Massive 안테나 기반 캐스케이드 도래각 추정 알고리즘에 관한 연구

Abstract

Estimating the accurate Angle-of-Aarrival (AOA) of a signal is one of the important issues of a modern wireless communication system based on beamforming. AOA of the signal is collected through an array antenna mounted on various systems existing on the ground, in the air, in space, etc. In order to satisfy the physical specifications of the system, various types of array antennas have been proposed, and an array antenna that improves the disadvantages of various existing array antennas is being actively studied. Various AOA estimation algorithms have been proposed as much as various array antennas, and AOA estimation algorithms are largely divided into a conventional method called a beam scan method and a subspace based method. In general, the AOA estimation performance of the subspace based method is superior to that of the conventional method. However, in the case of the subspace based AOA estimation algorithm, since there is no prior AOA information, the entire range can be searched or only a specific type of array antenna can be applied, which leads to poor operability. To solve this problem, in this dissertation, a cascade AOA estimation algorithm that can apply various array antennas while having AOA estimation performance similar to that of the subspace based AOA estimation algorithm is proposed. The proposed cascade AOA estimation algorithm consists of the Capon algorithm and the Beamspace MUSIC algorithm, and uses the two algorithms sequentially. Although the resolution of the Capon algorithm in the proposed algorithm is somewhat lower, the complexity is lower than that of the subspace-based AOA estimation algorithm, enabling fast AOA group estimation. Next, Beamspace MUSIC is applied to estimate the detailed signal AOA in the AOA group estimated through the Capon algorithm. Beamspace MUSIC algorithm has a very high resolution, although the complexity is rather high because beamspace processing is preceded. To evaluate the AOA estimation performance of the proposed cascade algorithm, computer simulations based on various scenarios are provided. The MUSIC algorithm, which has excellent resolution and can apply various array antennas, is the most representative algorithm used in the AOA estimation. However, if the system uses a Massive Array (MA) antenna in which multiple antenna elements are used, the multiple antenna elements are increases the computational complexity to the extent that real time estimation is difficult in case of MUSIC algorithm. In order to solve the problem of computational complexity of AOA estimation algorithms that may occur in MA antennas where multiple antenna elements are used, in this dissertation, a cascade AOA estimation algorithm based on a Flexible Massive Rectangular Array (FMRA) antenna is proposed. The FMRA antenna based cascade algorithm has the same basic structure as the cascade algorithm applying the general arrangement presented above. In order to reduce the algorithm complexity while maintaining the AOA estimation performance, the FMRA antenna based cascade algorithm estimates the approximate signal AOA group by activating some antenna elements among the entire antenna elements using Capon algorithm. The detailed signal AOA of the AOA group is estimated using the Beamspace MUSIC algorithm using all antenna elements. By selectively turning on/off antenna elements, it is possible to significantly reduce the computational complexity of algorithms that may occur due to multiple antenna elements, as well as dramatically reduce the computational complexity of algorithms according to the search range. A computer simulation for AOA estimation performance comparison of the proposed FMRA based cascade algorithm is provided, and a mathematical model for complexity analysis is presented. Based on the presented mathematical model, a computer simulation for comparison of computational complexity is provided. In the previously proposed cascade algorithm using an FMA antenna, a beamspace transformation matrix with a fixed size that is twice as large as the number of signals in a group is applied. This causes an unnecessary amount of calculation when estimating the AOA. To solve this problem, in this dissertation, a cascade AOA estimation algorithm based on an FMLA (Flexible Massive Linear Array) antenna to which an optimized beamspace transformation matrix is applied is suggested. For beamspace transformation matrix optimization, a beamspace MDL algorithm for estimating the number of signals is inserted between the Capon algorithm and the Beamspace MUSIC algorithm. Through beamspace transformation matrix optimization, the AOA can be efficiently estimated by optimizing the beamspace applied to the Beamspace MUSIC algorithm. Computer simulations based on various scenarios are provided to compare the AOA estimation performance and computational complexity.