Research On Dynamic Spectrum Management Based On Spectrum Sensing

ABSTRACT:Along with the development of wireless communication, the demand for radio spectrum is increasing, which causes the shortage in spectrum resource. However, the current static spectrum policy leads to spectrum resource waste. Dynamic spectrum sharing is efficient in improving spectrum utilization. Cognitive Radio is a key technology to implement dynamic spectrum sharing. Meanwhile, spectrum sensing introduces new challenges in spectrum management.Spectrum sensing is regarded as essential in cognitive radio. Cognitive radio-based spectrum management includes spectrum sensing, spectrum decision, spectrum sharing and spectrum handoff, all of which are investigated independently. Actually, there might be potential profit in considering them jointly, especially the joint design between spectrum sensing and others.The spectrum sensing-based dynamic spectrum management in clusted cognitive radio networks is investigated in this thesis. The innovations are as following:Spectrum sensing is modified to implent detection through wide band. Compressed spectrum sensing is proposed to help secondary users sense a wide band., in which compressive sampling is applied to reduce sampling rate. Two scenarios based on compressed spectrum sensing are discussed to explore the joint design of spectrum sensing and allocation.The spectrum decision considering spectrum sensing cost is explored, as well as a sequential cooperative spectrum sensing algorithm to suit the decision. Spectrum sensing cost function is modeled as a measurement of spectrum characteristic during spectrum decision process. A sequential cooperative spectrum sensing algorithm is proposed to minimize the spectrum sensing cost. Sequential detection is applied to handle the tradeoff between detection cost and accurancy. Distributed secondary users are activated to sample in parallel, which improves fairness in energy cost. Besides, a conditional activation and sampling scheme is proposed to reduce delay caused by sequential detection. The result could be applied in general scenarios requiring detection with low energy cost and limited delay.The joint design of spectrum sensing and handoff is explored. Hidden Markov Model is used for channel occupancy prediction. Performance parameters of spectrum sensing, i.e., false alarm probability and missing detection probability, are introduced as factors of elements of the Hidden Markov Model, to enhance prediction accuracy. Meanwhile, spectrum sensing algorithm is adjusted according to spectrum handoff performance. As a result, both interference to the primary users and spectrum processing delay of the secondary users are reduced.