QTSCH: A Lightweight Reinforcement Learning Based TSCH Scheduling Approach for TSCH Networks QTSCH: Pekiştirmeli Ö?grenme Tabanli Hafif Bir TSCH Zamanlama Yaklaşimi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636898
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: 6TiSCH, IIoT, Q-learning, reinforcement learning, scheduling, TSCH
- Ondokuz Mayıs Üniversitesi Adresli: Evet
Özet
This paper proposes QTSCH, a lightweight reinforcement learning based scheduling approach for Time Slotted Channel Hopping networks. The method selects between a shared common cell and a neighbor specific unicast cell for DATA packets using a per-node epsilon-greedy Q-learning mechanism. The proposed design preserves an Orchestra-like distributed slotframe architecture while making only the data transmission decision adaptive. In addition, the learning signal is updated using MAC-layer outcomes including successful transmission, collision, link failure, and wrong address events. Simulation results obtained in TSCH-Sim and compared with MSF and Orchestra show that QTSCH achieves lower latency, fewer queue drops, fewer MAC collisions, lower idle listening, and a lower radio duty cycle. The results demonstrate that QTSCH provides multi-metric improvement with low computational complexity.