A timing-accurate modeling and simulation environment for networked embedded systems
Proceedings of the 40th annual Design Automation Conference
Detailed OFDM modeling in network simulation of mobile ad hoc networks
Proceedings of the eighteenth workshop on Parallel and distributed simulation
Scalable Testbed for Next-Generation Wireless Networking Technologies
TRIDENTCOM '05 Proceedings of the First International Conference on Testbeds and Research Infrastructures for the DEvelopment of NeTworks and COMmunities
Genie: supporting the model driven development of reflective, component-based adaptive systems
Proceedings of the 30th international conference on Software engineering
| Hi-index | 0.00 |
This paper describes a co-simulation implementation of ns-2 and SPHERE. These simulators complement each other by providing support for a group of layers of the OSI communication model. In particular, the SPHERE simulator is responsible for the simulation of lower layers (link layer and physical layer), while the ns-2 simulator provides support for the simulation of the upper layers (network layer, transport layer and application layer). Each simulator has an action plane within the process of co-simulation. The aim of the co-simulation is to simulate heterogeneous handovers between LTE, WiMAX, WiFi and/or UMTS. In order to allow for media independent handovers, our implementation is based on an extension of the base IEEE 802.21 standard provided by NIST. This extension consists on the support of IEEE 802.21 entities on the network side, and on the integration between the IEEE 802.21 framework and the radio access technologies. In the case of LTE integration, this one needed to be provided through the SPHERE framework. From the same simulator, new events such as "link down" and "link up" needed to be implemented. The paper shows how the IEEE 802.21 was successfully used as bases for the integration of the ns-2 and SPHERE simulators. As expected when using co-simulation for different domains, e. g., network and physical layer, the simulation-time is largely dependent on the lowest layer in the stack. This layer usually has the smallest time-resolution and hence will require more processing resources per simulated second.