DECENT: Deep Learning Enabled Green Computation for Edge Centric 6G Networks

Abstract

Edge computing has received significant attention from academia and industries and has emerged as a promising solution for enhancing the information processing capability at the edge for next generation 6G networks. The technical design of 6G edge networks in terms of offloading the computationally extensive task is very critical because of the overgrowth in data volume primarily due to the explosion of smart IoT devices, and the ever-reducing size of these energyconstrained devices in IoT systems. Toward harnessing the benefits of deep recurrent neural network based on Long Short Term Memory (LSTM) in the design of next-generation edge networks, this paper presents a framework DECENT-Deep learning Enabled green Computation for Edge centric Next generation 6G neTworks. The data offloading problem is modeled as a Markov decision process considering joint optimization of energy consumption, computation latency, and offloading rate for network utility in 6G environment. The algorithm learns faster from previous longterm offloading experiences and solves the optimization problem with better convergence speed. Simulation results of the proposed framework DECENT shows that it maximizes the network utility by overcoming the challenges as compared to the state-of-the-art techniques.

Computation offloading techniques in edge computing: A systematic review based on energy, QoS and authentication

“…A 31.25% of researchers have used the development environments such as Microsoft Visual Studio, 82,89 UBUNTU, 67,111 Anaconda(spyder), 81 MATLAB, 37,54,56,59,68,69,72,73,77,103,105,116,126,138,141,142,146,154,158,162 IBM Ilog Cplex optimization Studio. 71 A 20.00% of researchers used DL and ML based libraries such as TensorFlow, 46,47,62,63,65,66,110,144,151,161 Keras, 47 scipy, 45 sckit tool, 113 Pytorch, 111,123 TVM deep learning compiler, 120 LIBSVM. 166 Python, 40,62,63,65,81,109,110,144,…”

“…71 A 20.00% of researchers used DL and ML based libraries such as TensorFlow, 46,47,62,63,65,66,110,144,151,161 Keras, 47 scipy, 45 sckit tool, 113 Pytorch, 111,123 TVM deep learning compiler, 120 LIBSVM. 166 Python, 40,62,63,65,81,109,110,144,151,161 C++, 99 and Java, 71,94 programming languages were considered by 15.00% of researchers. A 13.75% of the researchers have used various simulators and emulators as CloudSim, 34,74,158 EdgeCloudSim, 166 iFogSim, 148 Contiki Cooja environment, 108 Open AI Gym Toolkit, 144 SDN-BC emulated environment including Minine, 144 COOJA, 66 SimuLTE, 157 and SUMO.…”

“…Kashyap et al 151 proposed a DL‐enabled green computation for edge centric next generation 6G networks (DECENT) framework to solve the issue of rising data size generated by smart IoT devices and limited energy of 6G edge networks for offloading computationally demanding crucial tasks. The optimal data offloading problem was formulated as an MDP to represent the network utility maximization that optimizes ECp, computation latency, and offloading rate.…”

“…Kashyap et al151 proposed a DL-enabled green computation for edge centric next generation 6G networks (DECENT) framework to solve the issue of rising data size generated by smart IoT devices and limited energy of 6G edge networks for offloading computationally demanding crucial tasks.…”

Green Communication in Internet of Things: A Hybrid Bio-Inspired Intelligent Approach

“…Clustering is a robust and scalable approach to lower energy consumption with better network throughput [ 4 , 5 ]. It is widely studied as probability-based [ 6 , 7 , 8 , 9 ], weight-based [ 10 , 11 , 12 ] and heuristic-based approaches [ 13 , 14 , 15 , 16 , 17 , 18 ] for conservation of network energy.…”