Aksit, M., Say, H., Eren, M. A. and de Camargo, V. V. "Data Fusion Analysis and Synthesis Framework for Improving Disaster Situation Awareness" Drones 7, no. 9: 565, 2023 https://doi.org/10.3390/drones7090565
02.09.23
Abstract
We define disaster situation awareness as the ability of the authorities to effectively and efficiently detect the negative effects of disasters so that aid operations can be planned and executed in a timely manner.
In general, the objective is twofold: to understand the type and magnitude of the damage caused and to determine the conditions and locations of the persons that need help. The concept of situation awarenesshas been studied extensively and applied in several areas such as disaster management. Unmanned aerial vehicles (UAVs) (also known as drones), for example, can be used to detect the effect of disasters. UAVs flying over a disaster area can take, compare and analyze images obtained before and after the disaster. Although UAVs can be considered adequate for some purposes, they may fall short of detecting certain facts such as the location of persons under rubble. Moreover, UAVs may take a considerable amount of time before their missions are completed. Nevertheless, during the last decade, we have observed the introduction of the concept of so-called advanced air mobility (AAM) and its implementation by unmanned aerial systems (UASs), where swarms of UAVs cooperate together for a common mission. All these newtechnologies help in creating more effective disaster management systems.
Additionally to the use of UAVs, disaster-related data can be gathered from various sources. For example, dedicated sensors can be attached to physical objects such as residences, bridges, and roads to detect if the corresponding structure is damaged. Base stations of mobile network providers may supply information about the location and use of mobile phones. Dedicated systems can be brought to the disaster area, such as microwave radars to determine if there are living persons under rubble. Official registration databases can be consulted to estimate the possible victims of the collapsed residences. For brevity, in this article, such devices and systems are abstracted as data sources. Emergency control centers aim at minimizing the negative effects of disasters by detecting and monitoring disaster situations and by carrying out aid operations accordingly. The success of these centers depends on the accuracy and timeliness of the gathered data.
There is a large set of possible data sources, each with its advantages and shortcomings. Data fusion is a promising technique to get the best out of multiple data sources. Due to high investment costs, before designing and implementing data fusion systems, it may be beneficial to estimate the optimal set of data sources that give the best combined effectiveness, cost, and timing values of sensor fusion. This requires a set of tools for the analysis and/or synthesis of heterogeneous data fusion systems before they are installed. Implementation alternatives of data sources and fusion techniques are considered out of the scope of this article.
Unfortunately, there are almost no publications devoted to analysis and synthesis of prospective data fusion configurations in disaster/earthquake management. Moreover, most proposed solutions are problem- and/or system-specific. What is needed is a framework which enables us to define models for a large category of data fusion alternatives. These models can for example be formed manually by an expert or be computed by synthesis algorithms. The framework should be extensible to introduce new models of geographical elements, data sources, and analysis and synthesis algorithms.
The contributions/novelties of this article are as follows. First, to detect disaster situations, a novel domain model is defined for representing the relevant data sources. In this article, earthquakes are chosen as an example of disaster. Second, to represent geographical areas, an object-oriented model is defined. In addition, dedicated queries are introduced to create models of data fusion associated with a selected set of geographical entities. Third, a model-based framework is introduced to specify the candidate data sources for a given geographical area. Fourth, with the help of the framework, the effect
of various alternatives of data fusion can be estimated. Last, to synthesize the optimal set of data sources within specified constraints, algorithms are defined. To the best of our knowledge, such a framework has not been proposed by the research community before.