Personalisation/Profiling

PDFPrintE-mail

We implemented the pro ling scenario with the above described approaches (with IPv6 and without IP) to evaluate the feasibility, energy consumption, memory footprint and latency of each one of them. The scenario consists of a pro ling system in a WSN which is able to identify people, take decisions according to their pro le and make the resources of the WSN available to the Web.

Upon the arrival of a person in the proximity of an agent a message is transmitted to the pro ling server through the gateway. The message contains the pro le ID of the person hence the pro le service decides whether to allow the access of the person in the room or not. In addition, it provides the agents and the node-actuators customized information based on the user's pro le. According to the pro le and the needs of each person, several actions take place such as switching on/o a light and turning on/o a fan. The interconnections and the communication pattern of the nodes of the scenario are shown in te figure below.

In a distributed version of the profi ling scenario the agent is designed to take decisions according to its inputs and drive the node-actuators directly. In this case it is propagating the data to the gateway so that they are accessible from the Web. In a centralized version of the profi ling scenario the agent is acting as a forwarder without taking any decisions or actions. Then it is the responsibility of the pro filing server connected to the gateway to run the necessary services and drive the node-actuators.

We implemented a use case of the pro filing scenario interconnecting the three diff erent components; agent, gateway and node-actuator according to the figure below The agent initiates a "transaction" requesting the pro le from the server and transmitting the appropriate message to the node-actuator. We evaluated the above mentioned metrics performing 100 transactions for 20 times. The nodes were placed in 3 meters away from each other in an office environment. The transmission power of the nodes was set to their maximum. The evaluation of the two approaches took place at the same with the same conditions, allowing us to make accurate comparisons.

Calendar

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