Adaptive DBN Using Hybrid Bayesian Lichtenberg Optimization for Intelligent Task Allocation

Abstract

The cloud-based resources can accommodate massive energy sources and the main challenge arises when processing the terminal nodes associated with the Internet of Things (IoT). The main challenge the cloud faces when satisfying the user requests is the long delay, large bandwidth, and resource-constrained devices not capable of processing computational needs. These challenges are overcome in this work by using fog computing to process a wide range of IoT requests and workloads near the end user. The main motivation of this work is to enhance the intelligent task offloading decision by achieving a tradeoff between Quality of Service (QoS) and power consumption for a large number of mobile and fog nodes. This paper proposes an Adaptive Deep Belief Network (ADBN) that uses the Hybrid Bayesian Search and Lichtenberg Optimization (BSI-LO) technique to solve this complexity. The hybrid BSI-LO optimized ADBN architecture minimizes the energy consumption of fog devices operating on the network edge by taking different parameters such as battery lifetime, delay, workload, and power consumption into consideration and offers an effective intelligent task allocation decision. In this way, the proposed model handles a large number of requests raised in the IoT-Fog cloud network by efficient resource allocation. The hybrid BSI-LO algorithm is implemented to mutually optimize the resource allocation and offloading strategy. The experimental results confirm that the proposed methodology is effective in terms of improving the bandwidth, task weight cost, energy consumption, delay, and system utility of edge devices resulting in an efficient energy-saving strategy for the IoT-Fog-cloud architecture.