The Study of Reducing False Spectrum Sensing Probability in Cognitive Radio Using Reinforcement Learning

• Main Authors: Cheng-Wei Chang, 張政偉
• Other Authors: Wei-Yeh Chen
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/98366188208164261362

碩士 === 南台科技大學 === 資訊管理系 === 98 === As the vigorous development of wireless communications technology in recent years, there are more and more users accessing the wireless network, causing the spectrum scarcity becoming a severe problem, especially when the high data throughout is required. Nowadays, only the ISM band is unlicensed, most other spectrum bands need licensed to use. The Federal Communications Commission(FCC) reports indicated that the traditional static spectrum allocation resulted in low spectrum utilization. To solve this problem, the development of cognitive radio is an important technology. Cognitive radio is an intelligent communication technology which can enhance the efficiency of spectrum utilization by exploiting the unused resources of primary system, while not interfering with the transmission of primary users. Therefore fast and accurate spectrum sensing is very important in realizing a reliable cognitive radio. However, the error sensing of spectrum will result in low spectrum utilization. In the paper we proposed an reinforcement learning mechanism called DQ-Learning to reduce the false spectrum sensing probability. The simulation results showed that the proposed scheme can effectively reduce the false spectrum sensing probability, therefore increasing the successful channel access rate and system throughput.