Loading...
Traffic Offloading in LTE/Wi-Fi Heterogenous Wireless Networks Using Markov Decision Processes and Reinforcement Learning
Safavinejad, Ramin | 2021
243
Viewed
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 54645 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Ashtiani, Farid; Mirmohseni, Mahtab
- Abstract:
- We study an LTE-WiFi-Aggregation (LWA) network where some of the users support the LWA function, first standardized in 3GPP Rel. 13, and can simultaneously receive data from a WiFi access point and a cellular base station. In LWA networks, a base station can communicate with one or more Wi-Fi access points and if need be, it can transmit some of the user packets through the access points and their Wi-Fi networks. We consider an LWA network where some of the users are in the coverage of a base station and an access point, and support the LWA function. We aim to provide a scheduling policy for routing the LWA packets such that their expected delay is minimized. Three methods are proposed for this objective: Probabilistic scheduling, scheduling using Markov Decision Processes (MDPs) and scheduling using Reinforcement Learning (RL). While the Probabilistic method incurs the least overhead, it achieves the worst performance. The other two methods schedule the packets based on the conditions of the system at each time step, and can achieve up to 30 percent lower delays compared to the probabilistic method. The RL method, compared to the other two, requires no previous knowledge on the system characteristics, such as the number of users or the WLAN physical layer settings, and results show that it can work within a 10 percent performance margin of the MDP method
- Keywords:
- Reinforcement Learning ; Markov Decision Making ; Wireless Local Area Network ; Long-term Evolution (LTE)-Wireless Local Area Network (WLAN) Aggregation ; Traffic Offloading ; Heterogeneous Wireless Networks ; Delay
-
محتواي کتاب
- view