
Chapter 5. Evaluation 67
gateway driver
devicemanager
device_changed
visualization &
namespace
simulation
node
protocol_out device_changed_ns
switch_device_nsswitch_deviceprotocol_in
10.4ms 1.8ms
45.8ms 7.8ms
2.2ms
6.9ms
Figure 5.2.: The latency of the complete system. Each time was measured 100 times and has
been averaged.
During normal operation, the maximum number of received protocols during a certain time is
limited to a small number (see figure 3.6 on page 34). This number can, however, be greatly
increased by using the simulation node, which makes it possible to send messages faster than
they can be processed. As ROS topics are buffered internally, the input buffers are filled, and the
node is not jammed. Nevertheless, the message order can be interchanged due to small delays
in communication. As a result, the database entries may be faulty, for example when a device
is turned on and off and the off message is received first, it will be saved as on. However, this
problem only occurs during simulation and not during real operation. Due to the low message
throughput, the data traffic between the nodes can be neglected and was therefore not measured.
5.2. Home and office scenario
In addition to the previous benchmarks, the system has been tested in two real environments in
order to evaluate its long-term behaviour. The data was recorded with ROS-bag, which enables
them to be analysed for patterns and to be played them back in different environments. Figure 5.3
on the following page shows the structure of the office and home scenario with its devices.
The following device types were used:
• Remote control: A universal IT remote control with four channels (A, B, C and D) for
controlling devices all over the room.
• Contact switches: Simple reed contacts for detecting whether a door or window is open
or closed.
• Passive infrared sensor (PIR): A PIR sensor for presence detection.
• Power plugs: Switchable power plugs for turning electrical appliances on or off.
• Wall switch: Switches as inputs to turn devices on or off.
Kommentare zu diesen Handbüchern