1 RoboMap: Reliable Cooperative Mapping of Rugged Areas with Mobile Wireless Robots Nicolas Kourtellis Department of Computer Science and Engineering University of South Florida Tampa, Florida 33620 {nkourtel}@csee.usf.edu Abstract — The efficient delivery of files among mobile ad hoc nodes has always been a problem of optimization, based on many parameters that need to be taken into account. Consistency, communications availability, reliability on the data delivery, scalability of the network, bandwidth consumption, nodes characteristics and nodes failure modeling, are some of the factors that engage in today’s MANETs. In this paper, based on real-life robot scenarios, we tackle the data delivery problem, by trying to maximize the throughput of the ad hoc network, and at the same time achieve a high delivery ratio of files like local maps, from a group of Unmanned Ground Vehicles (UGV) and Unmanned Aerial Vehicles (UAV) to their base station. To support our approach, we simulate various scenarios involving these robots and from the experiments done, we derive critical values for the network parameters, such as maps sharing rate of updates, size of the network, communication scheme etc, so that network traffic is minimized along with maximization of the delivery of the files-maps to the base station of the robots. Index Terms — distributed data delivery, frequency of map files updating, average bandwidth consumption. I. INTRODUCTION n a Mobile Ad hoc Network (MANET), nodes have specific attributes, which distinguish them from normal stationary nodes. These special attributes can be classified into mobility and failure modeling. In particular, due to the mobility of each node, and the possible failures it can develop during its lifetime, the network topology may change rapidly and even unpredictably. To be more precise, the mobility of a node, offers freedom to move dynamically and arbitrarily in any direction, the node wishes to move to. These dynamic changes also alter the physical distances – links between the nodes, i.e. the topology of the network changes. That means, in the new position, the specific node potentially might not be able to communicate with the former neighborhood of nodes, so it needs to update it and establish a new neighborhood. On the other hand, we have the failure modeling. Depending on what is the role that the node is assigned to, different failure models apply, creating a complex system of factors that we need to take into account, so that the design of the system, leads to a robust and efficient configuration. Here, we intend to examine the possibility that a group of robots, of various kinds (terrestrial, aerial, water), can evolve into a MANET, of 100s of nodes, with all the advantages and disadvantages this conception could have. In order for this to be plausible, we need to study the data delivery models these specific nodes could apply, as well as potential behaviors in the mobility aspect of the network, and how all these can be blend efficiently into a reliable and working network. The interesting part of this effort is the nature of the robots: a robot has a more complicated – powerful system onboard, than a normal mobile node would have. In terms of CPU power, it has a powerful Pentium Processor, which in some cases is a Dual Core Processor, adding to the potential complicated computations it can perform, but also on the energy it will consume to do that. RAM and disc storage capabilities are usually close to a PC’s (GBs). Along with the computation platform, many hardware modules are installed, including sensors (GPS, Laser, Cameras, temperature etc), Network Cards and others. These features, force us, to separate a robot, from the typical Ad Hoc nodes we could encounter, and to classify it into a separate class of mobile ad hoc nodes. From here on we will address this specific type of Ad Hoc node as “robode” (robot + node), to emphasize the unique nature of this kind of powerful, complex and mobile nodes. II. MOTIVATION & PROBLEM DEFINITION A. Motivating Scenario To gain insight on the impending scenarios of the above idea, we next present a potential setup, which can appear in a group of robodes, and the possible set of interesting problems/questions that can result due to the robodes’ nature - characteristics. The setup is based on rear-life UGVs and UAVs, which implement the Distributed Field Robot Architecture (DFRA), which is further utilizing the JINI Architecture (based on Java). A DFRA description and applications can be found in [1] and [2], and also how the various modules/components/services cooperate, to achieve a higher cause/task. A region of operation (figure 1) is explored by a team of robodes with a shared task. The robodes can take a meaningful combination of roles from the following set:  Drone (forwarded node) I