A reliability evaluation of wireless sensor network simulator: Simulation vs. testbed
Guan, Zheng Yi (Alex)
View fulltext online
Permanent link to Research Bank record:http://hdl.handle.net/10652/1860
Wireless Sensor Network (WSN) is the new generation of sensor network which combines micro-electronic technology, embedded computing technology, distributed information processing technology, networking and wireless communication technologies. Wireless sensor network is composed of a huge number of sensor nodes that can cooperate to sense, collect and process the data. It has a series of features, such as the huge number of nodes, limited node hardware resource, power supply constraints, dynamic topology and self-organization, etc. The character of WSN allows it to be widely utilized in military surveillance, environmental observation and forecast, household appliance automation, medical care, space exploration, and other various commercial applications. Thus, these all contribute to a broad perspective of wireless sensor network. However, it also has brought a lot of challenges in the application areas – how to test and validate algorithms and protocols of wireless sensor network towards the large scale scenarios. Comparing simulators with test-bed, the accuracy and reliability of simulators becomes a strong issue. The purpose of this thesis is to present an effort took to examine the reliability of Castalia/OMNET++ simulator through simulating wireless sensor network. In this thesis, two test-bed experiments were setup by using two IRIS (Integrated Resource Information System) nodes and one MIB520 gateway to evaluate the simulator’s reliability. The results collected from IRIS test-bed experiments were compared with the results collected from the simulation experiments of the same scenarios on Castalia simulator. A lot of various parameters (traffic load, radio transmission power, and interference model) were considered in the simulation. The simulation results show that the Castalia simulator provide quite reliable results up to 2% over/under estimation of test-bed results. In addition, a simple tuning Castalia default value configurations can significantly increase accuracy.