@phdthesis{oai:uec.repo.nii.ac.jp:00008646,
 author = {佐藤, 光哉 and Sato, Koya},
 month = {2018-03-29},
 note = {2017, The growth in demand for mobile communication systems has exponentially increased data traffic during the last decade.  Because this exponential growth consumes finite spectrum resources, traditional spectrum utilization policies with exclusive resource allocation faces a limit. In order to develop novel spectrum resources, many researchers have shown an interest in spectrum sharing with cognitive radio (CR). This method allows secondary users (SUs) to share spectrum bands with primary users (PUs) under interference constraints for PUs. SUs are required to take into consideration the interference margin to the estimated interference temperature at PUs in order to protect communication quality of PUs. On the other hand, an excess interference margin decreases the spectrum sharing opportunity; therefore, it is important to manage the interference power properly.
 Spectrum estimation techniques in spectrum sharing can be categorized into two methods: spectrum sensing and spectrum database. Spectrum sensing uses the detection of PU signals to characterize radio environments.  To provide good protection, signal detection must be performed under the (strict) condition that the PU signal strength be below the noise floor, even under low signal-to-noise ratios (SNRs) and fading conditions. These fluctuations make it difficult for the SUs to achieve stable detection; thus, it is very challenging to accurately estimate the actual activity of the PU. The second method is based on storing information about spectrum availabilities of each location in spectrum databases. In this method, after
SUs query the database before they utilize the spectrum, the database provides spectrum information to the SUs.  Current databases usually evaluate white space (WS) based on empirical propagation models. However, it is well known that empirical propagation models cannot take into account all of the indeterminacies of radio environments, such as shadowing effects. Because SUs must not interfere toward PUs, the conventional database requires the SUs to set large margins to ensure no interference with PUs.
In this dissertation, we propose and comprehensively study a measurement-based spectrum database for highly efficient spectrum management. The proposed database is a hybrid system,  combining spectrum sensing and a spectrum database.  The spectrum database consists of radio environment information measured by mobile terminals.  After enough data are gathered, the database estimates the radio environment characteristics by statistical processing with the large datasets. Using the accurate knowledge of the received PU signal power, spectrum sharing based on PU signal quality metrics such as the signal-to-interference power ratio (SIR) can be implemented.
We first introduce the proposed database architecture. After we briefly discuss a theoretical performance of the proposed database, we present experimental results for the database construction using actual TV broadcast signals.  The experimental results show that the proposed database reduces the estimation error of the radio environment.
 Next, we propose a transmission power control method with a radio environment map (REM) for secondary networks.  The REM stores the spatial distribution of the average received signal power. We can optimize the accuracy of the measurement-based REM using the Kriging interpolation. Although several researchers have maintained a continuous interest in improving the accuracy of the REM, sufficient study has not been done to actually explore the interference constraint considering the estimation error.  The proposed method uses ordinary Kriging for the spectrum cartography. According to the predicted distribution of the estimation error, the allowable interference power to the PU is approximately formulated. Numerical results show that the proposed method can achieve the probabilistic interference constraint asymptotically, and an increase in the number of measurement datasets improves the spectrum sharing capability.
 After that,  we extend the proposed database to the radio propagation estimation in distributed wireless links in order to accurately estimate interference characteristics from SUs to PUs.  Although current wireless distributed networks have to rely on an empirical model to estimate the radio environment, in the spectrum sharing networks, such a path loss-based interference prediction decreases the spectrum sharing opportunity because of the requirement for the interference margin. The proposed method focuses on the spatial-correlation of radio propagation characteristics between different wireless links.  Using Kriging-based shadowing estimation, the radio propagation of the wireless link that has arbitrary location relationship can be predicted. Numerical results show that the proposed method achieves higher estimation accuracy than path loss-based estimation methods.
 The methods discussed in this thesis can develop more spatial WSs in existing allocated bandwidth such as TVWS, and can provide these WSs to new wireless systems expected to appear in the future.  Additionally, these results will contribute not only to such spectrum sharing but also to improvement of the spectrum management in existing systems.  For example, in heterogeneous networks (HetNets), a suitable inter-cell interference management enables transmitters to reuse the frequency efficiently and the user equipment can select the optimum base station. We anticipate that this dissertation strongly contributes to improving
the spectrum utilization efficiency of the whole wireless systems.},
 school = {電気通信大学},
 title = {Measurement-based Spectrum Database for Spatial Spectrum Sharing},
 year = {},
 yomi = {サトウ, コウヤ}
}