Edge Intelligent Joint Optimization for Lifetime and Latency in Large-Scale Cyber–Physical Systems
- 4 August 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 9 (22) , 22267-22279
Abstract
In recent years, the exploration on large-scale cyber–physical systems (CPSs) has become a fertile research field of significant impact. Large-scale CPS applications cover not only manufacturing and production areas but also daily living domains. Traditional solutions dedicated for large-scale CPSs mainly concentrate on the service latency or reliability optimization, but neglect the resultant negative impact on system lifetime. In this article, we conduct the first study on jointly optimizing the service latency and system lifetime subject to the constraints of reliability, energy consumption, and schedulability for large-scale CPSs. We propose an edge intelligent solution composed of offline and online phases. At the offline phase, the long short-term memory (LSTM) technique is leveraged to predict task offloading rates at individual user groups. Afterward, the multiobjective evolutionary algorithm with dual local search (DLS-MOEA) is exploited to determine optimal system static settings of computation offloading mapping and task replication number. At the online phase, an affinity-driven scheme incurring minimal system dynamic overheads is designed to deal with the inherent mobility of terminal users. We also build an algorithm validation platform upon which extensive simulation experiments are carried out. Experimental results show that our offline and online schemes outperform the state-of-the-art benchmarking methods by 27.1% and 43.5%, respectively.Keywords
Funding Information
- National Natural Science Foundation of China (62102164, 61802146)
- China Postdoctoral Science Foundation (2021T140272, 2021M691240)
- Fundamental Research Funds for the Central Universities (21621025)
- Basic and Applied Basic Research Foundation of Guangdong Province (2019A1515011017)
This publication has 46 references indexed in Scilit:
- vMCN2016
- Edge Computing: Vision and ChallengesIEEE Internet of Things Journal, 2016
- A Vertical and Cyber–Physical Integration of Cognitive Robots in ManufacturingProceedings of the IEEE, 2016
- Cost Efficient Resource Management in Fog Computing Supported Medical Cyber-Physical SystemIEEE Transactions on Emerging Topics in Computing, 2015
- Phase-Change Memory Optimization for Green Cloud with Genetic AlgorithmIEEE Transactions on Computers, 2015
- Throughput maximization for periodic real-time systems under the maximal temperature constraintACM Transactions on Embedded Computing Systems, 2014
- MOEA/D: A Multiobjective Evolutionary Algorithm Based on DecompositionIEEE Transactions on Evolutionary Computation, 2007
- Travel-Time Prediction With Support Vector RegressionIEEE Transactions on Intelligent Transportation Systems, 2004
- A fast and elitist multiobjective genetic algorithm: NSGA-IIIEEE Transactions on Evolutionary Computation, 2002
- Stochastic Decompositions in the M/G/1 Queue with Generalized VacationsOperations Research, 1985