10 June 2022 in IEEE Transactions on Intelligent Transportation Systems
As important services of the future sixth-generation (6G) wireless networks, vehicular communication and mobile edge computing (MEC) have received considerable interest in recent years for their significant potential applications in intelligent transportation systems. However, MEC-enabled vehicular networks depend heavily on network access and communication infrastructure, often unavailable in remote areas, making computation offloading susceptible to breaking down. To address this issue, we propose an MEC-enabled vehicular network assisted through aerial-terrestrial connectivity to provide network access and high data-rate entertainment services to a vehicular network. We present a time-varying, dynamic system model where high altitude platforms (HAPs) equipped with MEC servers, connected to a backhaul system of low-earth orbit (LEO) satellites, are used to provide computation offloading capability to the vehicles, as well as to provide network access for vehicle-to-vehicle (V2V) communications. Our main objective is to minimize the total computation and communication overhead of the joint computation offloading and resource allocation strategies for the system of vehicles. Since our formulated optimization problem is a mixed-integer non-linear programming (MINLP) problem, which is NP-hard, we propose a decentralized value-iteration-based reinforcement learning (RL) approach as a solution. In our Q-learning-assisted analysis, each vehicle acts as an intelligent agent to form optimal strategies for offloading and resource allocation. We further extend our solution to deep Q-learning (DQL) and double deep Q-learning to overcome the issues of dimensionality and the over-estimation of the value functions, as in Q-learning. Simulation results prove the effectiveness of our solution in successfully reducing system costs compared to baseline schemes.
18 April 2024 in IEEE Transactions on Wireless Communications
The emergence of fluid antenna systems (FAS) offers a novel technique for obtaining spatial diversity and leveraging interference fades for spectrum sharing in multiuser scenarios—a paradigm referred to as fluid antenna multiple access (FAMA). Nevertheless, as the number of users increases, the interference mitigation capability diminishes. To overcome this, opportunistic scheduling that prioritizes robust users proves to be an effective method for enhancing FAMA. This paper introduces a resilient decentralized reinforcement learning (RL) approach for opportunistic FAMA (O-FAMA), to autonomously select robust users and the port of each chosen user’s FAS jointly to maximize the network sum-rate. In order to enhance learning efficiency in this multi-agent environment, we propose a novel team-theoretic RL framework that includes a derivative network guiding the multi-agent learning of each solution’s policy networks. Our simulation results confirm the effectiveness of the proposed methodology.
16 September 2025 in IEEE Transactions on Wireless Communications
Based on our current understanding, extreme massive access over the same physical channel is only possible if an extra-large multiple-input multiple-output (XL-MIMO) antenna is used at the base station (BS) and instantaneous channel state information (CSI) is known at the BS side for precoding design. This casts doubt on scalability and challenges in device-to-device situations in which there is not a centralized, optimized BS for transmitting the user signals. To address this problem, we revisit the massive connectivity challenge by considering the case where no CSI is available at the BS and no precoding is used. In this situation, inter-user interference (IUI) mitigation can only be performed at the user terminal (UT) side. Leveraging the position flexibility of fluid antenna system (FAS), we adopt a fluid antenna multiple access (FAMA) approach that exploits the interference signal fluctuation in the spatial domain. Specifically, we assume that we have N spatially correlated received signals per symbol duration from FAS. Our main approach uses a simple heuristic port shortlisting method that identifies promising ports to obtain favourable received signals that can be combined via maximum ratio combining (MRC) to form the received output signal for final detection. On top of this, a pre-trained deep joint source-channel coding (JSCC) scheme is employed, which together with a diffusion-based denoising model (MixDDPM) at the UT side, can improve the IUI immunity. We refer to the proposed scheme as turbo FAMA. Simulation results show that with a physical FAS size of 20 wavelengths at each UT transmitting quaternary phase shift keying (QPSK) symbols, fast FAMA can support 50 users while turbo FAMA can handle up to 200 users if the required symbol error rate (SER) is 10-2. If a higher error tolerance is acceptable, say SER at 0.1, turbo FAMA can even serve up to 1000 users but fast FAMA is only able to handle 160 users, all remarkably achieved without CSI at the BS.
17 March 2026 in IEEE Open Journal of the Communications Society
Future wireless networks must deliver unprecedented spectral efficiency to accommodate massive numbers of interconnected devices; yet today’s multiple-access schemes remain fundamentally constrained by interference. Fluid antenna multiple access (FAMA) has emerged as a promising paradigm that tackles this bottleneck by introducing a new degree of freedom (DoF): receiver-side spatial fluidity enabled by fluid antenna systems (FAS). By dynamically repositioning the FAS aperture to exploit fine-grained channel variations, FAMA can mitigate inter-user interference and enable massive multiple access on the same time-frequency resource without precoding and any other interference suppression techniques. This paper presents a comprehensive survey of recent advances in FAMA. We first summarize the operating principles of FAS and then formalize FAMA, including representative network and channel models. We systematically categorize and review FAMA architectures—fast FAMA, slow FAMA, compact ultra-massive antenna arrays (CUMA), coded FAMA, and turbo FAMA—highlighting their design rationales, performance benefits, and applicability regimes. We further discuss practical considerations such as channel acquisition and tracking, hardware constraints, switching/positioning overhead, and integration with existing air interfaces. Finally, we outline open challenges and future research directions, including scalable resource allocation, robustness under mobility and correlation, protocol/standardization aspects, and cross-layer co-design. Overall, this survey provides a unified framework to understand FAMA’s potential and to guide both academic research and industrial innovation in this rapidly evolving area.
17 February 2026 in IEEE Wireless Communications Letters
Fluid antenna multiple access (FAMA) has recently emerged as a user-centric approach to massive connectivity. By exposing many reconfigurable “ports” on an antenna aperture, they exploit local spatial fading fluctuations at each user terminal (UT) to mitigate multiuser interference (MUI) without the need for precoding. While both turbo and fast FAMA promise ultra-massive access, its practicality is limited by the need to observe, at symbol rate, the complete set of per-port channel gains and received signals. This letter addresses this bottleneck with a novel plug-and-play attentional-copula extrapolator that reconstructs the joint per-symbol field of channel gains and received signals from a small subset of ports. The module couples monotone normalizing-flow marginals with a Transformer-based copula to capture both intra-port (channel-signal) and inter-port (spatial) dependencies, yielding drop-in estimates for existing turbo FAMA pipelines without retraining. Simulation results reveal that even with an extremely small port subset, the normalized mean-square error (NMSE) rapidly converges and the symbol error rate (SER) after turbo FAMA remains within a few percent of the full-channel state information (CSI) oracle across aperture sizes. By replacing exhaustive sampling with accurate learned extrapolation, the proposed scheme preserves turbo FAMA’s scalability while unlocking practical, low-overhead deployment—a viable pathway to extreme massive access.