An efficient placement of sinks and SDN controller nodes for optimizing the design cost of industrial IoT systems. (13th June 2018)

• Record Type: Journal Article
• Authors: Faragardi, Hamid Reza; Vahabi, Maryam; Fotouhi, Hossein; Nolte, Thomas; Fahringer, Thomas
• Other Names: Heilig Leonard guestEditor.; Lalla‐Ruiz Eduardo guestEditor.; Voß Stefan guestEditor.; Buyya Rajkumar guestEditor.
• Is Part Of: Software, practice & experience. Volume 48:Number 10(2018)
• Page Start: 1893
• Page End: 1919
• Publication Date: 2018-06-13
• Subjects: industrial IoT -- node placement -- parallel ant colony optimization -- software‐defined networking -- wireless sensor networks; Computer software -- Periodicals; Computer programming -- Periodicals; Computer programs -- Periodicals; 005.3
• DOI: 10.1002/spe.2593
• Languages: English
• ISSNs: 0038-0644

Abstract

Summary: Recently, a growing trend has emerged toward using Internet of Things (IoT) in the context of industrial systems, which is referred to as industrial IoT. To deal with the time‐critical requirements of industrial applications, it is necessary to consider reliability and timeliness during the design of an industrial IoT system. Through the separation of the control plane and the data plane, software‐defined networking provides control units (controllers) coexisting with sink nodes, efficiently coping with network dynamics during run‐time. It is of paramount importance to select a proper number of these devices (i.e., software‐defined networking controllers and sink nodes) and locate them wisely in a network to reduce deployment cost. In this paper, we optimize the type and location of sinks and controllers in the network, subject to reliability and timeliness as the prominent performance requirements in time‐critical IoT systems through ensuring that each sensor node is covered by a certain number of sinks and controllers. We propose PACSA‐MSCP, an algorithm hybridizing a parallel version of the max‐min ant system with simulated annealing for multiple‐sink/controller placement. We evaluate the proposed algorithm through extensive experiments. The performance is compared against several well‐known methods, and it is shown that our approach outperforms those methods by lowering the total deployment cost by up to 19%. Moreover, the deviation from the optimal solution achieved by CPLEX is shown to be less than 2.7%.