Wi-Fi Capacity Analysis for

802.11ac and 802.11n:

Theory & Practice

By Timo Vanhatupa, Ph.D.

Senior Research Scientist at Ekahau

Contents
Introduction Why capacity matters with Wi-Fi ...................................... 3
Part 1: Modeling the IEEE 802.11 WLAN Capacity ................................... 5
IEEE 802.11 MAC operation ......................................................... 6
IEEE 802.11 Physical Layer ......................................................... 7
How to estimate 802.11 capacity ................................................. 9
Example ..................................................................................10
Part 2: Capacity Analysis in Ekahau Site Survey.....................................13
The network modeling layer .......................................................13
The capacity estimation layer .....................................................14
The presentation layer ...............................................................15
Planning networks with capacity requirements ..............................18
References: .......................................................................................19
About the Author ...............................................................................20
Contact .............................................................................................21

Copyright 2015 Ekahau, Inc. All rights reserved.

Introduction Why capacity matters with

Wi-Fi
Just a few years ago, enterprise Wi-Fi networks were planned with coverage and perhaps
interference in mind. The number of wireless devices and the amount of transmitted data
was small and users could tolerate an imperfect Wi-Fi network. Today, 802.11 networks
should be designed primarily for capacity. The key drivers of improved network capacity and
reliability are:

More devices are joining the network

End users and business-critical applications rely on a Wi-Fi network that works 24/7

The amount of data transmitted wirelessly has signficantly increased thanks to:

Voice calls made over Wi-Fi

HD video streaming over Wi-Fi

File access and storage in the cloud

Thanks to the Bring Your Own Device (BYOD) trend, personal devices are allowed on
enterprise networks

When modeling the capacity of a single Access Point (AP) and client, the capacity limits of
each transmission technology are well known. Unfortunaltey, it is not a one-AP, single client
world. As a Wi-Fi network engineer, managing multiple access points on different floors
serving different types of client devices from different generations running various
applications, is a reality. Network management becomes incredibly complex unless
administrators have the right tools and some basic Wi-Fi knowledge.
There are two scenarios in which the estimation of network capacity is important:

When designing a completely new network, it is beneficial to evaluate the

coverage and capacity of the network without actually installing it. Installing either
too few or too many APs will lower the capacity of the network. Too few APs will
cause low signal strengths and adversely impact data rates whereas too many APs
will result in interference.

Capacity estimates can be useful when assessing installed networks as well.

However, measuring network capacity can be very difficult in practice. This is because

Copyright 2015 Ekahau, Inc. All rights reserved.

all normal network usage causes interference which skews measurements and a
complete analysis would require controlling all the traffic in the network, which is
practically impossible in large networks. However,
advanced tools such as Ekahau Site SurveyTM (ESS)
allow quick and comprehensive analysis of network

The rest of this whitepaper is organized as follows:

1. Understanding capacity: An overview to the
operation of IEEE 802.11 Wireless Local Area

As a Wi-Fi network
engineer, managing
multiple access points on
different floors serving
different types of client
devices from different
generations running various
applications, is a reality.

Network (WLAN) [1] and how capacity can be

estimated in 802.11 Wi-Fi networks. Understanding network operations is crucial for
understanding how ESS estimates the capacity of the network.
2. How Wi-Fi capacity can be seen and optimized: A high level overview to the
Ekahau Site Survey tool that allows you to design and maintain enterprise Wi-Fi
networks.

Copyright 2015 Ekahau, Inc. All rights reserved.

Part 1: Modeling the IEEE 802.11 WLAN

Capacity
The IEEE 802.11 WLAN consists of a set of APs and clients as presented in Figure 1. Each AP
and the clients associated to it form a Basic Service Set (BSS). Multiple BSS can be
connected via a Distribution System (DS) and usually they are connected to the Internet
[1].

Figure 1 Topology of an IEEE 802.11 Wireless Local Area Network

The modern IEEE 802.11 WLANs operate in the 2.4 GHz and 5 GHz frequency bands. The
basic channel is 20 MHz wide but 802.11n and 802.11ac allow usage of wider channels by
bonding several channels together. The number of available 20 MHz channels in the 2.4 GHz
and 5 GHz frequency bands are 11-13 and 19-25 respectively, depending on the regulatory
domain. All devices that are within the transmission range and share the same, or
overlapping, channel also share the channel transmission capacity.
IEEE 802.11 specifies operations in two layers of the Open Systems Interconnection (OSI)
model, namely in the data-link and physical layers as presented in Figure 2. The data-link
layer contains the Medium Access Control (MAC) protocol, which is responsible for
controlling transmissions in the BSS. The physical layer is responsible for modulating and
transmitting data. Each transmitted data packet goes through each layer and is
encapsulated with the layer specific framing. Finally the corresponding layer in the receiver

Copyright 2015 Ekahau, Inc. All rights reserved.

decodes the frame. The framing in each layer is needed for the operation of the layer but
this overhead decreases the effective capacity of the network.

Figure 2. Simplified view of OSI model and packet encapsulation.

IEEE 802.11 MAC operation

The 802.11 MAC protocol uses a Carrier Sense Multiple Access with Collision Avoidance
(CSMA/CA) method [1] illustrated in Figure 3. The basic principle in channel access is that
before sending, each station must listen to the channel and send only when the channel is
sensed free. Detection of collisions is difficult in a wireless environment and the following
additional method is used to avoid collisions. When the channel is sensed as free and
available, first the station waits for the duration of an InterFrame Space (IFS). The time
between the IFS and the next transmitted frame is called a contention window, which is
divided into transmission slots. After the IFS period has passed, the station starts a backoff
counter. The initial value for this counter is selected randomly between zero and contention
window minus one slots. When the counter reaches zero and the channel is still idle, the
station starts transmitting. If another station starts to transmit before that the counter is
stopped and the counting continues when the channel has been idle again for longer than
the IFS. When the counter reaches zero, the transmission can start.
In the MAC layer the transmitted data packet is called a MAC Protocol Data Unit (MPDU). It
starts with the MAC header followed by a MAC Service Data Unit (MSDU), which is the
actual payload data of the packet. The payload is followed by a Frame Check Sequence
(FCS), which is used for detecting transmission errors in the packet. Each correctly

Copyright 2015 Ekahau, Inc. All rights reserved.

transmitted packet is acknowledged with an ACK packet and retransmitted if no ACK packet
is received. Each transmitted MAC layer packet is wrapped inside a physical layer packet.

Figure 3 CSMA/CA access cycle

Several additional methods are used to avoid collisions. These are a dynamically changing
contention window, as well as Request to Send (RTS), Clear to Send (CTS) packets [2]. RTS
and CTS packets are omitted from the figure for simplicity.
The channel access method described above enables controlled usage of the transmission
channel simultaneously by several stations. However, as shown in the figure, it also requires
a considerable overhead before each sent frame and this causes problems, especially with
high loads and small packet sizes [3]. Distributed channel access also makes capacity
estimation very difficult.
Packet aggregation was introduced with 802.11n to reduce overhead in the MAC layer [4].
It allows transmission of several network layer packets in a single MAC layer packet. Packet
aggregation increases packet delay because the transmitter must wait until all packets are
collected before a MAC frame can be transmitted. Thus, the method has its limits with delay
sensitive applications, such as voice and video.

IEEE 802.11 Physical Layer

The IEEE 802.11 standard defines several physical layers. The latest new physical layer is
called VHT (Very High Throughput) and it is defined in the IEEE 802.11ac standard
amendment [5]. The support of VHT Physical layer was introduced to ESS in version 6.0.

Copyright 2015 Ekahau, Inc. All rights reserved.

The physical layer packet starts with a preamble, which is different for each physical layer.
The preamble is needed to synchronize transmission and control physical layer operation.
Figure 4 shows examples of the physical layer frame formats for High Rate / Direct
Sequence Spread Spectrum (HR/DSSS) defined in 802.11b, High Throughput (HT) mixed
format defined in 802.11n and VHT format defined in 802.11ac. As can be seen in the
picture, the overhead required in the preamble has decreased from 192 s to a minimum of
40 s.

Figure 4.Example physical layer frame formats with 802.11b, n and ac.

After the physical layer preamble, the data itself is transmitted using a selected Modulation
and Coding Scheme (MCS). Each physical layer defines a set of possible MCSs depending on
the quality of the signal between the transmitter and receiver. The MCSs differ on the
amount of error coding and density of data. This makes different MCSs suitable for different
signal conditions.
In the latest physical layers (802.11n and 802.11ac), data is transmitted using several
possible channel bandwidths. The normal channel bandwidth in 802.11 WLANs is 20 MHz,
whereas the 802.11n supports a 40 MHz bandwidth and the 802.11ac can utilize also 80
MHz and 160 MHz bandwidths. This allows transmission of much more data during the same
time period. The 802.11n and 802.11ac also support spatial multiplexing by using several
antennas to transmit multiple streams between the transmitter and receiver simultaneously.

Copyright 2015 Ekahau, Inc. All rights reserved.

This technology is called Multiple-Input and Multiple-Output (MIMO). If interference

conditions allow, the efficiency can be increased by reducing the guard interval between the
symbols from 800 ns to 400 ns. Using the short guard interval, the symbol length becomes
3.6 s. These parameters are presented in Figure 5, which provides a simple equation to
calculate the physical layer data rate. As can be seen, the physical layer data rate may
differ significantly based on the properties of the AP and the client device as well as the
signal conditions.

Figure 5. Key parameters of determining the IEEE 802.11 data rate.

How to estimate 802.11 capacity

The transmission time of an end user packet can be calculated based on the factors
described in the previous sections. For IEEE 802.11ac, the standard amendment [5]
provides equation 1 for the purpose, which is further described in the following section.
When the transmission time of the signal packet is known, the channel utilization of a data
stream can be calculated based on packet transmission time and packet rate. Although this
may sound simple, there are still several key parameters that must be estimated by ESS:

Packet error rate

Number of streams

Channel bandwidth

Additional overhead caused from multiple clients sharing the same channel

The amount of background traffic caused, for example, by beacons, probes and
management applications

Copyright 2015 Ekahau, Inc. All rights reserved.

Example
This section presents an example of the VHT physical layer transmission time calculation
with a full HD video stream application. The application is assumed to have a 1000 byte
packet size and a 10 Mbps transmission rate. Equations 1-4 can be used to calculate the
transmission time of a single data packet. The equations and the included variables are
described in detail in the 802.11ac standard amendment chapter 22.4.3 [5].

= _ + + + _ + +

(1)

_ = _ + _

(2)

_ = _ +

(3)

8 _ + +

(4)

First, the number of symbols ( ) needed to transmit the packet is calculated using
equation 4. The equation has the significant parameter , which defines the number of
data bits per symbol that are encoded with the selected modulation and channel bandwidth.
For example with a QAM-64 modulation, 5/6 coding rate, 40 MHz bandwidth and 2 spatial
streams the standard defines that is 1080. When a MAC frame is added to the data
packet, the DATA_LENGHT is 1034 bytes. Assuming that a Space-Time Block Code (STBC) is
not used, the equation becomes
8 1034+16+61

= 1

11080

= 8.

When a short guard interval is used, the transmission time can be calculated using equation
1. Consequently, in the physical layer, the transmission of a single 1000 byte packet takes

= 8 + 8 + 4 + 8 + 4 + 2 4 + 4 + 4

3.6 8
= 76
4

Copyright 2015 Ekahau, Inc. All rights reserved.

10

Similarly, the transmission of the 14 byte MAC layer ACK frame takes 48 s. Most of the
transmission time is consumed by the overhead, which causes that the transmission of the
ACK packet takes almost the same time as the transmission of the data packet. When we
add the MAC layer overhead including DIFS, estimated backoff time, and SIFS, the
transmission time of a single packet increases to approximately 200 s, depending on the
active contention window.
With a 10 Mbps rate, the application must send 1250 packets during one second. This is
further increased to 1375 packets/s by assuming that 10 % of the packets are
retransmitted on the MAC layer, which is not uncommon. Based on the single packet
transmission time and the amount of packets transmitted, the channel utilization caused by
the single HD video stream in the example is 27.5 %. High level presentation how the
transmission time is consumed is presented in Figure 6 a).

Figure 6. Transmission time distribution in the example.

MAC efficiency can be calculated by first calculating the theoretical maximum end-user
throughput (with 100 % channel utilization) and then dividing the maximum throughput by
the physical layer data rate as presented in [6]. In the example, the 10 Mbps throughput
caused 27.5 % channel utilization so the maximum throughput is 10 Mbps/0.275=36 Mbps.
The physical layer rate is defined in the standard for the selected modulation and coding

Copyright 2015 Ekahau, Inc. All rights reserved.

11

rate. In the example, the rate is 300 Mbps. Consequently, the efficiency is only 12 %, which
is mainly caused by the lack of frame aggregation. The efficiency of 802.11ac MAC depends
heavily on how many packets can be aggregated into a single frame in order to limit
transmission overheads. To compare efficiency with frame-aggregation, the Figure 6 b)
shows transmission time distribution when 10 packets are aggregated into a single frame
using A-MPDU mechanism. This increases MAC efficiency to 49 %. Selecting right number of
packets aggregated into single frame is a key parameter in capacity estimation because
with latency sensitive applications such as voice and video, aggregating a large number of
packets may cause unacceptable delay.

Copyright 2015 Ekahau, Inc. All rights reserved.

12

Part 2: Capacity Analysis in Ekahau Site

Survey
Ekahau Site Survey (ESS) displays capacity metrics on the floor plan as visualizations. The
ESS capacity visualizations are based on detailed modeling of capacity related parameters.
Capacity modeling can be divided into three distinct layers as shown in Figure 7. These are
the network modeling layer, the capacity estimation layer and the presentation layer.

Figure 7. Capacity modeling layers in Ekahau Site Survey

The network modeling layer

For measured networks, the key information for capacity modeling is taken from beacons,
probe responses and their information elements. This includes operational parameters such
as utilized channels, channel bandwidths, and the number of utilized streams.
For simulated networks, ESS models the WLAN capacity starting from a detailed definition of
APs, and the 3D environment. AP antennas as well as buildings, floors and walls are
modeled in 3D. Together, this allows a detailed estimation of the client signal in different
locations in the network using commonly available signal propagation models. Based on this

Copyright 2015 Ekahau, Inc. All rights reserved.

13

configuration, ESS also estimates the client interference based on 3D signal propagation
modeling. Interference estimation can also be done for measured networks when noise
information is not available from the utilized WLAN network interface card.

The capacity estimation layer

Client devices and client applications are modeled similarly for both simulated and measured
networks. The differences between various client devices are modeled by allowing the user
to select his own devices, including laptops, tablets, and mobile phones from different
vendors. Figure 8 shows how devices joining to the network can be configured.

Figure 8.Configuring devices that will be joining to network

The set of client devices in the network can be configured by the user and each application
is configured in detail by specifying packet types, lengths, packet transmission rates and
protocols as shown in Figure 9. This allows detailed modeling of different applications
including voice and video streams. Interactive applications such as voice and video are
sensitive to transmission delay. The maximum possible delay within each packet type is

Copyright 2015 Ekahau, Inc. All rights reserved.

14

taken into account based on the configured access category (IEEE 802.11e). This has an
effect on how many packets are aggregated into a single MAC layer packet.

Figure 9.Configuring how the average user behaves will improve the capacity estimation
accuracy

Data rates are calculated based on client device configurations and received signal strengths
and the Signal-to-Noise Ratios (SNR). The capacity of the network is modeled based on
selected client devices and their applications. For each packet stream on the network, ESS
calculates transmission time estimates based on the specifications in the IEEE 802.11
standard. This includes modeling of the 802.11 MAC protocol packet transmission process as
well as timings and modulations used in physical layers. Channel usage of the network is
modeled based on CSMA/CA method [1] used in IEEE 802.11 together with roaming logics,
which select APs each client device associates to. As a result, the algorithm gives the
channel usage estimates for each AP in the network.

The presentation layer

The capacity health visualization presented in Figure 10 shows the user if the capacity
requirement caused by client configurations are fulfilled in the network. The capacity

Copyright 2015 Ekahau, Inc. All rights reserved.

15

requirement is fulfilled when channel usage in the network is below 100% for all APs. The
algorithm also takes into account that part of the capacity is consumed by network
management applications including WLAN beacons, probes, association requests and so on.
The number of associated clients for each AP as well as the number of each type of
application is also compared to configured limits. This allows modeling of situations where
too many client devices are associated with a single AP although the data transmission
would be minimal.

Figure 10 Capacity health visualization shows if the network contains too much traffic, too
many voice calls or associations

Copyright 2015 Ekahau, Inc. All rights reserved.

16

The maximum throughput visualization presents the maximum possible throughput

available for a selected client device in a particular location when there is no other traffic in
the network. Because in a real network the capacity is shared between all users, this
throughput cannot be normally reached. However, the visualization gives valuable
information about the performance limits of the network.
The associated clients visualization presented in Figure 11 shows how clients are distributed
between APs based on the roaming logic used by ESS. The roaming logic selects the best
APs based on their data rate and SNR. It allows the user to see if some of the APs in the
network would be overloaded based on commonly used roaming logics.

Figure 11. Associated clients visualization shows how clients are distributed between APs.

Copyright 2015 Ekahau, Inc. All rights reserved.

17

The accuracy of the capacity estimation of the network relies heavily on how exact the
configuration of network applications and devices the user creates. The 802.11 specific
channel usage algorithms are modeled on a level which is detailed enough to take into
account the differences of capacity requirements of different applications (data, voice,
video) as well as the specifics of WLAN physical layers defined in the standard amendments
(802.11g, n) The latest physical layer integrated into ESS is IEEE 802.11ac. It includes
support for wide channel bandwidths (80, 160 MHz). The modeling is never a perfect image
of the real world but gives estimates that are reasonably accurate. This allows network
planning and modeling of the network capacity without actually installing the network.

Planning networks with capacity requirements

Network planning with ESS can be done in two ways:
1. Manually, by placing simulated APs on the map by hand or
2. Using Auto-Planner, where optimal AP placement and channel assignment are
suggested based on user requirements for capacity and coverage.

Both methods support capacity requirements and allow analysis using capacity
visualizations. If manual planning is chosen, the capacity visualizations can be used to
monitor when the capacity requirements are fulfilled. With automatic planning, the AutoPlanner selects the number of APs in such a way that the capacity requirements are fulfilled.
Sometimes the capacity requirements cannot be fulfilled with the selected AP type and
configuration. For example, there can be simply too many devices are streaming HD video
in a small area. In such cases, ESS selects the number of APs, which fulfill the requirements
as closely as possible. When the capacity requirements cannot be fulfilled, the user should
explore the following possibilities to increase the capacity:
1. Increase the amount of channels used
2. Use dual-band 802.11n/ac APs
3. Move as many clients to 5GHz as possible
4. If possible, get rid of legacy client devices
5. Lower the AP transmission power to decrease interference.
6. Use 802.11n and 802.11ac specific features, such as wider channels, to increase
capacity

Copyright 2015 Ekahau, Inc. All rights reserved.

18

References:
[1] IEEE Std 802.11-2012
[2] J. Hui and M. Devetsikiotis, A Unified Model for the Performance Analysis of IEEE
802.11e EDCA, IEEE Transactions on Communications, vol. 53, no. 9, pp. 14981510,
2005.
[3] D. Hole and F. Tobagi, Capacity of an IEEE 802.11b Wireless LAN Supporting VoIP, in
Proceedings of the IEEE International Conference on Communications (ICC04), June 2024,
2004, pp. 196201.
[4] J. Kolap, S. Krishnan, N. Shaha, Frame Aggregation Mechanism For High-Throughput
802.11N WLANs, International Journal of Wireless & Mobile Networks (IJWMN) Vol. 4, No.
3, June 2012
[5] IEEE P802.11ac/D5.0, January 2013
[6] E. H. Ong, et. al., IEEE 802.11ac: Enhancements for Very High Throughput WLANs,
IEEE International Symbosium on Personal, Indoor and Mobile Radio Communications
(PIMRC), 2011

Copyright 2015 Ekahau, Inc. All rights reserved.

19

About the Author

Dr. Vanhatupa received M.Sc. 1998 in Software Systems, and
Ph.D. 2008 in Digital and Computer Systems from Tampere
University of Technology. He worked several years as a software
engineer in a company developing network management
solutions. His Ph.D. research topics included performance
management and network planning in Wi-Fi networks. He joined
Ekahau in 2007 and currently he acts as a senior research
scientist in Wi-Fi tools team. His work includes developing
algorithms for network planning and analysis for Ekahau Site
Survey.

Copyright 2015 Ekahau, Inc. All rights reserved.

20

Contact
Ekahau East Coast
(Head Quarters)
1851 Alexander Bell Drive
Suite 105
Reston, VA 20191
Tel: 1-866-4EKAHAU
Fax: 1-703-860-2028
[email protected]

Ekahau Europe
(Sales, R&D, Product Management)
Hiilikatu 3
00180 Helsinki, Finland
Tel: +358-20-743 5910
Fax: +358-20-743 5919
[email protected]

Ekahau APAC
(Regional office)
B38, Tower 8, Imperial Cullinan
10 Hoi Fai Road, Tai Kok Tsui
Hong Kong
Tel: +852 9227 8406
[email protected]

Copyright 2015 Ekahau, Inc. All rights reserved.

21