Skip to Main Content
Majority of research in wireless ad-hoc networks is based on software tools simulating network environment under strictly controlled conditions, mainly due to its extreme cost, difficulty of adapting real-time topological changes in the environment and complexity of implementing a realistic testbed. In this paper, we present a testbed with real wireless task-oriented autonomous MANET based on VxWorks RTOS platform using Xilinx ML310 development boards with Virtex-II Pro FPGA devices and integrated gumstix/iRobot platform running embedded linux, as well as off-the-shelf laptops and desktops. As an example experiment, we consider the task of uniformly covering an unknown geographical terrain using autonomous MANET nodes with a limited communication range, which has many military missions such as search and rescue missions, surveillance tasks, locating and mapping chemical, and biological hazards. To achieve this objective, mobile nodes exchange one-hop neighbor information to decide their speed and directions without any central coordinator. Each node runs a genetic algorithm (GA) to select fitter speed and direction among an exponentially large number of choices for a better convergence toward a uniform distribution. The testbed experiments provide an effective research tool to demonstrate that our GA delivers acceptable network area coverage.