A preview of the PDF is not available
Evaluation and Analysis of Bio-Inspired Techniques for Resource Management and Load Balancing of Fog Computing
With the evolution of fog computing, processing takes place locally in a virtual platform rather than in a centralized cloud server. Fog computing combined with cloud computing is more efficient as fog computing alone does not serve the purpose. Inefficient resource management and load balancing leads to degradation in quality of service as well as energy losses. Traffic overhead is increased because all the requests are sent to main server causing delays which cannot be tolerated in health-care scenarios. To overcome this problem, the authors are consolidating fog computing resources so that requests are handled by cloudlets and only critical requests are sent to cloud for processing. Servers are placed locally in each city to handle the near-by requests in order to utilize the resources efficiently along with load balancing among all the servers, which leads to reduced latency and traffic overhead with the improved quality of service. Due to the limited data storage capacity available to Internet service providers and large-scale enterprises, the concept of resource sharing arises. The services can be given on lease to enterprises through Service Level Agreements (SLAs). Being the extension of the cloud computing, fog computing architecture brings the resources near end users. In order to get the services on lease, the enterprises are supposed to pay for the resources or services which are being used by them. For this, four nature inspired algorithms are analyzed in order to determine the efficient management of services or resources so that the cost of resources can be reduced and the billing can be attained through calculation of the utilized resources. Pigeon Inspired Optimization (PIO), Enhanced Differential Evolution (EDE), Binary Bat Algorithm (BBA) and Simple Human Learning Optimization (SHLO) are used to evaluate the energy consumed by the edge nodes or cloudlets that in turn can be used for estimating the bill through the Time of Use pricing variable. We evaluate the aforementioned techniques to analyze their performance regarding the bill calculation on the basis of fog servers usage. Simulation results demonstrate that BAT algorithm gives significantly better results than other three algorithms in terms of resource utilization and bill reduction.