No abstract

Abstract. When a disaster strikes, response teams can nowadays rely on recent advances in technology. This approach improves the definition of a disaster management strategy. The use of autonomous systems during rescue operations allows, for example, to reach places that may be inaccessible or dangerous to human rescuers. In this context, both the design and the configuration of an autonomous system, including its embedded instruments (e.g. sensors), play a very important role in the overall outcome of the rescue mission. An incorrect configuration can lead to the acquisition of inaccurate or erroneous data and may result in incorrect information provided to rescuers. How can we ensure that the configuration of the autonomous systems is correct for a target mission? We propose to validate this configuration by testing the behaviour of the autonomous systems and their equipment in a virtual environment. To do this, system, sensors, space environment (geometry, etc.), prevailing conditions at the intervention site (weather, etc.) and mission scenario must be modelled in a 3D simulation system. The results of these simulations allow to apply in real time the modifications required to better adapt the configuration to the objectives of the mission. These simulations must be performed prior to the deployment of rescue teams to speed the development of a rescue management strategy. In this contribution, we propose a protocol to enhance an existing simulation environment to make it adapt to support disaster management. Then, we validate it through a case study in which we show the approach to correctly configure a LIDAR for a realistic mission. Such simulations allowed us to quantitatively configure the parameters of the LIDAR mounted on an existing disaster management rover, in order to keep the energy consumption limited while guaranteeing a correct functioning of the system. Resuming, the expected results are: (i) the assessment of the suitability of system for the mission, (ii) the choice of the quantitative features which characterize such equipment, (iii) the expectation of mission success and (iv) the probability which the system survives and completes the mission.

In modern cloud data centers, reconfigurable devices can be directly connected to a data center's network. This configuration enables FPGAs to be rented for acceleration of dataintensive workloads. In this context, novel scheduling solutions are needed to maximize the utilization (profitability) of FPGAs, e.g., reduce latency and resource fragmentation. Algorithms that schedule groups of tasks (clusters, packs), rather than individual tasks (list scheduling), well match the functioning of FPGAs. Here, groups of tasks that execute together are interposed by hardware reconfigurations. In this paper, we propose a heuristic based on a novel method for grouping tasks. These are gathered around a high-latency task that hides the latency of remaining tasks within the same group. We evaluated our solution on a benchmark of almost 30000 random workloads, synthesized from realistic designs (i.e., topology, resource occupancy). For this testbench, on average, our heuristic produces optimum makespan solutions in 71.3% of the cases. It produces solutions for moderately constrained systems (i.e., the deadline falls within 10% of the optimum makespan) in 88.1% of the cases.