Q-Learning Based Scheduling for Delay Optimization in Wireless Network Simulation: A MATLAB Approach
Tulkin BotirovNavoi State University of Mining and Technologies, Navoi, UzbekistanBaxtiyor SodiqovNavoi State University of Mining and Technologies, Navoi, UzbekistanIlyos KalandarovNavoi State University of Mining and Technologies, Navoi, UzbekistanBobokul BotirovNavoi State University of Mining and Technologies, Navoi, UzbekistanBoburjon VafoevTashkent State University of Economics, Tashkent, Uzbekistan
2025
ABI
Annotatsiya
This paper investigates Q-learning based scheduling to minimize end-to-end delay in wireless networks using a MATLAB framework. We model traffic sources, finite queues, and a time-varying wireless channel; the agent observes queue occupancy, offered load, and channel quality to select service rates. Reward balances normalized throughput, delay, and loss, with ε-greedy exploration and online updates. Under diverse loads and channel conditions, the learned policy consistently outperforms a FIFO fixed-rate baseline, yielding lower delay and packet loss with competitive throughput. Results high-light SimEvents as a practical testbed for AI-driven control in next-generation networks.
Mavzular
Identifikatorlar
Iqtiboslar va manbalar
0 ta iqtibos13 ta foydalanilgan manba
Koʻrsatkichlar — AkademScholar · Tez orada