We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
The Monte Carlo tree search based bandwidth slicing allocation algorithm.
- Authors
Chang, Shun-Chieh
- Abstract
This study discusses various bandwidth allocation methods for network slicing and identifies their limitations. The heuristic-based methods were found to be insufficient in handling the randomness of varying channel quality and bandwidth requirements of each slice, while the AI-based methods were not sufficiently optimized. To address these limitations, this study proposes the MCTS-RA algorithm, which is based on Monte Carlo Tree Search methodology, to handle the randomness of each slice's channel quality and requested bandwidth and optimize the throughput. Additionally, MCTS-RA provides a generic performance index, i.e., MCTS-Level, to evaluate the performance gap of previous works to the optimal solution. The simulation results showed that MCTS-RA improved the overall throughput achieved by previous algorithms by around 10%. Furthermore, the results revealed that the MCTS-Level of previous algorithms was around 100–200, while the optimal solution's MCTS-Level was around 2000.
- Subjects
BANDWIDTH allocation; ARTIFICIAL intelligence; ALGORITHMS; TREES; BANDWIDTHS
- Publication
Wireless Networks (10220038), 2024, Vol 30, Issue 2, p835
- ISSN
1022-0038
- Publication type
Article
- DOI
10.1007/s11276-023-03521-3