Technical Solution Clinics
• Meet the Engineer
• Lunch time Table Topics
• DevNet zone related labs and sessions
• Recommended Reading: for reading material and further resources for this
session, please visit www.pearson-books.com/CLMilan 2015
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
116
Complete Your Online Session Evaluation
• Please complete your online session
evaluations after each session.
Complete 4 session evaluations
& the Overall Conference Evaluation
(available from Thursday)
to receive your Cisco Live T-shirt.
• All surveys can be completed via
the Cisco Live Mobile App or the
Communication Stations
BRKEWN-2000
© 2015 Cisco and/or its affiliates. All rights reserved.
Cisco Public
117