![](bg3.png)
The definition of transmission time is based on a
producer consumer model as shown in Figure 1. It is
the time duration from the beginning of the handing
over of the first user data byte of a packet at the
communication interface in the test producer, up to
the handing over of the last user data byte of the
same packet at the communication interface at the
test consumer. It may be necessary to transmit
several telegrams between the communication
modules e.g. for acknowledgment. Furthermore,
network elements such as base stations may be
involved in the communication producing additional
delays. All these delays are covered by the
transmission time.
In order to assess the reliability the retry rate is
analysed. It is defined as the number of WLAN
packets that are necessary to transmit the content of
one generated test packet.
The system under test implements a widely used
radio technology - Wireless LAN and its application
interface is very common - Ethernet. Therefore,
standard measurement equipment can be used to
implement the above mentioned test methodology.
Ethernet
Tester
Access Point
(Consumer)
Client
(Producer)
Ethernet Ethernet
Power
Splitter
Dynamic
Attenuation
Antenna Cabel
Power
Splitter
56 dB
WLAN Packet
Analyser
3 dB 3 dB
20 dB
Figure 2: Test architecture.
The test architecture is shown in Figure 2. The
devices under test (WLAN client and WLAN access
point) are put into separate radio proof enclosures.
The test boxes achieve an attenuation of 90 dB
against the outer environment. The radio interfaces
of access point and client are connected by an
antenna cable. It is led via a power splitter and a
dynamically changeable attenuator. This approach
excludes interferences to the radio communication.
The dynamic attenuation has been configured to
a constant value of 57 dB. Together with the
attenuation of splitter and antenna cable the total
attenuation is 60 dB as specified for the test cases.
Besides the relevance of this value for automation
applications it has been shown that in this way the
transmit signal is low enough to avoid an
overmodulation and high enough to be unaltered.
The second output of the power splitter is connected
to another static attenuation of 56 dB in order to
connect a WLAN packet analyser. A second power
splitter is used in order to provide the signals to two
channels of the WLAN packet analyser. This
increases the reliability of the packet monitoring.
The total attenuation between client and WLAN
analyser is 62 dB. The total attenuation between
access point and analyser is more than 133 dB. Thus
the analyser captures only the packets transmitted by
the client.
The test packets are generated by an Ethernet
tester which transfers the data to the WLAN client.
As mentioned before the user data length is 64
octets. The generated packets are compliant to
PROFINET-IO telegrams. The advantage is the
specific frame type and content which simplifies the
identification of packets during analysis.
The packets transferred by the WLAN client are
monitored by WLAN analyser. If the packets are
successfully received by the WLAN access point
they are transferred to the Ethernet tester and are
monitored there also.
During analysis in a first step the user data
packets are filtered out of the packets monitored by
the WLAN analyser. Since every user data packet
can be identified by a unique payload it can be found
within the user data packets monitored by the
WLAN analyser. If a user data packet is listed more
than ones it means retries has been initiated by the
WLAN client because of missing
acknowledgements. This way the retry rate can be
calculated.
In addition a timestamp is included within the
user data. With help of this timestamp the Ethernet
tester is able to calculate the transmission time for
every successfully received packet.
4 TEST RESULTS
4.1 Transmission Delay
The measured transmission time consists of random
and constant components (see also Rauchhaupt, L.,
Krätzig, M., 2008.). Examples for constant
components are the frame spacing times or the
signal propagation delay on the wireless medium.
These components have the same values in every
measurement of a sample. The random nature of the
transmission time is being caused by latency of
application interface and implementation, by the
IMPACT OF DIFFERENT BIT RATES ON PERFORMANCE CHARACTERISTICS OF INDUSTRIAL WLAN
SOLUTIONS
389