SEMU: A Framework of Simulation Environment for Wireless Sensor Networks with Co-simulation Model

Abstract

Abstract. This paper presents a framework of simulation environment (SEMU) which allows developers to understand the behavior of applications or protocols for a wireless sensor network (WSN) before deploying real nodes in a physical environment.For eliminating the gap between simulation and real deployment, SEMU has supported fast real code emulation by dynamic binary translation technique. SEMU also models the controlled environment as virtual operating system (Virtual OS) to coordinate the interactions of large number of nodes. In addition, we have proposed a co-simulation model to enhance the accuracy of pure software simulation.Then a further synchronization problem will be addressed and resolved by the co-simulation model. The evaluation results show SEMU is really a fast scalable WSN simulator with real code emulation.Index Terms: Simulator, wireless sensor networks, dynamic binary translation, hardware and software co-simulation IntroductionA wireless sensor network (WSN) is a network composed of a large number of sensor nodes, which are deployed in the environment. Recently, with the rapid development of WSNs, providing development tools such as simulation environment before deploying real nodes in physical environments is getting more important. A well simulation environment can help developers build their prototype models to know the interactions and the behavior of each node. In addition, most of WSN applications will deploy a large number of nodes in a simulation environment. However, the simulation speed depends on the simulation fidelity and scale. Therefore how to build up a fast scalable WSN simulation environment with the finegrained information is the main research problem in this paper.In this paper, a framework of simulation environment (SEMU) is presented. SEMU has a first version implementation and an extension model for hardware and software co-simulation. In order to extract the real behavior of each node, SEMU supports real applications to run on the virtual nodes. And, the new development trend shows us the powerful nodes [2], [4], [9] are also applied to WSNs.

Testing Wireless Sensor Networks with Hybrid Simulators

“…To ease study the behaviour of the specific application and protocol and help in understanding this behaviour before the actual deployment a framework of simulation environment (SEMU ) for WSN with co-simulation, hybrid, model was presented in [15]. The co-simulation model proposed in SEMU is similar to the three hybrid modes for WSN simulation as of that proposed for EmStar [6].…”

“…Then the event was distributed using the flooding protocol until all of the virtual nodes received it. It was reported that SEMU did run fast below 1250 virtual nodes [15]. When the number of virtual nodes is increased beyond 1250 it was reported that simulation time quickly increases due to the limited RAM that the simulation was carried on, which was 1.5 GB [15].…”

“…It was reported that SEMU did run fast below 1250 virtual nodes [15]. When the number of virtual nodes is increased beyond 1250 it was reported that simulation time quickly increases due to the limited RAM that the simulation was carried on, which was 1.5 GB [15]. Additionally, authors of [15] stated that scalability of SEMU is highly affected by the resource management of the Linux OS (i.e., the native OS).…”

Sensornet Checkpointing: Enabling Repeatability in Testbeds and Realism in Simulations

“…A number of sensor network simulators allow a mixture of simulated nodes and testbed nodes. This technique is often called hybrid simulation [12,15,23,24]. Although hybrid simulation is designed to increase the realism by running parts of the network in a testbed, repeatability is affected negatively.…”

“…To decrease the gap between simulation and testbed, hybrid simulation has been proposed [12,15,23,24]. Hybrid simulation contains both simulated and testbed sensor nodes.…”