16 October 2023 in IEEE Access
Accurate Channel State Information (CSI) is critical for maximizing the throughput of massive Multi-Input Multi-Output (mMIMO) systems. Due to the environment dynamics and user mobility, CSI aging is a major challenge to achieving the large throughput of mMIMO promised by theory. CSI prediction can be used to overcome this without increasing the signaling overhead. Motivated by the anticipated native support for Artificial Intelligence (AI) in the fifth generation and beyond cellular standards, we propose deep learning CSI prediction solutions based on 3-Dimensional (3D) Complex Convolutional Neural Networks (CCNN). These solutions provide improved capabilities for capturing temporal and spatial correlations, enhancing CSI prediction performance. In particular, they utilize the angle delay decomposition of previously observed CSI to predict the future one. In one architecture, the network, dubbed CSI Prediction Network (CSI-PNet), uses small kernels with circular padding to efficiently capture the correlation between propagation paths in the angle domain. This architecture can be further improved by the use of an attention-like model to vary the weights and enhance prediction performance adaptively. We also propose methods to enhance robustness to noise and time and frequency offsets. We tested these solutions using 3GPP-compatible simulations and field measurements in a commercial network. Our solutions demonstrate stable performance and significantly outperform several benchmarks, especially at low and medium speeds. They strike a balance between performance and architecture complexity, indicating suitability for actual implementation.
26 September 2017 in IEEE Transactions on Vehicular Technology
In this paper, we study the uplink performance in a 3D urban micro (3D-UMi) cell. The single-antenna users are distributed according to a 3D Poisson point process inside the cell. Each user has a probability to experience a line-of-sight (LoS) channel or a fading channel, and transmits to the multiantenna base station (BS) utilising fractional power control. The BS deploys a large 2D planar antenna array and performs 3D beamforming (BF) to receive the signals of all users in the cell sector. Under such a setting, we study the interference incurred by the co-channel users in the cell sector to a randomly located target user. In particular, we show that the interference incurred by the LoS interferers to a LoS target user is associated with the 3D BF antenna pattern. This type of interference is determined by the separation between the locations of the target user and interferer, and hence it is heavily dependent on the cell geometry. We exploit the massive MIMO regime and present a good approximation to model the LoS interference. We further resort to stochastic geometry to analyse the coverage probability and the rate per user in the cell.