Urgent Computing: Computations for Decision Support in Critical Situations (UC) Session 1

Time and Date: 15:45 - 17:25 on 12th June 2017

Room: HG E 33.3

Chair: Alexander Boukhanovsky

521 Simulation of emergency care for patients with ACS in Saint Petersburg for ambulance decision making [abstract]
Abstract: One of the stages of emergency medical care in case of ACS (if there are medical conditions for surgical intervention) is directly linked to the time between the first contact with the patient and the and inflating the balloon in the coronary artery (in a medical institution). Time of the operation start a medical facility depends on the time of patient delivery to hospital, as well as on the waiting time in the queue in the institution. This paper describes a development of ambulance model for obtaining aggregate estimation of these 2 periods of time. The estimation time is obtained by means of described in the article the decision support system (DSS) in the ambulance service. Unlike modern navigation systems DSS takes into account ambulance vehicle behavior (the ability to exit into oncoming traffic) and availability of free operation rooms. With the help of the described simulation model of the ambulance service we carried out the time distribution analysis (between the first contact with the patient and surgical intervention in case of ACS) in St. Petersburg, Russia. Simulation scenario uses real data on the work of ambulance service in the city.
Ivan Derevitskiy, Evgeniy Krotov, Daniil Voloshin, Alexey Yakovlev, Sergey V. Kovalchuk and Vladislav Karbovskii
362 Smart levee monitoring and flood decision support system: reference architecture and urgent computing management [abstract]
Abstract: Real-time disaster management and decision support systems rely on complex deadline-driven simulations and require advanced middleware services to ensure that the requested deadlines are met. In this paper, we propose a reference architecture of an integrated smart levee monitoring and flood decision support system, focusing on the decision support workflow and urgent computing management. The architecture is implemented in project ISMOP in which controlled flooding experiments are conducted using a full-scale experimental smart levee. While the system operating in the ISMOP project monitors a~test levee, it is designed to be scalable to large-scale flood scenarios.
Bartosz Balis, Tomasz BartyƄski, Marian Bubak, Daniel Harezlak, Marek Kasztelnik, Maciej Malawski, Piotr Nowakowski, Maciej Pawlik and Bartosz Wilk
565 Firemap: Dynamic Data-Driven Predictive Wildfire Modeling and Visualization Environment [abstract]
Abstract: Wildfires are destructive fires over the wildland that can wipe out large areas of vegetation and infrastructure. Such fires are hard to control and manage as they can change directions almost instantly, driven by changing environmental conditions. Effective response to such events require ability to monitor and predict the behavior of the fire as fast as they change. The WIFIRE project builds an end-to-end cyberinfrastructure for real-time and data-driven simulation, prediction, and visualization of wildfire behavior. One goal of WIFIRE is to provide the tools to predict a more accurate rate of a spreading wildfire. To this end, WIFIRE has developed interfaces for ingesting and visualizing high-density sensor networks to improve fire and weather predictions, and has created a data model for wildfire resources including sensed and archived data, sensors, satellites, cameras, modeling tools, workflows, and social information including Twitter feeds for wildfire research and response. This paper presents WIFIRE's Firemap web platform to make these geospatial data and products accessible. Through a web browser, Firemap enables geospatial information visualization and a unified access to geospatial workflows using Kepler. Using GIS capabilities combined with scalable big data integration and processing, Firemap enables simple execution of the model with options for running ensembles by taking the information uncertainty into account. The results are easily viewable, sharable, repeatable, and can be animated as a time series.
Daniel Crawl, Jessica Block, Kai Lin and Ilkay Altintas
593 Performance-aware scheduling of streaming applications using genetic algorithm [abstract]
Abstract: The main objective of Decision Support Systems is detection of critical states and response on them in time. Such systems can be based on constant monitoring of continuously incoming data. Stream processing is carried out on the basis of computing infrastructure and specialized frameworks such as Apache Storm, Flink, Spark Streaming. However, to provide the necessary system performance at high load incoming data, additional data processing mechanisms are required. In particular, the efficient scheduling of streaming applications plays an important role in the data stream processing. Therefore, this paper is devoted to investigation of genetic algorithm to improve the performance of data stream processing system. The proposed genetic algorithm is developed and integrated into Apache Storm platform, and its efficiency is compared with heuristic algorithm for scheduling of Storm streaming applications.
Pavel Smirnov, Mikhail Melnik and Denis Nasonov
365 Towards an operational database for real-time environmental monitoring and early warning systems [abstract]
Abstract: Real-time environmental monitoring, early warning and decision support systems (EMEWD) require advanced management of operational data, i.e. recent sensor data required for the assessment of the current situation. In this paper, we evaluate the suitability of four data models and corresponding database technologies -- Document database MongoDB, relational database PostgreSQL, Dictionary data server Redis, and time series database InfluxDB -- to serve as an operational database for EMEWD systems. For each of the evaluated databases, we design alternative data models to represent time series data, and experimentally evaluate them. Then we perform comparative performance evaluation of all databases, each using the best model. We designed performance tests reflecting realistic conditions, using mixed workloads (simultaneous read and write operations) and queries typical for smart levee monitoring and flood decision support system. Overall the results of the experiments allow us to answer interesting questions, such as: (1) how to best implement time series in a~given data model? (2) What are the reasonable limits of the operational database sizes? (3) What are the performance limits for different databases?
Bartosz Balis, Marian Bubak, Daniel Harezlak, Piotr Nowakowski, Maciej Pawlik and Bartosz Wilk