An Architecture-Based Approach for Compute-Intensive Pervasive Systems in Dynamic Environments

Abstract

Distributed systems have continued to evolve and we note two important trends: the dramatically increasing level of dynamism in contemporary distributed systems and the convergence of mobile computing with cloud computing. The end result is that it is very difficult to achieve the required level of scalability and dependability in a systematic way when considering pervasive systems that are software-and compute-intensive and whose functionality is typically augmented by static cloud infrastructure resources. This work discusses relevant challenges and requirements for integrating cloud computing with pervasive systems operating in dynamic environments. We present a set of requirements using a holistic case study and describe a reference architecture to address these requirements.

A survey on fog computing for the Internet of Things

“…A distributed intelligent intermediate layer is required to add extra functionalities to the system, doing some processing of data when devices gather them and before sending them to the network and, eventually, to the cloud. There are some work proposing to move part of the resources towards the network edge, such as Cloudlet [33], Edge Computing [34], and Follow-Me Cloud [35]. Cloudlet efforts propose to create a cluster of servers near endpoints in order to satisfy real-time and location-awareness requirements.…”

“…Virtualization techniques allow the deployment of different IoT applications. Hypervisor is the most extended technology [29,33,123], because of its flexibility and the number of IoT applications that can be deployed using it, since the required operating system could be included in the VM. Nonetheless, for environments in which the fog nodes have fewer computing capabilities or the kind of applications to deploy are known, the containers technology presents additional advantages [123].…”

“…Normally, the data storage approaches implemented follows a centralized model, storing the data in the fog node. Nevertheless, different approaches [33,93,124] also implements distributed information storage. This improves the mobility of the users along the network and the storage capacity, but increases the network overload.…”