Effective Load Balancing Strategy (ELBS) for Real-Time Fog Computing Environment Using Fuzzy and Probabilistic Neural Networks

• Published in: Journal of network and systems management Vol. 27; no. 4; pp. 883 - 929
• Main Authors: Talaat, Fatma M., Ali, Shereen H., Saleh, Ahmed I., Ali, Hesham A.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.10.2019; Springer Nature B.V
• Subjects: Algorithms; Caching; Cloud computing; Communications Engineering; Computer Communication Networks; Computer Science; Computer Systems Organization and Communication Networks; Computing time; Failure rates; Fuzzy logic; Fuzzy sets; Fuzzy systems; Health care; Information Systems and Communication Service; Load balancing; Networks; Neural networks; Operations Research/Decision Theory; Parameters; Real time; Search algorithms; Servers; Standardization; Strategy; Traffic signs; Fog computing; Average Turnaround Time; Cloud computing; Priority; Healthcare system; Failure Rate; Load balancing; Real time task scheduling; Quality of Service
• ISSN: 1064-7570; 1573-7705
• DOI: 10.1007/s10922-019-09490-3

Fog computing (FC) is an extension of cloud computing, however, it utilizes the resources close to the edge of the network. FC is a valuable choice to support real time applications such as healthcare, industrial systems, and intelligent traffic signs. However, Fog is a new emerging computing paradigm and still needs standardization in many issues especially in load balancing. This paper presents a new Effective Load Balancing Strategy (ELBS) for FC environment, which is suitable for Healthcare applications. ELBS tries to achieve effective load balancing in Fog environment via real-time scheduling as well as caching algorithms. It introduces several rules to accomplish reliable interconnections among fog servers. Moreover, the proposed ELBS guarantees a suitable interconnection among fog servers and both cloud and dew layer servers. ELBS is composed of five modules namely: (i) Priority Assigning Strategy (PAS), (ii) Data Searching Algorithm (DSA), (iii) External Data Requesting Algorithm (EDRA), (iv) Server Requesting Algorithm (SRA), and (v) Probabilistic Neural Network based Matchmaking Algorithm (PMA). PAS assigns a priority to each incoming Process (P) by considering three predefined parameters, which are; Predefined Priority (PP), Deadline Time (DT), and Task Size (TS). All those parameters are the inputs to a fuzzy inference system to assign the process priority. DSA is an algorithm to provide the required data for each arrived process in its fog region. EDRA is an algorithm used to search for the required data for each process in the neighbor servers. SRA is responsible for searching for the FS with the required capabilities for the incoming process. ELBS uses PMA to assign the process to the most appropriate server. It also defines a perfect methodology for a reliable connectivity among nodes. ELBS has been implemented and compared against recent load balancing techniques using iFogSim. Experimental results have shown that ELBS outperforms recent load balancing techniques as it achieves the lowest Average Turnaround Time and Failure Rate. Accordingly, ELBS is a suitable strategy to achieve load balancing in fog environment as it guarantees a reliable execution for real time applications.