The CTI test-bed

PDFPrintE-mail

The CTI test-bed. Apart from the already existing smart room, CTI testbed has been scaled out to 21 motes deployed over 7 offices at the premises of our building. The motes are organized in a mesh topology and communicate with the main server of the testbed via multihop data propagation. We would like to note that several of these additional motes are running Contiki OS; therefore the testbed consists of sensor motes running both TinyOS and Contiki. The testbed has also been upgraded from CoAP-03 to CoAP-12 while additional virtual resources have been implemented (i.e. average/min/max temperature and humidity) that combine readings from multiple sensors in order to provide a more abstract piece of information.

The CTI Demo Room at Patras. As part of the CTI test-bed, a demo smart/green room is developed at the PROKAT building of the CS Department at the University of Patras. In our demo-room we will demonstrate selected green/smart building scenarios. To be more specific, by using the data sensed the sensors will control (via actuators) the various devices (lights, air-condition, heater, appliances, etc.) in order to improve the energy efficiency in the building and improve the comfort levels of the users, e.g. turn off the light/air-conditioning when people walk out of the room, open the curtains when there is enough sun light outdoors etc. For that reason, we will use actuators able to be controlled by wires or wirelessly such as: heater electro-valve, air-condition controller, blinds controller, window controller, switch for electrical appliances and light controller (dimmer).
For the purposes of developing a demo room, we have installed the electromechanical infrastructure that is necessary for enabling the control of the indoor (lights and dimmer) and the outdoor light level (motor, curtains/blinds). Furthermore, we have implemented a control panel which is the hardware interface between the wireless sensor network and the electromechanical infrastructure of the demo room and enables the wireless network to control the demo room.

This demo room will be (partially) open to FIRE users who will be able to connect to it via a web interface and test it.


Comparison to the state of the art protocols. Using the extended CTI testbed we have conducted research on our energy balanced RPL (EB-RPL) that was initially evaluated using the SenseWall testbed. This new set of experiments gave us the opportunity to evaluate EB-RPL under real operating conditions in a building environment (this includes obstacles such as walls, moving people, interference from other networks like WiFi and cellular). The experiments ran over a period of two months for each routing protocol. EB-RPL virtually partitions the network into five sectors (as shown in the Figure above) and each node of a sector either transmits directly to the sink or to a node in the next sector. As far as energy balance property is concerned, the main findings show that EB-RPL is more efficient than RPL, which tends to overuse nodes closer to the Sink, resulting to early network disconnections. This is due to the fact that nodes have limited transmission range and all traffic has to be served through the small number of nodes that lie closer to the sink. We note that RPL, even if it is a tree-like protocol, has an energy balancing mechanism, by supporting data propagation via alternative paths for each node to the sink. The next Figure depicts the energy dissipation of the two protocols. The upper part of the Figure shows the values of energy dissipation of each node by ID. The lower part shows the energy map of the network according to the virtual sector topology. Darker nodes are those with lower residual energy, whereas the brighter ones are those with more energy. Note that the energy dissipation was measured by the voltage difference of each mote’s battery supply. It is evident that the RPL-EBP protocol brings off the energy balance property, since there is a color uniformity over the network.

In order to achieve increased energy balance in the network, EB-RPL has to spend more energy. Nodes may sometimes transmit their data directly to the sink, by increasing their transmission range online, thus spending more energy on transmissions but achieving lower delivery latency. Long range transmissions, which are absent from RPL, require higher energy supplies, The next Figure depicts average energy consumption and data delivery rate for the two protocols.

Balance of the energy saving vs end-user comfort. We have conducted research on user comfort in buildings. We have thoroughly studied corresponding bibliography on the Sick Building Syndrome and the Air Quality Monitoring. We have also identified ways of quantifying the comfort an end-user receives while inside a building. For instance, the Predicted Mean Vote (PMV) averages human comfort over a large group considering six key factors, i.e., air temperature, radiant temperature, humidity, air velocity, activity, and clothing level. This metric is evaluated using the following formula:

PMV = (-8:6479 + 0:2431 C) + (0:3442 - 0:0073 C)Tair

Where Tair corresponds to the air temperature and C is a constant factor that depends on the rest five factors mentioned above.

Energy saving percentage. The smart room has been equipped with two types of power meters in order to monitor energy consumption for the entire smart room as well as per electrical device inside the room. One C11 ABB power meter has been installed on the main line that powers the entire room while four Efergy ecotouch power meters monitor lights, the air-conditioning unit, the ventilation and four computer monitors. The ABB meter is able of monitoring the active power, the voltage, the current and the power factor with a total accuracy of 1% in terms of actual power consumption. Measurements are obtained in the form of electrical pulse (where each pulse denotes a 10Wh consumption) through a Solid State Relay switch. Data are propagated to the server via a TelosB mote, using the installed IPv6 network. The ecotouch power meters by Efergy are of less accuracy as they only measure apparent power, not including reactive power. Their basic use is to provide an estimation of the energy consumption and to inform the user on his energy profile rather than acting as a precise measuring tool. However, in conjunction with the ABB power we are able to obtain measurements for each device with an error in the range of 1% to 10%.

For a period of four weeks we monitored the energy consumption of the room. Then, based on the results, we came up with sophisticated scenarios that combine sensor measurements (temperature, air humidity, air quality), automations (air-conditioning, ventilation, lights, curtains) and sophisticated algorithms (in-door localization algorithms, user identification mechanisms) towards better comfort/energy trade-offs.

Calendar

Last month August 2017 Next month
S M T W T F S
week 31 1 2 3 4 5
week 32 6 7 8 9 10 11 12
week 33 13 14 15 16 17 18 19
week 34 20 21 22 23 24 25 26
week 35 27 28 29 30 31