A Testbed for Emulation of MANETs in Hostile Scenarios Marco Pratesi #1 , Andrea Colarieti *2 , Fortunato Santucci #*3 , Andrea Di Cola ##4 , Sebastiano Schillaci **5 # University of L’Aquila (Italy) Dpt. of Electrical and Information Eng. 1 pratesi@ing.univaq.it * Center of Excellence DEWS, L’Aquila (Italy) 2 andrea.colarieti@westaquila.com 3 santucci@ing.univaq.it ## University of L’Aquila (Italy) 4 andreadicola@gmail.com ** Thales Communications, Chieti Scalo (Italy) 5 sebastiano.schillaci@it.thalesgroup.com Abstract—The present work focuses on emulation and valida- tion of protocol stacks for MANETs (mobile ad-hoc networks) that must support multimedia applications in a tactical envi- ronment. We considered a network consisting of some tens of mobile nodes, with randomly variable speeds in the order of some tens of km/h. Nodes are deployed on a scenario of a few km 2 and have a radio range of some hundreds of meters. Commonly used Internet-based applications (based on the TCP-IP paradigm) have to be supported; they offer a multimedia traffic and are associated with QoS requirements (e.g. bounded latency and low losses). We developed a testbed for emulation and experimental validation of such scenario with a major routing protocol in the MANET context, i.e. OLSR (Optimized Link-State Routing). I. I NTRODUCTION MANETs have emerged as attractive solutions for support- ing autonomous communications in many challenging envi- ronments, wherein their intrinsic features of self-configuration and adaptation are particularly appealing. In particular, there have been increasing interests in the context of emergency and tactical communications. Protocol design has been a research objective in recent years, and routing protocols have received much attention, with AODV [1] and OLSR [2], [3] representing two major alternatives derived as wise adaptations of their originating counterparts in fixed networks. While there is a significant amount of literature dealing with performance analysis of these protocols, performance assessment in a real and challenging scenario certainly requires a careful evaluation of achievable performance. There are few papers discussing trustworthiness of simulation results and diversity of results among different wireless network simulators. An interesting paper [4] compares simulation results of OPNET, NS2 and GloMoSim. The paper states that there is a scarcity of real ex- periments demonstrating correctness of wireless network simu- lators results. Tools like MNE (Mobile Network Emulator) can help testing and debugging protocols for real testbeds, but it is not clear if these tools are suitable for protocol performance evaluation [5]. While numerous simulation environments and models have been developed for mobile routing and network- ing analysis, there is a growing interest in development of real-low-cost, flexible wireless mobile test environments. The Naval Research Laboratory (NRL) developed Mobile Network Emulator (MNE), a test suite providing direct and indirect support for node motion control and modeling [6], [7]. The emulation environment has been implemented with widely available, off-the-shelf commercial hardware within a commer- cial open source environment. The tool allows experimenters, developers, and researchers to test actual mobile networking technology and scenarios under more controlled laboratory conditions with low cost and repeatability. In [8], [9], the authors proposed a Linux kernel–based switch that connects multiple hosts according to a controllable dynamic topology with a controllable bit error rate on the links. In the present work, we intend to report on our ongoing work in a research project that relates to the deployment of a MANET in a tactical scenario. We describe how the emulation environment we developed can supplement more abstract simulation studies and how emulation and simulation can be integrated for more detailed and carefully planned experiments, ensuring reliable results under a highly dynamic network topology. As a preliminary step, we represented and simulated in the Opnet Modeler environment [10] an ad-hoc network with fixed nodes, to calculate relevant performance metrics for routing protocols of interest in a simple scenario. In particular, routing traffic overhead, network throughput, and delay metrics have been considered for OLSR, AODV, and DSR. OLSR performs better in terms of delay, hence it is more suited to meet our application scenario, that involves delay-sensitive applications and QoS requirements. Subsequently, other simulations have been run for more complex scenarios and mobility and traffic models, and results confirmed that OLSR is a good choice for the environment at hand. Then, we focused on development of a testbed for emulation and experimental validation of the 978-1-4244-3584-5/09/$25.00  2009 IEEE ISWCS 2009 234