Mobile ad hoc networks allow for the spontaneous formation of communication networks without dedicated infrastructure. Ad hoc networks are not yet ready for large-scale deployment, because several unsolved research challenges persist. Evaluation methods such as analytical modeling, simulation, emulation, and real world experiments aid in addressing these challenges. There is a strong need for tools to support the task of modeling and evaluation to allow for protocol validation, performance analysis, or proofof-concept implementations. The choosing of appropriate tools is a time-consuming process, which is often unnecessarily repeated, due to limited knowledge-transfer. We contribute an extensive survey covering real world and emulation testbeds to simplify the choice of appropriate research tools and methodologies in the domain of mobile ad hoc networks. In particular, we identify the key attributes of the aforementioned classes of testbeds and thoroughly discuss the state-of-the-art in literature to form a comprehensive classification of available testbeds.

No abstract

Abstract. We investigate the interaction of mobile robots, relying on information provided by heterogeneous sensor nodes, to accomplish a mission. Cooperative, adaptive and responsive monitoring in Mixed-Mode Environments (MMEs) raises the need for multi-disciplinary research initiatives. To date, such research initiatives are limited since each discipline focusses on its domain specific simulation or testbed environment. Existing evaluation environments do not respect the interdependencies occurring in MMEs. As a consequence, holistic validation for development, debugging, and performance analysis requires an evaluation tool incorporating multi-disciplinary demands. In the context of MMEs, we discuss existing solutions and highlight the synergetic benefits of a common evaluation tool. Based on this analysis we present the concept of the MM-ulator : a novel architecture for an evaluation tool incorporating the necessary diversity for multi-agent hard-/software-in-the-loop simulation in a modular and scalable way.

Future buildings and environments are envisioned to provide ambient intelligence, adapting to a user's preferences based on information about his context and status. Smart heterogeneous sensor networks are well suited data sources for such environments, because they allow for dynamic adaptation to newly added sensor types and novel tasks. Realistic verification of protocols and algorithms for smart networks poses special constraints on testbeds, necessitating support for heterogeneous platforms and mobility in the network. These distinct requirements limit the usability of many known testbeds of purely homogeneous nature.In this paper, we determine a minimum set of premises for smart heterogeneous sensor network testbeds and evaluate existing architectures with respect to these requirements. We then present our tubicle node platform, an integrated sensor network node providing inherent support for heterogeneity and fulfilling the determined set of requirements in their entirety. A set of twenty tubicles forms the basis for our TWiNS.KOM testbed. Specifically designed for heterogeneity, the architecture allows rapid validation of smart sensor network algorithms and quick experimental setup.

No abstract