Volume 18, No. 6, 2021

Resource Optimization In Fog Computing With ShiftInvariant Deep Convolutive Load Balancing

S. V. Nethaji , Dr. M. Chidambaram


Fog computing is a well-known computing paradigm that uses fog environments to deliver computation and storage services, effectively managing decentralised resources. Internets of Things (IoT) applications choose fog computing nodes to meet end-user needs. Fog computing is a notion that lies somewhere between IoT devices and the cloud paradigm, with the purpose of lowering job scheduling and load balancing latency. Load balancing is what determines the effectiveness of resource allocation and management solutions. Various solutions have been proposed to resolve issues about cloud load balancing. Using population-based methods, on the other hand, did not improve load balancing efficiency or resource use. Multi objective chaotic salp swarm resource optimised shift-invariant deep convolutive load balancing (MCSSROSIDCLB) is a new technique for effective load balancing with a short makes pan. The MCSSROSIDCLB approach uses numerous processing layers to achieve correct results, including input, more than one hidden layer, and output layer. The MCSSROSIDCLB technique connects user-requested tasks to Fog nodes according to resource availability. The input layer collects the number of tasks before passing it on to the hidden layer. The load balancer evaluates fog node resource availability by taking end-user requests. The load balancer finds the resource-optimized fog node based on CPU, memory, and bandwidth in the second hidden layer by using Multi objective chaotic salp swarm optimization. Finally, the load balancer distributes incoming workloads to the resourceoptimized fog node in the third hidden layer. At the output layer, this technique allows for more resource-optimized load balancing. Experiments on characteristics including load balancing efficiency, make span, and memory consumption are carried out for a variety of user tasks. The observed data and discussion reveal that the MCSSROSIDCLB technique improves loadbalancing efficiency and reduces make span and memory utilisation when compared to state-ofthe-art alternatives.

Pages: 3845-3861

Keywords: Fog computing, load balancing, resource allocation, shift-invariant convolutive deep learning, Mult iobjective chaotic salp swarm optimization

Full Text