Energy-Efficient Load Balancing Algorithms in Fog Computing

Abstract

Fog computing is an amalgamation of many network technologies working at edges of networks, thus differentiating itself from cloud computing with an increased focus on traffic load balancing at edges. Fog computing combines shared, geographically distributed and heterogeneous resources to achieve high computational performance. The objective of fog computing is to provide enhanced responsiveness with reduced latency, that too near to the user (at one hop distance). Fog computing offloads huge tasks to cloud and performs latency-sensitive tasks with the help of collaborative heterogeneous fog nodes within the fog network, nearer to the end devices. This helps in reduction of excessive traffic on the network along with the optimal usage of the available resources. These resources may belong to homogeneous or heterogeneous environments like devices working in different institutions, different domains and may pose tasks that requires high computations. One of the major challenge in such heterogeneous and complex computing environments is devising energy-efficient load balancing algorithm(s) in fog environment. Such algorithm(s) should be efficient, robust, and scalable with optimal use of available resources. To meet the growing traffic needs on the Internet as well as for optimal or minimal energy utilization between available fog nodes present in the fog zone, energy-efficient load balancing algorithm(s) are the current needs in fog computing environment. Efficient use of such algorithms will produce better Quality of Service (QoS) parameters (such as latency, responsiveness, availability, bandwidth, scalability, storage, energy consumption etc.) and increases performance of the system. This research work mainly focuses on ‘Energy-Efficient Load Balancing Algorithms in Fog Computing’, which deals with designing of energy efficient fog load balancer and optimizing the fog network paths for better traffic management. Initially, an in-depth review of existing models, approaches and algorithms has been done. During the literature review, it has been observed that existing load balancers can be improved to meet the current emerging technologies necessities, and for that, there is a need to have proper design and optimal fog load balancer algorithm, embedded into our devised fog load balancer. The empirical results of various designs of fog load balancers shows that fuzzy based traffic management component is most appropriate for working with the imprecise nature of load passing through the interconnects of the network. In this research work, a fuzzy-based fog load balancer is devised using different levels of design (3-level, 5-level and 7-level) and tuning of fuzzy controls. This fuzzy logic based algorithm has been implemented for conducting load analysis of interconnects for managing traffic. Use of optimization algorithms is one of the way to improve efficiency of the network in terms of responsiveness, latency and to avoid wastage of energy. Nano-Caches are integrated for delivering contents efficiently, using search-based optimization techniques which are energy and response aware in nature. An algorithm namely Modified Teaching Learning Based Optimization (MTLBO) is devised and implemented in fog zone to find efficient route for forwarding contents using Nano-Caches and subsequently to improve content retrieval time. Mathematical distribution model of traffic/load is used for simulation process. MTLBO is compared with existing algorithms, namely, Teaching Learning Based Optimization (TLBO) algorithm and Simulated Annealing (SA) algorithm. The design of experiments (DOE) has been carried out to observe number of iterations, learning rate and network size. The analysis shows that 3-level design is energy efficient for load balancing in fog zone due to reduced number of intervals in fuzzy design, reduced overheads in provisioning and improved responsiveness. Higher levels (5-level and 7-level) lead to creation of redundant fuzzy rules and wastage of resources. The optimization results show that Modified Teaching Learning Based Optimization (MTLBO) approach is better than Teaching Learning Based Optimization (TLBO) approach as it has less overheads in terms of memory (considering number of fog caches) and network size for delivering contents at remote areas. In comparison to the Simulated Annealing (SA) algorithm, MTLBO performs better in terms of execution time, overhead in terms of memory, and scalability as function of network size. To validate the research work, various applications have been considered to check the evaluation of proposed energy efficient fog load balancer. The case presented in this thesis gives insights on how fog devices and data mining can be used for bringing people into main fold of the economy. The inclusiveness index of the person is computed on the basis of four aspects: fitness, her/his inner social circle, her/his reliability to remain in a place, and call analysis. While computing the fitness index, it was found that Naive Bayes (NB) algorithm has the maximum accuracy with respect to K-Nearest Neighbors (KNN), Decision Tree (DT) and Linear Discriminant Analysis (LDA). For computing inner social circle, Louvain algorithm helped to compute stability and strength of socio-economic ties of an individual. For geospatial and call analysis, insights from knowledge discovery algorithm such as FP-Growth helped to arrive at decision to qualify the person for inclusive program. The thesis ends with details on how to automate the inclusiveness index computation using machine learning. The research indicates that energy is the key constraint for implementing such programs. Hence, a theoretical analysis about energy efficiency is also explained in the thesis. In this research work, a unique system of computing inclusiveness score has been introduced and implemented using fog load balancer in fog network.