Research On Wideband Spectrum Sensing For Cognitive Radios

Cognitive Radio (CR), as a kind of intelligent spectrum sharing technique, can detect and access the allocated but unoccupied spectrum of licensed primary user (PU). Thus, the CR technique offers an efficient way to solve the dilemma of the frequency resource scarcity and the increasing demand on available spectrum. The first and key task in CR technique is to perform accurate and fast spectrum sensing, in order to determine the spectrum occupancy of licensed PUs and identify potential transmission opportunities for secondary CR users. With the rapid development of radio access technologies and related services, the task of spectrum sensing is to get more unoccupied spectrum opportunities from much wider spectrum. Therefore, the wideband spectrum sensing is becoming a new but crucial issue in CR field and catch attentions in both research and industry societies. This dissertation mainly focuses on discussing and solving several problems in wideband spectrum sensing for CRs and is also a collection of results achieved during the author participating the following projects:"Research of Spectrum Detection based on Time-domain Probability Distribution Curve Estimation of Spectrum Usage" supported by NSFC grant number 60772110, "Research of Spectrum Resource Sharing System" supported by 863 Project grant number 2009AA011802, etc.The main work in this dissertation includes the following parts:Chapter 2 surveys the existed research on CR spectrum sensing. Chapter 3 discusses the local wideband compressed spectrum sensing in the lack of prior knowledge about sparsity order. Chapter 4 discusses the cooperative spectrum sensing under complicated circumstances. Chapter 5 discusses the performance analysis of block diagonal structure sampling matrices which are adopted in the proposed algorithms in Chapter 3 and 4. The main contents and contributions of this dissertation are issued as follows:1. In the survey of existed CR spectrum sensing, the specific spectrum sensing techniques are sorted into different categories. Within each category, spectrum sensing algorithms and schemes are summarized. In the meanwhile, a multihead-selected based cooperative spectrum sensing method and optimal decision rule are proposed to cope with fading channel effects. After discussing the traditional narrowband scenario, the wideband spectrum sensing is then introduced as a brand new research field, and followed with the problems and challenges in such a field which are the issues this dissertation will focus.2. In the local CR wideband spectrum sensing, the problem of sampling wastage caused by the lack of prior knowledge about sparsity order is discussed firstly. Next, the concept of sparsity order estimation is introduced in order to acquire the accurate sparsity of the wideband spectrum. And the gap between the required number of samples for sparsity order estimation and that for sparse signal reconstruction is observed and quantified. Capitalized on such a gap, a two-step compressed spectrum sensing (TS-CSS) algorithm is then proposed to avoid the sampling wastage. Simulation results show that the overall sampling cost caused by TS-CSS algorithm can be minimized adaptively, given the desired sensing performance.3. In the part of cooperative spectrum sensing for wideband, fading channels, existing noise, channel state information unknown, hardware limited CR system to detect wideband spectrum, the system and signal models are first set up for the collaborated CR system. According to the established models, a low rank property of the received spectrum matrix is recognized, and the rank order is the same as the size of the nonzero support of the monitored wideband spectrum. Capitalized on such a nice low rank property, a cooperative spectrum sensing based on matrix rank minimization (CSS-MRM) algorithm is then proposed. The CSS-MRM algorithm offers an efficient tradeoff between the detection diversity gain and the sampling cost reduction. Simulation results show that the CSS-MRM algorithm can reduce the sampling cost while improve the sensing performance.4. In both the TS-CSS algorithm and the CSS-MRM algorithm, a type of block diagonal structured matrices is used corresponding to the two-step procedure and the collaborated processing matrices of individual CR users. To characterize such a type of matrices, a family of generalized block diagonal (GBD) structure is modeled. The restricted isometry property of such structured matrices is established to reveal the required number of samples for perfect sparse signal reconstruction with high probability. Simulation results show that the GBD matrices enjoy several nice structural benefits in implementation at minimal extra cost in terms of the number of samples.