(The) study on cooperative spectrum sensing using soft combination in cognitive radio networks

Abstract

Recently, Cognitive Radio (CR) has been highlightened as an epochal technology which enables to resolve the spectrum scarcity problem resulted from the fixed spectrum allocation policy. It can be implemented by allowing secondary users (SU) to utilize the spectrum band licensed by primary users (PU) without imposing any harmful interference to the PU. To guarantee no interference to the PU, the most important function is a spectrum sensing which is required to be accurate and fast. Although there have been serveral spectrum sensing algorithms such as matched filter method, energy detection method, cyclostationary detection, it may be difficult to detect the activity of PU due to the characteristics of wireless channel method by only single SU. In this context, cooperative spectrum sensing allows us to achieve robustness on spectrum sensing without drastic requirements on single SU. Based on the type of fusion rules at a fusion center, cooperative spectrum sensing can be classified into centralized detection and decentralized detection. While the former has optimal detection performance on condition that transmitted information from individual SU should not be processed, the latter is required to only one bit for reporting individual information in hard decision fusion rule, but accepts the performance loss is an inevitable aspect. In this thesis, we focused on the soft decision fusion rule for decentralized detection, which is based on the two-bit soft decision fusion rule. From the standpoint of implementation of fusion rule, the soft decision fusion rule has two aspects considered carefully, which is the quantization at the local SUs and fusion rule at the fusion center. Due to the difficulties in obtaining optimal fusion rule considering followed aspects, we focused on suboptimal two-bit soft decision fusion rule which is based on the form of polynomial. Moreover, based on this fusion rule, we proposed the modified two-bit soft decision fusion rule in wh...