IoT has revolutionized various industries with interconnected smart embedded devices featuring sensors and actuators. However, providing a sustainable power supply for these devices is a significant hurdle. The finite lifespan of traditional batteries necessitates frequent replacements, leading to increased maintenance costs, downtime, and environmental concerns associated with battery disposal. Energy harvesting offers an admirable alternative by converting ambient energy into usable electrical energy, enabling IoT devices to operate autonomously and sustainably, reducing maintenance needs, and enhancing reliability. Despite their diversity, energy harvesters face common limitations such as variable energy availability, limited power output, and high initial costs. Recognizing these limitations is essential for informed decisions regarding their use in sustainable IoT applications and their communication performance. Due to the lack of a comprehensive and inclusive study in this context, this paper gives an extensive survey on energy-harvesting technologies and their applications in IoT. By drawing insights from over 400 recent publications, this paper first introduces a taxonomy by categorizing the existing technologies based on their energy sources into five general domains, i.e., photovoltaic, vibration and kinetic, radio frequency, thermoelectric, and chemical and biological. Then it expands this categorization into 11 subdomains and delves into the working principles of every technology. Besides surveying potential IoT applications, this comprehensive study covers simulation-facilitating mathematical models, pros, cons, and Commercial-Off-The-Shelf (COTS) modules for each harvesting technology. It also includes evaluative simulations to analyze their output power behavior, focusing on their impact on wireless communication performance. Finally, by discussing the existing research trends, the survey aims to provide informed decision-making and foster the development of innovative, sustainable IoT solutions for a green future.

Surgical Internet of Medical Things (IoMT) applications impose stringent Quality of Service (QoS) demands, requiring ultra-low latency and high reliability for critical data streams like control signals or high-definition video. Conventional resource orchestration methods often fall short, failing to guarantee performance for high-priority surgical tasks, particularly under dynamic network conditions, including varied wireless channel quality, and potential congestion. This paper introduces the Priority-Aware SDN Orchestration (PASO) framework, leveraging Software-Defined Networking (SDN) for centralized network visibility and control to address these critical challenges. PASO implements a novel joint optimization strategy focused primarily on maximizing the completion rate of critical surgical tasks that adhere strictly to their maximum delay constraints. A key distinguishing feature is its dynamic resource reallocation mechanism, which employs the preemption of lower-priority, non-medical traffic to proactively secure network resources and guarantee QoS for vital surgical flows, even during periods of high network load. A computationally efficient, priority-aware greedy algorithm operationalizes this strategy on the SDN controller. Comparative simulations against Random, Particle Swarm Optimization (PSO), Simulated Annealing (SA), and a Deep Reinforcement Learning (DRL) agent as baselines demonstrate PASO’s substantial effectiveness. It significantly reduces end-to-end delay (e.g., up to 70% less than PSO under high load) and drastically improves the completion rate for critical surgical tasks (over 80% higher than PSO under high load), while also promoting balanced network utilization. PASO offers a robust solution for dependable, low-latency surgical IoMT operations through its targeted optimization and preemptive resource management capabilities.

As latency-sensitive IoT applications proliferate, fog computing must meet stringent real-time requirements by performing computation near the edge. However, existing task assignment and offloading approaches are predominantly reactive, relying on current system states without anticipating future load fluctuations—leading to frequent congestion, suboptimal resource utilization, and missed task deadlines under bursty workloads. Although a few studies have explored workload prediction, most employ simplistic statistical models or heavy recurrent networks (e.g., LSTM) that are unsuitable for real-time deployment on resource-constrained fog nodes. To address these limitations, this paper introduces Convolutional Assignment Strategy for Tasks (CAST), a Convolutional Neural Network (CNN)-based, predictive task assignment framework designed for resource-constrained fog networks, enabling proactive, deadline-aware, and load-balanced decision-making. CAST operates within a three-layer architecture consisting of IoT devices, fog nodes, and a centralized controller. At the IoT layer, tasks are executed locally or offloaded to fog nodes based on task size, deadline, and device constraints. Fog nodes queue tasks using Earliest Deadline First (EDF) scheduling while periodically reporting workload to the controller. The centralized controller predicts future workload using a CNN trained on time-series workload matrices representing all fog nodes. Unlike Recurrent Neural Networks (RNNs) or Transformers, which incur higher computational latency due to sequential processing or quadratic self-attention complexity, the CNN leverages the spatio-temporal grid of workload matrices to efficiently extract hierarchical features capturing both temporal dynamics and spatial correlations between nodes. Its computational efficiency and parallelizable nature make it suitable for real-time execution on resource-constrained networks. Predicted workloads guide the controller in classifying nodes as crowded or uncrowded, enabling deadline-aware task assignment and dynamic load balancing. Tasks are then offloaded among fog nodes or, if necessary, to the cloud to prevent congestion, minimize queuing delays, and maximize the number of tasks completed within their deadlines. CAST is a robust mechanism, because the CNN model can be periodically retrained to adapt to evolving workload patterns. Simulation results show that CAST reduces average task response time by 30%, improves the deadline-hit ratio by 48%, and achieves more balanced workload distribution compared to state-of-the-art heuristic and modern Deep Reinforcement Learning (DRL) baselines, while maintaining low prediction and decision overhead. Furthermore, Pareto efficiency analysis confirms that CAST represents the optimal solution for latency-critical applications, occupying the high-performance frontier compared to energy-conservative baselines.

Non-Volatile Memory Express (NVMe) is revolutionizing cloud storage by offering high throughput, low latency, and efficient resource utilization. Understanding its performance characteristics under diverse workload conditions is crucial for optimizing storage configurations in cloud environments. Accordingly, this study leverages NVMeVirt, a powerful software-defined NVMe emulator, to investigate the impact of various workload characteristics on critical performance metrics. We analyze the influence of I/O distribution patterns (sequential, random, and Zipfian), queue depths, and the number of concurrent jobs on throughput, latency, and IOPS. We also analyze the energy consumption of both NVMe and SATA SSDs. Results demonstrate that NVMeVirt effectively emulates NVMe SSDs, exhibiting high throughput and low latency across diverse workloads. Sequential workloads consistently achieve higher throughput and significantly lower latency compared to random and Zipfian workloads, particularly at higher queue depths. Specifically,NVMeachieves up to 8.72% higher throughput and 34.89% lower latency than SATAfor sequential workloads. This highlights the importance of aligning storage configurations with applications’ anticipated data access patterns. Our analysis also reveals the potential for optimizing energy efficiency by carefully adjusting workload parameters and storage configurations. Random and Zipfian workloads, due to increased seek operations and internal SSD activity, demonstrate higher energy consumption, underscoring the need for efficient data management strategies in cloud environments. This study provides valuable insights into the performance characteristics of NVMe SSDs and their advantages over SATA SSDs under various workload conditions. It highlights the importance of considering data access patterns and storage configurations for optimizing performance and energy efficiency in cloud-like environments. The findings of this paper can provide researchers with an optimized design and deployment of NVMe-based storage systems in cloud data centers to meet the demands of modern data-intensive applications effectively.