GPU Software Implementation of Spectrum Sensing Algorithms for Cognitive Radio

• Main Authors: Chu-Han Lee, 李居翰
• Other Authors: Sao-Jie Chen
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/65677434068921679349

碩士 === 國立臺灣大學 === 電子工程學研究所 === 100 === In 1999, Mitola proposed the idea of cognitive radio (CR), which is a promising technology to achieve efficient spectrum utilization. Cognitive radio is more flexible and intelligent than traditional wireless communication techniques. Cognitive radios have the ability to sense their operating environment and automatically switch between different standards. A cognitive radio system needs to sense the primary user radio spectrum fast and accurately. Various detection approaches have been proposed for spectrum sensing, such has energy detection, waveform-based sensing, and cyclostationarity-based sensing methods. In this Thesis, we implement two kinds of spectrum sensing techniques, waveform-based detection and cyclostationary-based sensing methods. Both of these algorithms have the ability to separate the signal of interest from the noise or interference and own a high computation complexity. In order to reduce the computation time and increase the detection speed, we implemented these algorithms on an GPU (Graphic Processing Unit) platform using CUDA (Compute Unified Device Architecture) 4.0. By efficiently using the parallel processing power of CUDA, our methods showed tremendous speed-up over the sequential implementations in a multi-standards environment. In the end, we also compared our results with the results of other multi-core device to show that GPU is a promising platform to implement high-speed parallel algorithms.