Loading [a11y]/accessibility-menu.js
Location Aided and Machine Learning-Based Beam Allocation for 3D Massive MIMO Systems | IEEE Conference Publication | IEEE Xplore

Location Aided and Machine Learning-Based Beam Allocation for 3D Massive MIMO Systems


Abstract:

3D massive multiple-input multiple-output (MIMO) technology is considered as one of the key technologies of 5G. Compared with traditional 2D massive MIMO systems, the ele...Show More

Abstract:

3D massive multiple-input multiple-output (MIMO) technology is considered as one of the key technologies of 5G. Compared with traditional 2D massive MIMO systems, the elevation angles of signal propagation are introduced in 3D massive MIMO systems, which make it fully utilize the advantages in horizontal and vertical dimensions at the same time, further increase the freedom of scheduling and resource allocation, and significantly improve the system capacity. To improve the system performance further, beamforming can form high-gain beams to reduce signal interference between user equipments (UEs). Traditionally, beam allocation is regarded as an optimization problem. Since most beam allocation problems are non-convex, it is difficult to obtain an optimal solution in real time. With the rise of artificial intelligence technology, machine learning has become one of the most promising tools. In order to optimize the beam allocation better in a short time, a location aided and machine learning based beam allocation algorithm (LMLBAA) is proposed in 3D massive MIMO systems. The algorithm absorbs the idea of k-NN or SVM to construct the multi-classifier, which uses collected location information of UEs as feature vectors and corresponding precoding codeword indexes as classes. When a new UE joins up, the base station (BS) will select a suitable codeword for the UE to perform precoding and form the corresponding serving beam according to the decision result of the multi-classifier. Simulation results show that the more location information of UEs is collected during the initialization process, the better the performance of the proposed algorithm will be. Meanwhile, as the transmit signal-to-noise ratio (SNR) increases, the average available sum rate will increase and approach the results of the exhaustive search method.
Date of Conference: 24-28 June 2019
Date Added to IEEE Xplore: 22 July 2019
ISBN Information:

ISSN Information:

Conference Location: Tangier, Morocco

Contact IEEE to Subscribe

References

References is not available for this document.