Advanced Air Mobility (AAM) seeks to establish a next-generation air transportation system by leveraging autonomous unpiloted aerial vehicles (UAVs) to transport passengers and cargo between locations previously underserved or unserved by traditional aviation. Achieving AAM at scale requires overcoming significant challenges in airspace management, classification, and traffic control to safely accommodate the increasing volume of UAV operations. This paper presents a comprehensive design for air corridors to facilitate efficient aerial transport and formulates a multi-UAV coordination problem within these corridors. The objective is to enable each UAV to autonomously make control decisions based on local observations gathered from onboard sensors. This decentralized control approach is modeled as a multi-agent partially observable Markov decision process (POMDP), aiming at minimizing UAV travel time while ensuring adherence to corridor boundaries and collision avoidance. To address the complexities posed by varying state dimensions and types, we propose a novel Hybrid Transformer-based Multi-agent Reinforcement Learning (HTransRL) architecture. HTransRL integrates a customized transformer model into an actor-critic network, effectively processing both sequential and non-sequential observed states of varying sizes while capturing their correlations. This enables safe and efficient UAV navigation. Simulation results show that in test environments similar to or simpler than training scenarios, HTransRL achieves a successful arrival rate exceeding 90% in worst-case test scenarios. In test environments more complex than training scenarios, HTransRL demonstrates superior scalability compared to two baseline methods, achieving higher arrival rates and comparable travel times.

To reap the benefit of big data generated by the massive number of Internet of Things (IoT) devices while preserving data privacy, federated learning (FL) has been proposed to enable IoT devices to train machine learning models locally. That is, instead of sharing the local data sets, different clients in terms of IoT devices only need to upload their local models to a centralized FL server. Client selection in FL is critical to maximize the number of qualified clients, who can successfully upload their local models to the FL server before the predefined deadline. Normally, client selection is coupled with wireless resource management owing to the fact that different clients need to share the same spectrum to upload their local models. The existing solutions of joint optimizing client selection and resource management are designed based on frequency-division multiple access (FDMA) or time-division multiple access (TDMA), which do not consider the dynamics of the clients and lead to low bandwidth utilization. In this article, we propose the Nonorthogonal-Multiple-Access (NOMA)-based resource allocation for client selection in FL to dynamically and jointly optimize client selection for each global iteration as well as the transmission power of each selected client in each time slot within a global iteration. We design the deep-reinforcement-learn-based client selection in NOMA-based federated learning (DREAM-FL) algorithm to solve the problem. Extensive simulations are conducted to demonstrate that DREAM-FL can select more qualified clients and has higher model accuracy than FDMA and TDMA-based solutions.

Free space optics (FSO), which uses light as the carrier to transmit data in free space, has been demonstrated as a secure and high-speed solution for long distance and line-of-sight wireless communications. Applying FSO as fronthaul/backhaul communications between base stations (BSs) and the gateway can significantly increase the fronthaul/backhaul link capacity. Traditionally, the gateway has to be equipped with multiple FSO transceivers, each of which is used to communicate with a BS by establishing a dedicated FSO. In this paper, we propose to use a steerable FSO system, where the gateway is equipped with a steerable FSO transceiver to communicate with multiple FSO transceivers at different BSs in a time division multiplexing manner. Applying the steerable FSO system can reduce the number of FSO transceivers at the gateway, and thus reduce the capital cost of implementing an FSO based fronthaul/backhaul network. We formulate the transmission scheduling problem in the steerable FSO system to optimize the active time for each FSO link associated with the steerable FSO transceiver such that the overall delay of transmitting a packet from geo-distributed BSs to the steerable FSO transceiver at the gateway is minimized, while guaranteeing the latency requirements of the BSs. We propose the laTency aWare transmIssionScheduling for sTeerable FSO (TWIST) algorithm, which is designed based on Sequential Quadratic Programming, to efficiently solve the proposed problem. The performance of TWIST is validated via extensive simulations.

Federated learning (FL) is a collaborative machine learning framework to enable different clients such as Internet of Things (IoT) devices to participate in a machine learning model training process, while preserving data privacy. Client selection is critical to determine the performance of FL. Most of the existing client selection methods aim to maximize the number of selected clients, who can upload their local models before the deadline, in each global iteration, thus potentially accelerating the model convergence rate. However, these methods ignore the fact that most of the IoT devices are powered by on-board batteries and harvested green energy from the environment to prolong battery life. Hence, clients selected by these methods may not have sufficient energy to upload their local models in a global iteration or are unable to participate in the training process in the near future due to battery drainage. In this paper, we propose a novel client selection, entitled “EnerGy-AwaRe CliEnt SElection for Green FeDerated Learning (GREED)”, to optimize the trade-off between maximizing the number of selected clients and minimizing the energy drawn from batteries for the selected clients, while ensuring that all the selected clients have sufficient energy to upload their local models before the deadline. The performance of GREED is validated via extensive simulations.

Federated learning (FL) has been proposed to efficiently and privacy-preserving distributed machine learning architecture for the Internet of Things (IoT). In a wireless FL system, clients in IoT devices train their local models over the local data sets. The derived local models are uploaded to an FL server to generate a global model, broadcasted to the clients in the next global iteration for further training. Owing to the heterogeneous feature of the clients, client selection is critical to determine the overall training time. Traditionally, the objective of client selection is to select the maximum number of clients who can derive and upload their local models before the deadline in each global iteration. However, selecting more clients increases the energy consumption of the clients. Moreover, selecting the maximum number of clients is unnecessary as having fewer clients in early global iterations and more clients in later global iterations have been proved to achieve higher model accuracy. Hence, this article proposes to dynamically adjust and optimize the tradeoff between maximizing the number of selected clients and minimizing the total energy consumption of the clients by selecting suitable clients and allocating appropriate resources in terms of CPU frequency and transmission power. We formulate the joint client selection and resource management problem and design the energy and latency-aware resource management and client selection (ELASTIC) algorithm to efficiently solve the problem. Extensive simulations are conducted to demonstrate the performance of ELASTIC.

Federated learning (FL) is a distributed machine learning technique to address the data privacy issue. Participant selection is critical to determine the latency of the training process in a heterogeneous FL architecture, where users with different hardware setups and wireless channel conditions communicate with their base station to participate in the FL training process. Many solutions have been designed to consider computational and uploading latency of different users to select suitable participants such that the straggler problem can be avoided. However, none of these solutions consider the waiting time of a participant, which refers to the latency of a participant waiting for the wireless channel to be available, and the waiting time could significantly affect the latency of the training process, especially when a huge number of participants are involved in the training process and share the wireless channel in the time-division duplexing manner to upload their local FL models. In this paper, we consider not only the computational and uploading latency but also the waiting time (which is estimated based on an M/G/1 queueing model) of a participant to select suitable participants. We formulate an optimization problem to maximize the number of selected participants, who can upload their local models before the deadline in a global iteration. The Latency awarE pARticipant selectioN (LEARN) algorithm is proposed to solve the problem and the performance of LEARN is validated via simulations.