Research On Cooperative Spectrum Sensing Algorithms For Dense Cognitive Network

Dense heterogeneous network is envisioned to efficiently enhance network capacity which benefits from spectrum spatial reuse.However,one of the key bottlenecks is that the complex interference in network results in damage of spectrum efficiency.By way of regarding macro cells as primary network and applying cognitive radio technology to dense small cells would enable to reuse licensed spectrum resources of macro cells and overcome the co-channel interference between macro cells and small cells.The network consisting of dense small cells with cognitive ability is called dense cognitive network.Spectrum sensing is the basic function of cognitive radio technology and traditional spectrum sensing algorithms usually utilize sample covariance matrix to estimate unknown covariance matrix.However,dense deployment of small cells causes high-dimensional characteristic of sensing data and changes statistical characteristics of the received signal which increases estimated error of sample covariance matrix in high-dimensional scene.Therefore,the performance of traditional spectrum sensing algorithms degrades and it is necessary to study high-dimensional spectrum sensing algorithms for dense cognitive network.The paper is supported by the National Science Funding project:"Sensing and utilization of polarization resource in dense heterogeneous wireless environment"(No.61571062).In this paper,high-dimensional spectrum sensing algorithms are studied for dense cognitive network based on high-dimensional estimated theory.Estimated eigenvalues based cooperative spectrum sensing algorithm and estimated covariance matrix based cooperative spectrum sensing algorithms are proposed respectively,and the main research contents are as follows:1.Key technologies and applicable scenarios of cognitive network and dense network are summarized respectively.Then,existing independent spectrum sensing algorithms and cooperative spectrum sensing algorithms are classified and summarized by analyzing and comparing their advantages,disadvantages and applicable scenarios.Finally,the paper concludes existing high-dimensional spectrum sensing algorithms for dense cognitive network and points out their problems.2.Considering that dense small cells cooperate to sense primary signal transmitted by single antenna of macro cell,consistent-estimated eigenvalues based cooperative spectrum sensing algorithm is proposed aiming at the problem that estimated performance of sample eigenvalues for eigenvalues of high-dimensional covariance matrix degrades.Proposed algorithm utilizes well-estimated consistent estimators of eigenvalues to solve the estimated problem of eigenvalues of high-dimensional covariance matrix.Furthermore,algorithm achieves better sensing performance by using the rank-one structure of transmitted signal.Finally,theoretical derivation and simulation results show that proposed algorithm achieves better sensing performance when eigenvalue splitting condition is satisfied and it exhibits 1.9dB better performance than existing high-dimensional spectrum sensing algorithms.Therefore,proposed algorithm can be applied to sense primary signal transmitted by single antenna of macro cell so as to realize the coexistence with the macro cell network.3.Considering that dense small cells cooperate to sense primary signal transmitted by multiple antennas of macro cell,two oracle approximating shrinkage estimator based cooperative spectrum sensing algorithms are proposed in the case of ideal and non-ideal noise aiming at the problem that estimated performance of sample covariance matrix for high-dimensional covariance matrix degrades.Two algorithms solve the estimated problem of high-dimensional covariance matrix by utilizing well-estimated oracle approximating shrinkage estimator and improve sensing performance.Furthermore,the paper analyzes expression of false alarm probability and threshold of the proposed algorithm under ideal noise.Finally,in case of ideal noise,simulation results show that proposed algorithm respectively exhibits 1.2dB and 0.9dB better performance than existing high-dimensional spectrum sensing algorithms and consistent-estimated eigenvalues based cooperative spectrum sensing algorithm and it is valid when sample dimension is larger than number of sample.In case of non-ideal noise,proposed algorithm exhibits 1.2dB better performance than traditional spectrum sensing algorithm considering non-ideal noise and it is valid when sample dimension is larger than number of sample.Therefore,proposed two algorithms can be respectively applied to sense primary signal transmitted by multiple antennas of macro cell base station for case of ideal noise and non-ideal noise so as to realize the coexistence with the macro cell network.