POLSTM: Poplar optimization-based long short term memory model for resource allocation in cloud environment

Abstract

Due to the evolution of Internet of Things (IoT) paradigm, the number of devices is growing day by day which arises the stringent requirements for various communications. The standard cloud computing model is not capable of efficiently hosting IoT tasks due to the high latency associated with it. Hence, this work utilizes the mobile edge computing and fog concept to process the input tasks independent of the cloud layer. Generally, the high computation costs and energy consumption resulted in the applications such as power-hungry and computation-intensive, which become massive challenges for an IoT device. Motivated by the previous research, we mainly plan to investigate the resource allocation issues that arise in the MEC-enabled IoT–Fog–cloud architecture. We propose a Poplar Optimization algorithm (POA) based on Attention 1DCNN-LSTM architecture (POA-A1DCNN-LSTM) for improving the total utility of the MEC servers by optimizing the energy consumption and task delay. Initially, the POA algorithm is implemented for cluster head (optimal fog node) selection using the node degree, node distance, and residual energy. Next, the A1DCNN-LSTM architecture is employed for task offloading by selecting the fog node with minimal task length and processing delay. The performance of the proposed method is validated by average latency, user satisfaction, network lifetime, energy consumption, average time delay, and normalized system utility. The experimentation results revealed that the proposed method attained better effectiveness in different metrics by achieving 1.5 ms, 0.65, 60 s, 0.45 j, 385 ms, and 0.98 compared to state-of-the-art methods.The experimentation outcomes demonstrate the effectiveness of the proposed POA-A1DCNN-LSTM architecture to decrease task completion delay and offer effective task scheduling as well as resource allocation performances in the MEC-enabled IoT–Fog–cloud network.