This paper investigates the outage probability in edge intelligence (EI)-enabled networks that support embodied agents, such as humanoid robotic assistants (HRAs) powered by foundation models (FMs), within indoor hotspot (InH) environments where base stations are typically mounted on walls or ceilings in corridors. The analysis considers both line-of-sight (LOS) and non-line-of-sight (NLOS) conditions in internet of things (IoT) environments. We propose a novel cache-aware inference offloading (CAIO) strategy that combines edge caching and cloud-based retrieval-augmented generation (RAG) to deliver timely and accurate FM inference. Closed-form expressions for outage probability are derived under Rayleigh fading for both CAIO and a cloud-only (CLD) benchmark. The system model incorporates realistic assumptions, including binomial point processes for user distribution and practical backhaul latency modeled using Gamma distributions. Unlike prior work that primarily addresses latency or energy efficiency, the present analysis explicitly considers FM inference accuracy and processing dynamics at both edge and cloud. Analytical and simulation results identify optimal BS inference accuracy, FM content popularity, and cache size under varying maximum RAG latency values to ensure reliable IoT operation, showing that CAIO consistently outperforms CLD across all scenarios for real-time FM inference for HRAs.

In this paper, the inference freshness of generative artificial intelligence (gen-AI) services in industrial Internet-of-Things (IIoT) networks is investigated. A freshness metric termed the peak age of inference (PAoIF) is proposed to quantify inference freshness by accounting for peak age of information and delays due to transmitting inference requests and results. A cache-aware offloading (CAO) strategy which employs multi-access edge computing in IIoT networks is also proposed for timely inference delivery. Leveraging novel closed-form expressions for PAoIF violation probability within the proposed CAO and benchmark strategies for IIoT, this study analyzes the impact of PAoIF violation age, gen-AI service request rate, and average transmission rate, on inference freshness. We identified scenarios where the proposed CAO strategy exhibits a lower PAoIF violation probability compared to the benchmark strategies under consideration. Furthermore, we show that PAoIF violation probability under the proposed CAO strategy is minimized via optimizing the average transmission rate in IIoT networks employing servers with limited computing resources. Therefore, the analysis shows that the proposed CAO strategy is a viable technique towards enabling inference freshness for gen-AI services in IIoT networks.

This paper investigates the maximization of average energy efficiency (AEE) in MEC-enabled Industrial Internet of Things (IIoT) networks through a cache-based association strategy (CBAS). The system comprises autonomous mobile robots (AMRs), small-cell base stations (SBSs), and macro-cell base stations (MBSs) operating under line-of-sight obstructions, fading, and interference. To jointly optimize transmit power, SBS cache size, MBS cache size, and SBS intensity, parameters critical to CBAS efficiency, we propose a deep Q-network (DQN)-based approach that circumvents the computational burden of solving NP-hard formulations through exhaustive search. In addition to performance benchmarking, we enhance explainability by analyzing the learned state embeddings using t-SNE. Simulation results show that our approach achieves near-optimal AEE with significantly lower overhead, making it scalable and suitable for real-time IIoT deployments.

Minimizing the Age of Information (AoI) violation probability is essential for reliable and timely data delivery in MEC-enabled IIoT networks. This study proposes the Joint Caching and Power Control (JCPC) scheme, leveraging Deep Deterministic Policy Gradient (DDPG) to optimize caching decisions and power allocation under system-level AoI and power constraints. Simulations validate that JCPC outperforms baseline strategies, significantly reducing AoI violation probability and improving the efficiency of MEC-enabled IIoT networks.

This article investigates the Peak Age of Information (PAoI) violation probability of status update strategies in cache-enabled Industrial Internet of Things (IIoT) networks. Specifically, a PAoI characterization framework for PAoI violation probability analysis in cache-enabled IIoT networks is presented for status updates with and without a zero wait (ZW) policy. Furthermore, multicast-aware (MA) and hierarchical-aware (HA) status update strategies are also proposed for cache-enabled IIoT networks. Using newly derived PAoI cumulative distribution functions (CDFs) and a status update transmission delay probability density function (PDF), closed-form PAoI violation probability expressions are obtained for the MA and HA status update strategies. Extensive analysis shows that the ZW policy for status updates in cache-enabled IIoT networks can experience high PAoI violation probability when high transmission delay is encountered during the delivery of acknowledgement messages. Furthermore, the PAoI violation probability of the proposed MA and HA status update strategies are dependent upon the status update transmission delays, PAoI violation threshold, wireless link timeout threshold, and the popularity of wireless sensors (WSs). We show that the proposed MA and HA status update strategies exhibit lower PAoI violation probability over other benchmark strategies. Importantly, the proposed MA status update strategy with and without the ZW policy attains the lowest PAoI violation probability even in scenarios where status updates are requested from a large pool of WSs. Therefore, the analysis shows both the proposed MA and HA status update strategies as promising solutions toward the monitoring and control of mission-critical IIoT applications.

An important role played by industrial Internet of Things (IIoTs) networks is supporting the operations of autonomous mobile robots (AMRs) by leveraging multiple-access edge caching servers. At the same time, judicious content caching strategies are essential for minimizing content retrieval delays and the costs associated with updating caches. In this study, a novel multiagent reinforcement learning (MARL)-based cache update strategy termed multiagent cache update (MACU) is proposed. MACU leverages the multiagent deep deterministic policy gradient (MADDPG) framework and aims to optimize cache updates to reduce Age of Information (AoI), minimize events where AoI exceeds acceptable levels, and ensure that the costs associated with performing cache updates are kept low. Furthermore, to mitigate the computational complexity of training agents with MACU, the MACU with global critic (MACU-GC) variant is introduced, which diverges from traditional MADDPG by employing a singular global critic for enhanced training efficiency. Extensive numerical evaluations showcase the proposed strategies’ superiority over conventional deep reinforcement learning-based caching methods, achieving significant improvements in AoI costs, caching costs, and AoI violation costs, while effectively reducing content retrieval latency, enhancing hit rates, and optimizing AoI and link load metrics.

In this paper, the problem of energy-efficient uncrewed aerial vehicle (UAV)-assisted multi-access task offloading is investigated. In the studied system, several UAVs are deployed as edge servers to cooperatively aid task executions for several energy-limited computation-scarce terrestrial user equipments (UEs). An expected energy efficiency maximization problem is then formulated to jointly optimize UAV trajectories, UE local central processing unit (CPU) clock speeds, UAV-UE associations, time slot slicing, and UE offloading powers. This optimization is subject to practical constraints, including UAV mobility, local computing capabilities, mixed-integer UAV-UE pairing indicators, time slot division, UE transmit power, UAV computational capacities, and information causality. To tackle the multi-dimensional optimization problem under consideration, the duo-staggered perturbed actor-critic with modular networks (DSPAC-MN) solution in a multi-agent deep reinforcement learning (MADRL) setup, is proposed and tailored, after mapping the original problem into a stochastic (Markov) game. Time complexity and communication overhead are analyzed, while convergence performance is discussed. Compared to representative benchmarks, e.g., multi-agent deep deterministic policy gradient (MADDPG) and multi-agent twin-delayed DDPG (MATD3), the proposed DSPAC-MN is validated to be able to achieve the optimal performance of average energy efficiency, while ensuring 100% safe flights.

In this paper, the problem of energy-efficient unmanned aerial vehicle (UAV)-assisted computation offloading over the Terahertz (THz) spectrum is investigated. In the studied system, several UAVs are deployed as edge servers to aid task executions for multiple energy-limited computation-scarce terrestrial user equipments (UEs). Then, an expected energy efficiency maximization problem is formulated, aiming to jointly optimize UAVs’ trajectories, UEs’ local central processing unit (CPU) clock speeds, UAV-UE associations, time slot slicing, and UEs’ offloading powers. To tackle the considered multi-dimensional optimization problem, the duo-staggered perturbed actor-critic with modular networks (DSPAC-MN) solution in a multi-agent deep reinforcement learning (MADRL) setup, is proposed and tailored, after mapping the original problem into a stochastic (Markov) game. Compared to representative benchmarks in simulations, e.g., multi-agent deep deterministic policy gradient (MADDPG) and multi-agent twin-delayed DDPG (MATD3), the proposed DSPAC-MN can achieve the optimal performance of average energy efficiency, while ensuring 100% safe flights.

In the context of the Industrial Internet of Things (IIoT), multiple access edge computing (MEC) enables the provision of computational resources closer to users. The work presented in this paper explores a common IIoT application scenario wherein autonomous mobile robots (AMR) operate in a MEC-enabled network and depend on multimedia content stored at the MEC servers over the course of operation. The age of information (AoI) metric is used to measure the freshness of the cached content from the perspective of the AMR. At the same time, the energy cost associated with refreshing the cache files is simultaneously considered as well. This paper delves into achieving an optimal trade-off between minimizing the weighted AoI cost and the energy expended for cache refreshing. We address this problem by introducing a cache refreshing-deep deterministic policy gradient (CR-DDPG) algorithm, a model-free deep reinforcement learning method, to optimize both AoI and energy usage. Various simulation studies are conducted to evaluate the proposed CR-DDPG algorithm, and the results demonstrate that CR-DDPG consistently outperforms its baseline counterparts, rendering it a robust approach for cache-refreshing in dynamic IIoT environments.

The mean peak age of information (PAoI) of computation offloading in multi-access edge computing (MEC) enabled industrial Internet-of-Things (IIoT) networks is investigated in this paper. Specifically, the mean PAoI and average energy consumption of energy-aware computation offloading (ECO) and cloud-only (CL) computation offloading are derived in closed-form. Numerical and simulation results show that the mean PAoI and average energy consumption of the ECO and CL strategies are dependent on the task generation rate and local computation capability experienced in the MEC-enabled IIoT networks. Specifically, the ECO strategy is shown to achieve lower mean PAoI even at high task generation rates in the MEC-enabled IIoT network. Furthermore, we show that the ECO strategy achieves lower average energy consumption than the CL strategy even when local computing capability in the MEC-enabled IIoT network is limited.

This paper investigates the energy efficiency of computation offloading strategies in multi-access edge computing-enabled (MEC-enabled) Internet-of-Vehicles (IoV) networks. First, the energy efficiency of computation offloading strategies in the MEC-enabled IoV network are derived in closed-form. Thereafter, a multi-agent deep reinforcement learning based (MADRL-based) energy efficiency maximization algorithm is proposed to enable computation offloading strategies to attain maximum energy efficiency in the MEC-enabled IoV network. It is shown through extensive analysis that the maximum attained energy efficiency hinges on the choice of task size and transmission timeout threshold, with a computation offloading strategy that jointly considers transmission and computation latencies outperforming existing strategies. It is also shown that the proposed MADRL-based energy efficiency maximization algorithm achieves near-optimal energy efficiency in the MEC-enabled IoV network, making it a promising solution towards achieving energy efficient MEC-enabled IoV networks.

The average energy efficiency of computation offloading in multi-access edge computing (MEC) enabled Internet-of-Vehicles (IoV) networks with wired backhaul links is investigated in this paper. Specifically, the cumulative distribution function (CDF) for computation delay over wired backhaul link is derived in closed-form. The average energy efficiency of computation offloading in MEC-enabled IoV networks with wired backhaul links is derived in closed-form, and is compared against non MEC-enabled IoV networks with wired backhaul links. Numerical and simulation results show that MEC-enabled IoV networks can attain at least 10 times higher energy efficiency than non MEC-enabled IoV networks. Furthermore, we show that energy efficiency can be maximized through the proper selection of task size and transmission deadline in the MEC-enabled IoV network.

In this work, we propose SVDNet, a novel deep learning (DL) architecture that utilizes singular value decomposition (SVD), for transmit power control in a multiuser multiple input multiple output (MU-MIMO) system. We propose a novel method of training SVDNet in a supervised manner for the power control task by using binary cross-entropy loss functions. SVDNet requires fewer computations than traditional power control algorithms such as weighted minimum mean squared error (WMMSE). Our simulation results show that the proposed SVDNet provides over a 50% increase in sum-rate performance as compared to similar supervised DL-based power control schemes while being significantly more computationally efficient than WMMSE.

To serve advanced use-cases in industrial internet of things (IIoT) setups, communication and computation over wireless networks have faced overlapping resource management challenges. Two crucial resources in this context are radio re-sources and computational resources. The problem to achieve the ultra-low latency for mission critical applications is motivating enterprises to invest in offloading capability of computation heavy tasks while retaining the bandwidth efficiency of edge nodes. This work proposes a novel multi-hop offloading framework powered by deep reinforcement learning to aid the edge nodes in making intelligent decisions on task offloading. The proposed method is benchmarked against existing state of the art techniques to measure task completion delay and algorithmic runtime.

A performance analysis of base station (BS) association strategies is conducted in this paper for an industrial fifth generation (5G) network comprising a mobile actuator operating in the presence of interference. Specifically, outage probability expressions are derived with a stochastic geometry framework for the random association (RA), strongest association (SA), closest association (CA), and closest with strongest association (CSA) strategies. An analysis reveals that the SA strategy exhibits the lowest outage probability at the expense of significantly higher overheads. In contrast, the CSA strategy has significantly less overheads and exhibited outage probability that is higher than the SA strategy but lower than the CA and RA strategies. Thus, the SA strategy is suitable for industrial use cases with very low outage probability requirements and high overhead tolerance while the CSA strategy can be adopted for industrial use cases with low overhead and higher outage probability requirements. It is also noted that the CSA strategy can be further improved by introducing more BS tiers.

In this paper, non-orthogonal multiple access (NOMA) is investigated as a potential solution to address spectrum scarcity in unmanned aerial vehicle (UAV) communications. In particular, the ergodic capacity of uplink (UL) NOMA is characterized within a stochastic geometry framework for a multi-UAV network comprising a ground station (GS) and multiple UAVs. Our analysis reveals that UL NOMA enables fairness, in terms of ergodic capacity, to be achieved in the multi-UAV network. When compared against conventional UL orthogonal multiple access (OMA), we show that the ergodic capacity and ergodic sum capacity is higher for the proposed UL NOMA transmissions. Thus, the proposed scheme enables higher throughput to be achieved in multi-UAV networks while simultaneously improving spectrum efficiency.

The impact of cellular interference on uplink unmanned aerial vehicle (UAV) communications is analyzed in this paper for a multi-UAV network. Specifically, the outage probability and finite signal-to-noise ratio (SNR) diversity gain of the multi-UAV network is characterized in the presence of interference from uplink UAVs and cellular base stations. We demonstrate that the operational altitude of the multi-UAV network determines the severity of cellular interference across low-to-high transmit power regimes. In particular, UAVs operating at higher altitudes are less affected by cellular interference, with higher finite SNR diversity gains and lower outage probability floors observed. Thus, interference mitigation techniques may have to be considered when multi-UAV networks are operating at low altitudes in areas with cellular networks.

In this paper, we demonstrate the necessity of robust cellular interference cancellation when co-locating uplink (UL) unmanned aerial vehicle (UAV) communications with cellular networks. Through a stochastic geometry based evaluation framework, we show through bit error rate (BER) and ergodic capacity analysis that multi-user interference (MUI) is the main limiting factor in UL UAV communications after cellular interference cancellation. It is also observed that high altitudes and weak residual cellular interference corresponds to lower BER. In contrast, low altitudes and weak residual cellular interference results in higher ergodic capacity. Hence, advanced interference cancellation techniques should be considered to suppress the presence of strong MUI in the multi-UAV network.

Unmanned aerial vehicles (UAVs) are set to feature heavily in upcoming fifth generation (5G) networks. Yet, the adoption of multi-UAV networks means that spectrum scarcity in UAV communications is an issue in need of urgent solutions. Towards this end, downlink non-orthogonal multiple access (NOMA) is investigated in this paper for multi-UAV networks to improve spectrum utilization. Using the bivariate Rician shadowed fading model, closed-form expressions for the joint probability density function (PDF), marginal cumulative distribution functions (CDFs), and outage probability expressions are derived. Under a stochastic geometry framework for downlink NOMA at the UAVs, an outage probability analysis of the multi-UAV network is conducted, where it is shown that downlink NOMA attains lower outage probability than orthogonal multiple access (OMA). Furthermore, it is shown that NOMA is less susceptible to shadowing than OMA.

As unmanned aerial vehicles (UAVs) are expected to play a significant role in fifth generation (5G) networks, addressing spectrum scarcity in UAV communications remains a pressing issue. In this regard, the feasibility of full-duplex non-orthogonal multiple access (FD-NOMA) UAV communications to improve spectrum utilization is investigated in this paper. Specifically, closed-form outage probability expressions are presented for FD-NOMA, half-duplex non-orthogonal multiple access (HD-NOMA), and half-duplex orthogonal multiple access (HD-OMA) schemes over Rician shadowed fading channels. Extensive analysis revealed that the bottleneck of performance in FD-NOMA is at the downlink UAVs. Also, FD-NOMA exhibits lower outage probability at the ground station (GS) and downlink UAVs than HD-NOMA and HD-OMA under low transmit power regimes. At high transmit power regimes, FD-NOMA is limited by residual SI and inter-UAV interference at the downlink UAVs and FD-GS, respectively. The impact of shadowing is also shown to affect the reliability of FD-NOMA and HD-OMA at the downlink UAVs.

A hybrid-duplex aeronautical communication system (HBD-ACS) consisting of a full-duplex-enabled ground station (GS) and two half-duplex (HD) air stations (ASs) is proposed as a direct solution to the spectrum crunch faced by the aviation industry. The closed-form outage probability and finite signal-to-noise ratio (SNR) diversity gain expressions in aeronautical communications over Rician fading channels are derived for a successive interference cancellation (SIC) detector. Similar expressions are also presented for an interference ignorant (II) detector and the HD-equivalent modes at GS and ASs. Through the outage and finite SNR diversity gain analysis conducted at the nodes, and system level, residual self-interference (SI) and inter-AS interference are found to be the primary limiting factors in the proposed HBD-ACS. Further investigations revealed that the II and SIC detectors in the proposed HBD-ACS are suitable for the weak and strong interference scenarios, respectively. When compared with the HD-ACS, the proposed HBD-ACS achieves a lower outage probability and higher diversity gains at higher multiplexing gains when operating at low SNRs. The finite SNR analysis also showed the possibility of the proposed HBD-ACS being able to attain interference-free diversity gains through proper management of the residual SI. Hence, the proposed HBD-ACS is more reliable and can provide a better throughput compared with the existing HD-ACS at low-to-moderate SNRs.

With the growing popularity of unmanned aerial vehicles (UAVs), spectrum management is a pressing issue, particularly for multi-UAV systems. To this end, a hybrid-duplex (HBD) UAV communication system (UCS) consisting of a full-duplex (FD) enabled ground station (GS), and legacy half-duplex (HD) UAVs is proposed in this paper. To model the fading and shadowing environment commonly encountered in UAV communications, a mix of Rician and Rician shadowed fading is assumed. In particular, novel power series approximations of the Rician shadowed fading power probability density function (PDF), and cumulative distribution function (CDF) are presented, along with closed-form outage probability expressions. Performance analysis shows that the proposed HBD-UCS exhibits lower outage probability than the HD-UCS when shadowing is encountered at low signal-to-noise ratios (SNRs). Also, inter-UAV interference has a stronger influence on outage probability decay at low SNR regimes, with lower inter-UAV interference corresponding to a sharper decline in outage probability.

Air traffic growth is expected to rise sharply in the near future, causing a swell in data communication demands on the congested aeronautical spectrum which cannot be met with existing aeronautical communication systems. To this end, a hybrid-duplex (HBD) aeronautical communications system consisting of a full-duplex (FD) enabled ground station, and legacy half-duplex (HD) air- stations is proposed as a direct solution to boost spectral efficiency. In particular, the system level outage probabilities and transmission power requirements of the proposed HBD and HD aeronautical communication systems are analyzed. Performance analysis shows that the proposed HBD system can attain better outage performance in the en route scenario. The proposed HBD system is also able to support an equivalent data rate as an HD system with lower transmit power and signal-to- interference-plus-noise-ratio (SINR) requirements and longer transmission range when inter-aircraft interference and residual self-interference is kept sufficiently low.

The demand for high data rate communication is continuously increasing. To address this, studies on bit error rate improvements, cooperative diversity and transmit diversity have been topics of interest. In this paper, a novel Quad State- Paired QPSK (QS-PQPSK) modulation scheme is proposed. The proposed technique is simulated and compared with other main PSK-based modulation schemes. Preliminary simulation results show that in the presence of only Additive White Gaussian Noise (AWGN), QS-PQPSK performs better than 8-PSK in terms of symbol error rate and effective throughput. In a Rayleigh flat fading channel with AWGN, QS-PQPSK outperforms 8-PSK in terms of bit and symbol error rates and effective throughput. The performance of the proposed method also surpasses the other main PSK-based methods in most cases. This highlights the proposed QS-PQPSK technique as a viable spectrally efficient modulation technique for high data rate applications.

Electroencephalography (EEG) signals have widely been used for developing Brain Computer Interface (BCI) systems. BCI systems generally record, process and extract informative features hidden in brain signals, ultimately aiming towards "human thought translation". A number of EEG based BCI studies focus on estimation and enhancement of cognitive functions such as attention, memory and creativity, assessing mental workload, fatigue, etc. Detection of discriminative EEG features associated with presentation of familiar and non-familiar images is not well-studied so far, though it is worthy to explore its usability even in EEG-based authentication systems. In this paper, a set of time-frequency based EEG features are investigated, while the subjects are exposed to familiar and unfamiliar visual stimuli (images) for a fixed time period during the experimental paradigm. The results show that combination of features such as band power values, signal peaks, activity and mobility of the signal gives an average accuracy of 70.71% in classifying between familiar and unfamiliar images among 7 subjects. Further investigation is necessary to improve the classification performance and to reduce the effects of intersubject and intra-subject variability of EEG signals during feature extraction.