A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Load Balancing and Resource Allocation in Smart Cities using Reinforcement Learning
Technology is being adopted by many municipalities to improve their services and operations through so-called “smart city applications”. Development of these smart city applications are increasingly looking to leverage edge and cloud computing to support these initiatives. This approach brings a number of challenges, including the allocation of tasks to that process data from sensors to computing systems, dealing with the data that may be time sensitive, handling frequently starting and finishing tasks, dynamically changing resource utilization of computing elements, etc. This paper focuses on the challenge of allocating tasks to resources to try to ensure balanced loads on processing elements in dynamic environments. A reinforcement learning approach is introduced to address this problem. The reinforcement learning agent uses a novel Multi-Observation Single-State model based on observed features from multiple sources at a single step. A model of a smart city computational infrastructure is introduced and is used to define our reinforcement learning algorithm for task allocation. We illustrate the agent behavior through simulations.
Load Balancing and Resource Allocation in Smart Cities using Reinforcement Learning
Technology is being adopted by many municipalities to improve their services and operations through so-called “smart city applications”. Development of these smart city applications are increasingly looking to leverage edge and cloud computing to support these initiatives. This approach brings a number of challenges, including the allocation of tasks to that process data from sensors to computing systems, dealing with the data that may be time sensitive, handling frequently starting and finishing tasks, dynamically changing resource utilization of computing elements, etc. This paper focuses on the challenge of allocating tasks to resources to try to ensure balanced loads on processing elements in dynamic environments. A reinforcement learning approach is introduced to address this problem. The reinforcement learning agent uses a novel Multi-Observation Single-State model based on observed features from multiple sources at a single step. A model of a smart city computational infrastructure is introduced and is used to define our reinforcement learning algorithm for task allocation. We illustrate the agent behavior through simulations.
Load Balancing and Resource Allocation in Smart Cities using Reinforcement Learning
AlOrbani, Aseel (author) / Bauer, Michael (author)
2021-09-07
233693 byte
Conference paper
Electronic Resource
English
Monthly Electric Load Forecasting Using Transfer Learning for Smart Cities
| DOAJ | 2020
Load Balancing Algorithm on the Immense Scale of Internet of Things in SDN for Smart Cities
| DOAJ | 2021
Review Pages: Smart Cities Challenges: Smart Environment for Sustainable Resource Management
| DOAJ | 2014