Veri tabanlı spektrum tahsisi ile daha güvenilir ortam erişim kontrolüne doğru

Most of the deployed wireless networking technology relies on Wi-Fi communications that depends on IEEE 802.11 standard. IEEE 802.11 consists of a set of protocols for medium access control (MAC) and physical layer (PHY) communications. Distributed Coordination Function (DCF) included in IEEE 802.11 employs a single channel in the MAC layer. However, network traffic load on a single channel leads to high packet collision and performance degradation. Single-Radio Multi-Channel (SRMC) Medium Access Control (MAC) protocols aim to transmit in parallel on distinct channels while reducing the hardware cost. Initially, we consider dynamic multi-channel access problem with independent and stochastically identical channels. We formulate the problem as a Restless Multi-Armed Bandit (RMAB) process and propose a novel load-adaptive frequency hopping method, namely Index-Based Channel Hopping (IBCH) algorithm, with the goal of optimal spectrum allocation and throughput maximization. We classify previous efforts by their spectrum decision mechanisms and rendezvous characteristics as single- and multi-rendezvous protocols. Multi-rendezvous protocols have the capability of supporting simultaneous handshaking on different channels whereas with single-rendezvous protocols only asynchronous channel negotiations are allowed. Later on, we compare performance of IBCH with various protocols from both rendezvous classes at varying transmission range, network size and spectrum availability. Simulation results show that IBCH significantly outperforms other spectrum assignment methods from both rendezvous classes. In the second part of the thesis, we focus on IEEE 802.22 based Wireless Regional Area Networks (WRANs). IEEE 802.22 WRAN standard is aimed at using cognitive radio (CR) to avoid user interference and congestion. Cognitive Radio (CR) is a promising technology for emerging intelligent wireless communication systems due to its effiiv cient utilization of the frequency bands. CR systems enhance temporal and spatial efficiency, therefore, ease spectrum scarcity by exploiting temporary idle periods in the frequency usage, known as spectrum holes. Dynamic Spectrum Access (DSA) improves spectrum utilization by allowing secondary users (SUs) to opportunistically access temporary idle periods in the primary user (PU) channels. Previous studies on utility maximizing spectrum access strategies mostly require complete network state information, therefore, may not be practical. Model-free reinforcement learning (RL) based methods, such as Q-learning, on the other hand, are promising adaptive solutions that do not require complete network information. In the second section, we tackle this research dilemma and propose deep Q-learning originated spectrum access (DQLS) based decentralized and centralized channel selection methods for network utility maximization, namely DEcentralized Spectrum Allocation (DESA) and Centralized Spectrum Allocation (CSA), respectively. Actions that are generated through centralized deep Q-network (DQN) are utilized in CSA whereas the DESA adopts a non-cooperative approach in spectrum decisions. We use extensive simulations to investigate spectrum utilization of our proposed methods for varying primary and secondary network sizes. Our findings demonstrate that proposed schemes outperform model-based RL and traditional approaches, including slotted-Aloha and Whittle index policy, while %87 of optimal channel access is achieved.