查看文章

Cognitive radio (CR) is the next-generation wireless communication system that has been proposed to address spectrum scarcity in the traditional static spectrum assignment policy. The static spectrum assignment policy allocates frequency bands to licensed users or primary users (PUs) for their exclusive usage, and so unlicensed users or secondary users (SUs) are forbidden from accessing the licensed channels. CR uses dynamic spectrum access to solve this problem, which have two main approaches to access channels. Firstly, SUs access the channel in an opportunistic manner, while the PUs are oblivious to the presence of SUs. Secondly, SUs negotiate with PUs for channel access in a collaborative manner in order to achieve mutual benefit, such as Quality of Service (QoS) enhancement for both PUs and SUs.

To date, research has been primarily focused on simulation-based investigation. There were only a perfunctory effort to investigate the network layer of CR networks on a real testbed environment. This research work is a pioneering effort to examine opportunistic and collaborative channel access approaches at the network layer using real testbed implementation. There are three major contributions in this thesis. Firstly, a channel selection scheme is implemented in multi-hop CR network using reinforcement learning (RL) on a USRP/GNU radio platform. Secondly, route selection schemes are implemented in a multi-hop CR network using RL and SL with the objective of improving QoS performance. Thirdly, addresses the challenges of network-layer implementation using USRP/GNU radio platform. Analyzes the outcomes and …