Coexistence between 5G cellular networks and incumbent radar systems is necessary for an increasing number of spectral bands, including highly valuable spectrum such as the C-band. This paper presents a novel coexistence framework that intelligently adjusts 5G antenna parameters to mitigate interference reaching known radar systems, while simultaneously maximizing cellular network performance. The framework leverages Gaussian process regression and differential evolution to navigate high-dimensional, non-convex spaces while effectively managing uncertainty. We propose a practical approach that utilizes user RSRP measurements to characterize communication interference on radar, addressing the non-cooperative nature of radar systems. Evaluation on AT&T Labs' high-fidelity simulator demonstrates over a 12% increase in sum-log-rate and around a 3.6 dB increase in median SINR compared to the exhaustive search with common parameter configurations across all base stations, while decreasing interference on radar to its lowest achievable level in our simulation setup.

This work investigates the joint optimization of coverage, capacity, and cell load by tuning several cell-specific antenna and cell association parameters via data-driven methods. We are particularly focused on the complexities of macrocell and small cell coexistence, and demonstrate an automated learning method whereby macrocells and small cells can strategically adapt their coverage areas. Coupled with adaptive offloading using a tunable small cell bias, we demonstrate significant throughput and coverage improvement in a realistic 5G network simulator developed by AT&T Labs. Concretely, we formulate an optimization problem to maximize network coverage and the application-layer data rate experienced by users, accounting for delays from congestion, cell loading, and packet retransmissions. We propose an algorithm that approaches the optimum via Gaussian process models and the evolutionary search: efficiently navigating the high-dimensional, nonconvex space while managing uncertainty. Our results show that the joint optimization of antenna tuning and load balancing - exemplified by load-aware cell shaping - more than doubles the cell edge throughput and increases the cell edge SINR by 8 dB, compared to bias-only optimization. Furthermore, our algorithm and overall approach appear viable for implementation.

We propose a novel framework for optimizing antenna parameter settings in a heterogeneous cellular network. We formulate an optimization problem for both coverage and capacity– in both the downlink (DL) and uplink (UL)– which configures the tilt angle, vertical half-power beamwidth (HPBW), and horizontal HPBW of each cell’s antenna array across the network. The novel data-driven framework proposed for this nonconvex problem, inspired by Bayesian optimization (BO) and differential evolution algorithms, is sample-efficient and converges quickly, while being scalable to large networks. By jointly optimizing DL and UL performance, we take into account the different signal power and interference characteristics of these two links, allowing a graceful trade-off between coverage and capacity in each one. Our experiments on a state-of-the-art 5G NR cellular system-level simulator developed by AT&T Labs show that the proposed algorithm consistently and significantly outperforms the 3GPP default settings, random search, and conventional BO. In one realistic setting, and compared to conventional BO, our approach increases the average sum-log-rate by over 60% while decreasing the outage probability by over 80%. Compared to the 3GPP default settings, the gains from our approach are considerably larger. The results also indicate that the practically important combination of DL throughput and UL coverage can be greatly improved by joint UL-DL optimization.

In this paper, we jointly optimize the capacity and coverage of both uplink and downlink transmissions by tuning the downtilt angle, vertical half-power beamwidth (HPBW), and horizontal HPBW of each cell's antenna array across a heterogeneous cellular network. We formulate an optimization problem and propose a novel sample-efficient algorithm to solve this non-convex problem. We evaluate our framework on a state-of-the-art cellular system-level simulator developed by AT&T Labs by comparing it with the 3GPP baseline. Example results tuned to optimize uplink coverage and downlink rate indicate that jointly optimizing the uplink and downlink directions improves uplink median and 5% outage SINR by (i) 1.6 dB and 4.5 dB, respectively, compared to downlink only-optimization and by (ii) 6.7 dB and 14.6 dB compared to the 3GPP baseline. Simultaneously, we can increase downlink median and outage SINR by comparable amounts compared to uplink-only optimization, but with larger gains in median SINR and downlink sum-rate. Our results indicate that there are significant gains to be harvested from site-specific data-driven base station parameter optimization, and they can be achieved in a scalable and automated fashion.

Outdated and inaccurate channel state information (CSI) prevents the transmitter from adapting to current channel conditions and impairs reliable communication. This paper proposes a deep learning-based methodology for real-time prediction of future CSI, which enables it to be fed back in advance and mitigates the effect of the feedback and processing delay. Predicting the future CSI at the user equipment may also help the gNodeB (gNB) send reference signals less frequently and may help reduce the reference signal transmission overhead. Our approach utilizes Long Short-Term Memory (LSTM) type recurrent neural networks (RNN) to predict the CSI from past channel estimations. However, the knowledge of past channel gains is assumed to be limited, and only partial channel data is available. We propose a new algorithm that utilizes LSTMs and remedies incomplete channel history problem, which we denote by RNN-I. We evaluate the performance of the proposed framework by comparing it with two conventional techniques: persistence and auto-regressive (AR) model-based prediction methods. The results show that the proposed RNN-I algorithm performs well under strong channel mismatch scenarios, while AR methods diverge and can not keep up with fast varying channels if the mismatch in the channel characteristics used in training and inference is significant.

Finding an optimum configuration of base station (BS) antenna parameters is a challenging, non-convex problem for cellular networks. The chosen configuration has major implications for coverage and throughput in real-world systems, as it effects signal strength differently throughout the cell, as well as dictating the interference caused to other cells. In this paper, we propose a novel and sample-efficient data-driven methodology for optimizing antenna downtilt angles. Our approach combines Bayesian optimization (BO) with Differential Evolution (DE): BO decreases the computational burden of DE, while DE helps BO avoid the curse of dimensionality. We evaluate the performance on a realistic state-of-the-art cellular system simulator developed by AT&T Labs, that includes all layers of the protocol stack and sophisticated channel models. Our results show that the proposed algorithm outperforms Bayesian optimization, random selection, and the baseline settings adopted in 3GPP by nontrivial amounts in terms of both capacity and coverage. Also, our approach is notably more time-efficient than DE alone.