Article Preview
TopIntroduction
In the last couple of years along with the emergence of Internet of Things (IoT), the conventional network computing paradigm has undergone through a series of changes. From a report given by CISCO on future consumption of network bandwidth, it has been estimated that by 2020 the number of internet connected things will be up to 50 billion (Cisco (2012)), which has been depicted in Figure 1. The whole existing unconnected network is going to be converted into a large connected network and the entire internetworking era of IoT is moving to be transformed into Internet of Everything (IoE)(Preden et al. (2015)).The network computing paradigm is evolving from simple Internet Protocol to 5G technology, which is illustrated in Figure 2. Computing in IoT environment with a number of data generating devices is producing an enormous amount of structured and unstructured data (Buyya and Dastjerdi (2016)). But the present cloud platform is not meant for handling such a volume, variety and bandwidth of data generated from these applications, which are counted in Exabyte. Cloud computing (CC) can provide scalable storage and processing resources to cater to the need of IoT applications. But the other problem is to do processing on all these data, it requires high network bandwidth to carry all these data to the cloud server end. Which may be a major concern for some delay-sensitive and emergency applications like health monitoring system, defense application etc. To overcome all these problems Fog computing (FC) is devised as a solution, Shi and Dustdar (2016) that uses the physical computing device resources, connected nearer to IoT sensors for local data spacing and to do some initial computations. The name ’Fog Computing’ is given by Salvatore Stolfo (StolfoShachtman (2012)) which is an extended computing edition of CC. It includes major features like computing, facilitating networking, storage assistance and infrastructure as a strength for end-user computing paradigm. Apart from these, it also supports latency constraints for the specific systems. CC provides all these services on the basis of a pay per use model, but, fails to handle a large amount of data generated from the real-time processing of IoT devices. All the edge devices like network switches, bridges, routers and access point, are now capable of processing some data and are generating huge online data. CC fails to handle all these real-time data by maintaining the network latency and communication delay (A.T.Thien et al. (2016), A. TFC Mahmud et al. (2018)).
Figure 1. Future estimation of number of connected devices by 2020, Source: CISCO
Fog computing is defined as the distributed paradigm which candeliver cloud-identical services/facilities at the edge of the network layer (Dastjerdiet al.(2016)). This definition is also provided by CISCO. Some authors extended this definition by adding some more characteristics like heterogeneity, ubiquity, and decentralization into it and described it as a collection of ubiquitous, heterogeneous, decentralized devices which work in cooperation with the help of core network (Rahmani et al.(2018)). In past few years, comprehensive studies and surveys are done which provides an extensive review of the existing literature. Redowan Mahmud and Rajkumar Buyya had given a review on the challenges present in FC and provided a detailed taxonomy of FC basing on the challenges. Their focus is mainly based on the challenges in FC (Mahmud et al.(2018)). A. T. Tran and R.C. Palacios (A.T.Tran & R.C. Palacios (2016)) had given a report on the application areas with some proposed architectures and protocols used in different domains of FC. F.A.Kraemer et al. (F.A.Kraemer (2017)) have presented a review on the use of FC in healthcare. They have emphasized on different deployment scenarios, placement of fog computing-based task.
Figure 2. Evolution of network computing paradigm
In contrast to the pre-existing work, the key focus of the present paper is on surveying the existing architecture and services used so far in the field of healthcare including the market trends and future demands. The study has been done in the following three dimensions considering the up to date literature.