Resource aware placement of IoT application modules in Fog-Cloud Computing Paradigm

TLDR: The result of this work can serve as a Micro-benchmark in studies/research related with IoT and Fog Computing, and can be used for Quality of Service (QoS) and Service Level Objective benchmarking for IoT applications.

About: This article is published in Immunotechnology.The article was published on 2017-05-08 and is currently open access. It has received 288 citations till now. The article focuses on the topics: Edge computing & Cloud computing.

Figures

Fig. 2. Directed Acyclic Graph (DAG) of the application deployed, with relevant tuples indicating the values used for simulation of the algorithm. The number of modules as well as sensors/actuators are synchronous to the needs of a typical IoT application, and may vary as per the use case and application size; the modules can be conveniently linked to various logics to convey the various stages of application processing, as are the ones conveyed by the color pair encoding here.

Fig. 5. Huge effect of efficient module mapping on end-to-end latency

TABLE II EXPERIMENTAL NETWORK CONFIGURATIONS

TABLE I EXPERIMENTAL NETWORK CONFIGURATIONS

Fig. 3. One of the network topologies used for deployment iteration. The simulation has been varied over three such topologies with varied workloads, but essentially the same standardized network structure.

TABLE III EXPERIMENTAL NETWORK CONFIGURATIONS

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Resource aware placement of iot application modules in fog-cloud computing paradigm" ?

In this paper, the authors present a Module Mapping Algorithm for efficient utilization of resources in the network infrastructure by efficiently deploying Application Modules in Fog-Cloud Infrastructure for IoT based applications. The result of this work can serve as a Micro-benchmark in studies/research related with IoT and Fog Computing, and can be used for Quality of Service ( QoS ) and Service Level Objective benchmarking for IoT applications. 

Q2. What are the future works mentioned in the paper "Resource aware placement of iot application modules in fog-cloud computing paradigm" ?

In future work, the authors plan to include further dynamic characteristics of the DAG once the application has been deployed. The authors also plan to look into the scheduling policies of resources on Fog Devices after the application deployment. 

Q3. What are the devices of primeconsideration to us here in Fog-Cloud architecture?

Amongst the available networking devices in the network infrastructure, the devices of primeconsideration to us here in Fog-Cloud architecture are the gateways that connect the devices in the bottom most layer (Tier 1/IoT layer) to the Internet. 

Q4. What is the way to describe a distributed computing environment?

Distributed computing environment calls for distributed components, which would give better results with multi-component applications, for which DDF is one of the best approaches available. 

Q5. What are the characteristics of the application that has been designed and modeled in the paper?

The application that has been designed and modeled in the paper is motivated from standardized realistic IoT scenarios like health care [21] and latency-critical gaming [22]. 

Q6. What is the function that is used to locate the module?

The Control Loop of the algorithm (for loop/line5) runs for all the modules of the application that need to be placed, and calls the function LOWERBOUND (Algorithm 2) in each iteration, which searches for the eligible network node meeting the requirement of the module (constraint specified in equation 12). 

Q7. What are the main aspects of the work done by the authors?

Other aspects include the programming model approach to be followed for development of Fog based IoT applications [6] to its scalability for large scale geographical distribution [7], [8], as well as a context aware realtime data analytics platform for the Fog [9]. 

Q8. How did the proposed approach affect the network usage?

4. Staggering decrease in network usage via proposed approachThere was also a huge effect of efficient module mapping on end-to-end latency, with highly favourable results towards Fog-Cloud placement as per the designated approach as shown in Fig. 

Q9. What is the function that is used to sort the network nodes and modules?

Taking the set of network nodes N and set of application modules V as input, it first sorts the network nodes and modules in ascending order as per their capacity and requirement respectively. 

Q10. What is the definition of an edge in a DAG?

The set of all edges in the DAG of an application is denoted by E.E = {ei | ei = 〈vi, vj〉} ∀vi, vj ∈ V (10)There are two types of edges possible in a DAG— periodic and event based. 

Q11. What is the definition of the SenseProcess-Actuate Model?

The application works on the SenseProcess-Actuate Model, where the information collected by sensors is emitted as data streams, which is processed and acted upon by application modules running on Fog and Cloud layer, and the resultant commands (or Outputs) are sent to the actuators. 

Q12. What is the impact of the proposed approach on the latency in IoT applications?

The authors present the impact of an evolving paradigm that is Fog Computing towards solving the problem of latency in time critical IoT applications, while also accounting for the pressure-2024681012Fog-Cloud Placement Traditional Cloud Placement Fog-Cloud Placement Traditional Cloud Placement Fog-Cloud Placement Traditional Cloud PlacementConfig-1(2FG/Workload-4Devices) Config-2(4FG/Workload-8Devices) Config-3(6FG/Workload-12 Devices) 

Q13. What is the function that maps the network nodes?

Algorithm 1 ModuleMapping Algorithm: Fog-Cloud Placement Input : Set of Network nodes N and Application modules V Output : Mapping of modules on to network nodes1: function MODULEMAP(NetworkNode nodes[], AppModule modules[]) 2: Sort(nodes[]),Sort(modules[]); . in ascending order 3: Map < NetworkNode,AppModule[ ] > moduleMap; . 

Q14. What is the main idea of the paper?

The authors outlined the key characteristics that impact the performance of such IoT applications, and have classified and kept into account the static part while increasing the network efficiency and broadening the scope of such applications. 

Q15. What is the definition of an application operator?

In context of analytics applications such as stream analytics or event based analytics, these modules are usually termed as application operators.