Andreas F. Molisch’s research while affiliated with University of Southern California and other places

Performance of video streaming, which accounts for most of the traffic in wireless communication, can be significantly improved by caching popular videos at the wireless edge. Determining the cache content that optimizes performance (defined via a revenue function) is thus an important task, and prediction of the future demands based on past history can make this process much more efficient. However, since practical video caching networks involve various parties (e.g., users, isp, and csp) that do not wish to reveal information such as past history to each other, privacy-preserving solutions are required. Motivated by this, we propose a proactive caching method based on users' privacy-preserving multi-slot future demand predictions -- obtained from a trained Transformer -- to optimize revenue. Specifically, we first use a privacy-preserving fl algorithm to train a Transformer to predict multi-slot future demands of the users. However, prediction accuracy is not perfect and decreases the farther into the future the prediction is done. We model the impact of prediction errors invoking the file popularities, based on which we formulate a long-term system revenue optimization to make the cache placement decisions. As the formulated problem is NP-hard, we use a greedy algorithm to efficiently obtain an approximate solution. Simulation results validate that (i) the fl solution achieves results close to the centralized (non-privacy-preserving) solution and (ii) optimization of revenue may provide different solutions than the classical chr criterion.

Channel models that represent various operating conditions a communication system might experience are important for design and standardization of any communication system. While statistical channel models have long dominated this space, machine learning (ML) is becoming a popular alternative approach. However, existing approaches have mostly focused on predictive solutions to match instantaneous channel realizations. Other solutions have focused on pathloss modeling, while double-directional (DD) channel representation is needed for a complete description. Motivated by this, we (a) develop a generative solution that uses a hybrid Transformer (hTransformer) model with a low-rank projected attention calculation mechanism and a bi-directional long short-term memory (BiLSTM) layer to generate complete DD channel information and (b) design a domain-knowledge-informed training method to match the generated and true channel realizations' statistics. Our extensive simulation results validate that the generated samples' statistics closely align with the true statistics while mostly outperforming the performance of existing predictive approaches.

Channel models that represent various operating conditions a communication system might experience are important for design and standardization of any communication system. While statistical channel models have long dominated this space, machine learning (ML) is becoming a popular alternative approach. However, existing approaches have mostly focused on predictive solutions to match instantaneous channel realizations. Other solutions have focused on pathloss modeling, while double-directional (DD) channel representation is needed for a complete description. Motivated by this, we (a) develop a generative solution that uses a hybrid Transformer (hTransformer) model with a low-rank projected attention calculation mechanism and a bi-directional long short-term memory (BiLSTM) layer to generate complete DD channel information and (b) design a domain-knowledge-informed training method to match the generated and true channel realizations' statistics. Our extensive simulation results validate that the generated samples' statistics closely align with the true statistics while mostly outperforming the performance of existing predictive approaches.

This paper introduces AutoBS, a reinforcement learning (RL)-based framework for optimal base station (BS) deployment in 6G networks. AutoBS leverages the Proximal Policy Optimization (PPO) algorithm and fast, site-specific pathloss predictions from PMNet to efficiently learn deployment strategies that balance coverage and capacity. Numerical results demonstrate that AutoBS achieves 95% for a single BS, and 90% for multiple BSs, of the capacity provided by exhaustive search methods while reducing inference time from hours to milliseconds, making it highly suitable for real-time applications. AutoBS offers a scalable and automated solution for large-scale 6G networks, addressing the challenges of dynamic environments with minimal computational overhead.

Recent advances in deep learning have successfully replaced classical algorithms with machine learning models based on neural networks (NNs). This is particularly prevalent in proactive caching. As NNs grows more capable as their size in terms of storage and computation increases, NN-based proactive caching achieves performance improvement. Nonetheless, there remain challenges in implementing NN-based proactive caching in realistic environments with dynamic user movement. These are due to the fixed structure of NNs that should expand the input size to match the dimensions of the input with the dimensions of the NN’s input units. To address these challenges, this paper proposes a scalable proactive caching framework, named slimmable federated reinforcement learning (SlimFRL). By adopting slimmable neural networks (SNNs) in FRL, our SlimFRL easily adjusts the widths of the SNNs during training according to the number of users. Moreover, due to the scalability of SNNs, our SlimFRL can set the appropriate input dimension while not using imputation, leading to performance improvement. This paper also validates the performance and advantages of SlimFRL in terms of reward and additional cost functions. Additionally, this paper proposes several training algorithms for SlimFRL and corroborates their superiority with convergence analysis and various experiments.

Recent interest in Device-to-Device (D2D) communication systems, particularly by Public Safety Organizations (PSOs), arises from scenarios where no infrastructure is available. One notable scenario is Indoor-to-Outdoor (I2O), where emergency responders inside buildings communicate with command posts on the street. We aim to understand the propagation channel for designing wireless systems in such contexts. We report findings from a comprehensive measurement campaign in the public safety frequency band near 800 MHz, assessing multiple-input-multiple-output (MIMO) channels between users across five floors of a California office building and a multi-antenna base station at street level. This study provides insights as well as statistical channel models for pathgain, delay spread, angular spreads, and power distribution among Multi-Path Components (MPCs) and their dependence on indoor-user height. In Line-of-Sight (LoS) scenarios, 10 dB pathgain variation was observed between ground and the 5th floor, while Non-Line-Of-Sight (NLoS) cases showed no such variation. Delay spreads exhibited a similar trend in LoS scenarios, with a 10 ns variation based on RX height, while NLoS scenarios remained heightindependent. Angular spreads were consistently large for indoor units, regardless of height. Analysis of the probability density function of the MPC powers using a 5D κ parameter revealed approximately 8 dB for LoS scenarios and −1 dB for NLoS scenarios, assuming beamforming for maximum received power.

Cell-free massive MIMO (CF-mMIMO), where each user equipment (UE) is connected to multiple access points (APs), is emerging as an important component for fifth-generation (5G) and sixth-generation (6G) cellular systems. Accurate channel models based on measurements are required to optimize the design and deployment of such systems. This paper presents an extensive measurement campaign for CF-mMIMO in an urban environment. A new “virtual AP” technique measures channels between 80 UE locations and more than 20, 000 possible microcellular AP locations. Measurements are done at 3.5 GHz carrier frequency with 350 MHz bandwidth (BW). The paper describes the measurement setup and data processing, shows sample results and their physical interpretation, and provides statistics for key quantities such as pathloss, shadowing, delay spread (DS), and delay window. We find pathloss coefficients of 2.9 and 10.4 for line-of-sight (LOS) and non line-of-sight (NLOS), respectively, where the high LOS coefficient is mainly because larger distance leads to more grazing angle of incidence and thus lower antenna gain in our setup. Shadowing standard deviations are 5.1/16.6 dB, and root mean squared (RMS) DSs of −80.6/−72.6 dBs¹. The measurements can also be used for parameterizing a certain type of channel model, namely Cell-free massive MIMO for Urban Non-stationary Environment with Correlations (CUNEC), which will be reported in future work.

Growing demand for high data rates is driving interest in the upper mid-band (FR 3) spectrum (6-24 GHz). While some propagation measurements exist in literature, the impact of vegetation on link performance remains under-explored. This study examines vegetation-induced losses for an urban scenario across 6-18 GHz. A simple method for calculating vegetation depth is introduced, along with a model that quantifies additional attenuation based on vegetation depth and frequency, divided into 1 GHz sub-bands. We see that excess vegetation loss increases with vegetation depth and higher frequencies. These findings provide insights for designing reliable, foliage-aware communication networks in FR 3.

The upper midband, designated as Frequency Range 3 (FR3), is increasingly critical for the next-generation of wireless networks. Channel propagation measurements and their statistical analysis are essential first steps towards this direction. This paper presents a comprehensive ultra-wideband (UWB) double-directional channel measurement campaign in a large portion of FR3 (6-14 GHz) for urban microcellular environments. We analyze over 25,000 directional power delay profiles and providing key insights into line-of-sight (LoS) and obstructed line-of-sight (OLoS) conditions. This is followed by statistical modeling of path loss, shadowing, delay spread and angular spread. As the first UWB double-directional measurement campaign in this frequency range, this work offers critical insights for spectrum allocation, channel modeling, and the design of advanced communication systems, paving the way for further exploration of FR3.

Growing demand for high data rates is driving interest in the upper mid-band (FR 3) spectrum (6-24 GHz). While some propagation measurements exist in literature, the impact of vegetation on link performance remains under-explored. This study examines vegetation-induced losses in an urban scenario across 6-18 GHz. A simple method for calculating vegetation depth is introduced, along with a model that quantifies additional attenuation based on vegetation depth and frequency, divided into 1 GHz sub-bands. We see that excess vegetation loss increases with vegetation depth and higher frequencies. These findings provide insights for designing reliable, foliage-aware communication networks in FR 3.