Abstract
Cognitive radio (CR) is proposed as a promising technique to solve the spectrum resource scarcity problem. Such radios are enabled to understand the surrounding radio environment and react accordingly to the operational conditions. Spectrum sensing has been regarded as a critical technique to realize CR functionality due to its low cost, simple implementation and high agility. However, there are still many technique challenges that to be solved in spectrum sensing in order to meet the special and strict requirements of CR networks (CRN). Collaborative spectrum sensing offers an efficient and effective solution to these challenges by utilizing spatial diversity among spaced CR nodes. However, additional energy is consumed due to collaboration and thus energy-efficient schemes should be designed to balance the trade-off between the achievable system throughput and the energy consumption. This thesis first presents a comprehensive performance analysis and comparison between different stand alone sensing techniques in terms of reliability, delay and complexity. The analytical analysis results are validated by simulation and the appropriate application scenarios are presented for each sensing technique. Based on the obtained results, CR networks employing heterogeneous sensors (CRNHS) are considered and sensor allocation schemes based on the modified particle swarm optimization (PSO) algorithm are proposed to maximize the secondary system energy efficiency in multi-band collaborative spectrum sensor networks. It is demonstrated that the proposed algorithm, namely PSO-SAP, improves the energy efficiency of the multi-band secondary system dramatically. When limited battery power is taken into account, a network throughput maximization problem is formulated and we propose an ant colony optimization (ACO)-based energy-efficient sensor scheduling scheme, namely ACO-ESSP, to maximize the network throughput in CRNHSs. The proposed algorithm is demonstrated to improve the network throughput efficiently and effectively compared with a greedy-based approach and a Genetic algorithm.