Avaliação de Fog Computing para suporte a sistemas de localização interna de equipamentos hospitalares

Abstract

The traditional method of hospital management is difficult to allocate resources to the entire area in a unified way, which leads to a great redundancy of manpower and waste of resources in this aspect. The Internet of Things (IoT),emerged as a communication paradigm that involves organized systems of computing devices, mechanical and digital devices, objects, living or non-living things that are capable of transferring information without involving interaction human-human or human-machine. In this way, the smart hospital based on IoT systems can provide much more efficient and refined management for people and regions in the hospital area. Hospital automation can be offered by one of two alternatives, either through centralized computing or distributed computing, add that centralized systems are not wanted mainly because when receiving a large amount of data they would have two large drawbacks: high latency and high network usage. However, edge devices generally have less computing power than cloud datacenters, which forces the strategy of inserting an intermediate layer into the system, giving rise to the concept of Fog Computing. Faced with this problem and the results observed through systematic mapping, we decided to propose and analyze an architecture based on Fog Computing to assist in the management of hospital resources, supporting the location of mobile equipment (such as a defibrillator, cardiopulmonary resuscitator, echocardiogram, among others) internally. We simulate two application models (display after storage and display before storage) in three scenarios (Edge, Fog and Cloud Computing), progressively increasing the network load with the increase of devices transmitting data every 1s, 2s, 5s and 10 seconds. Using the iFogSim simulator, it was observed that Model 1 (display after storage), in scenario 2 (Cloud Computing) obtained the best result for application latency. In Model 2 (display before storage), the shortest time presented was scenario 3 (Fog Computing). This work also observed that changing the processing capacity in MIPS of the equipment located in the Fog Computing did not cause significant latency gains.