Aug 2026· Electronics· Vol 15, pp. 3689· 1 citation· 41 references
TL;DR
A systematic literature review was conducted following PRISMA guidelines, analyzing peer-reviewed studies published between 2021 and 2026 across major databases, and reveals that hardware–software co-design and custom neural accelerators are crucial for overcoming operational bottlenecks.
Abstract
The exponential growth of the Internet of Things (IoT) has generated massive data streams traditionally processed by centralized cloud architectures, which increasingly face latency, bandwidth, and privacy limitations. Shifting artificial intelligence to resource-constrained edge nodes, known as TinyML, offers a robust decentralized alternative, though it introduces severe memory, compute, and energy bottlenecks. To map this transition, a systematic literature review was conducted following PRISMA guidelines, analyzing peer-reviewed studies published between 2021 and 2026 across major databases. The analysis identifies primary architectural paradigms and evaluates the efficacy of state-of-the-art model compression techniques, such as quantization, pruning, and knowledge distillation. Furthermore, the findings reveal that hardware–software co-design and custom neural accelerators are crucial for overcoming operational bottlenecks, while also highlighting persistent security and privacy challenges in on-device learning. Ultimately, while deploying complex models on microcontrollers is increasingly viable, achieving optimal performance demands holistic optimization strategies. This review synthesizes current research gaps and provides a strategic roadmap to guide future interdisciplinary efforts toward resilient, energy-efficient, and secure next-generation intelligent edge systems.
A comprehensive review of edge computing as a modern trend in information technology, including the convergence of edge computing with artificial intelligence (Edge AI), 6G networks, digital twins, and serverless edge architectures is presented.
Awiti Gideon Appiah, Dr. Lazarus Kwao, Benjamin Opoku Atuahene· International Journal of Cre...· 0 citations
Internet of Things and Edge architectures have become increasingly popular during the last few years, leading to more complex architectures that cover multiple real-life use cases in various domains. These complex architectures often employ edge and IoT storage platforms that optimize the storage, processing, transfer, and general governance of the data produced and used by the applications hosted in these architectures. The present work proposes a novel approach for intelligent self-optimization in an edge cloud employing IoT storage nodes. Intelligent refers to the usage of constraint-aware optimization algorithms that adapt placement decisions to the evolving state of the network rather than relying on manual configuration. It aims to proactively place docker and virtual machine images in specific nodes to minimize the transfer delays, the bandwidth used, and the occupied memory in the edge nodes. The solution presented builds on our earlier work by including the state of the edge network at each point in time, creating a time series of graphs, and applying four distinct optimization methods to optimize the image placement at each timestep. The evaluation was performed in simulated scenarios, testing IoT edge networks of 64 nodes with mixed ethernet and Wi-Fi connections. The results indicate that integer linear programming solutions consistently achieve the minimum number of image replicas across all evaluated topologies, reducing the average hosts to clients ratio to 0.36–0.49 compared to 0.87–2.03 for the next best algorithm, while greedy and approximation algorithms deliver quick and cost effective placements in under 15 ms per timestep across all network configurations.
The needs of scenarios such as the rapidly expanding Internet of Things (IoT) ecosystem or real-time services are
examples of situations where a centralized cloud model may not suffice‚ as all data resides on the cloud and traditional cloud
computing practices have limitations․ Low latency‚ high bandwidth and privacy issues arise due to the ever-growing amount of
data being generated by connected devices․ One way to address these issues is to move computation‚ storage and networking
resources closer to the edge where data is generated․ Edge computing can minimize latency in latency-sensitive applications and
can also help reduce network congestion by offloading processing from the core network․ This paper surveys the recent works in
edge computing‚ classifying the existing body of work in edge computing into four categories based on four dimensions: physical
architecture‚ algorithmic optimization techniques‚ performance objectives‚ and security mechanisms․ The survey provides
examples of representative works for each dimension‚ analyses the corresponding performance-energy trade-offs based on
processing performance‚ energy consumption‚ and security objectives‚ and presents future research challenges such as
hardware heterogeneity‚ network dynamics‚ and the development of energy-efficient edge systems․ The paper also outlines
research directions to enable the development and deployment of cloud-to-edge and cloud-to-thing computing environments in
the future․
K. M. Meena Kumari, Rajesh Kumar, Naveen Monga· International Journal for Re...· 0 citations
As the deployment of Internet of Things (IoT) ecosystems accelerates, safeguarding distributed networks against pervasive security and privacy threats has become a paramount concern. Integrating robust cryptographic protocols directly onto resource-limited edge devices offers a promising line of defense. However, severe hardware constraints historically complicate practical implementation. Because finite-field arithmetic fundamentally dictates the speed and efficiency of these cryptographic primitives, optimizing underlying multiplication techniques remains critical. To address these challenges, this paper presents an innovative, highly regular bit-serial systolic architecture tailored specifically for Dickson modular multiplication in binary extension fields. This is achieved via a streamlined systolic mapping over GF(2l) using dependency graph extraction, scheduling vectors, and projection directions. With localized pathways, the structure is highly optimized for VLSI integration. The performance and effectiveness of the proposed system are thoroughly evaluated and validated through comprehensive simulation results. Based on analytical and gate-level modeling, the design significantly enhances efficiency, lowering area by at least 162.8%, power by at least 214.3%, Area–Time Product by at least 5%, and Time–Power Product by at least 25.6%. These findings confirm that the proposed architecture substantially outperforms state-of-the-art bit-serial multipliers across these key evaluation metrics. Consequently, this solution serves as an ideal cryptographic engine for tightly constrained IoT hardware and embedded nodes, reinforcing secure and energy-aware data processing. By fostering resilient infrastructure and green digital practices, the work directly supports sustainable digital transformation and robust edge computing security.
The integration of Edge Artificial Intelligence (AI), Federated Learning (FL), and the Internet of Things (IoT) is fundamental to the development of next-generation intelligent computing paradigms. While these technologies are advancing rapidly, their convergence into a unified Edge-FL-IoT framework presents significant complexity due to fragmented architectures and diverse operational requirements of the system. The primary objective of this paper is to provide a comprehensive analysis of the convergence of Edge AI, Federated Learning (FL), and the Internet of Things (IoT). To achieve this, we present a multi-dimensional framework for classifying and evaluating existing Edge-FL-IoT systems according to architectural design, communication protocols, and privacy protection strategies. The analysis demonstrates that the integration of Edge AI and FL addresses key limitations of traditional centralized AI, particularly by alleviating bandwidth constraints and mitigating data ownership vulnerability. In addition, the study identifies and categorizes unresolved technical challenges within Edge-FL-IoT environments, including resource heterogeneity among edge nodes, non-independent and identically distributed (non-IID) data and emerging security threats. The paper concludes by outlining promising directions for future research and discussing major technological trends, thereby providing a comprehensive roadmap for the design of secure, efficient, and scalable decentralized intelligent system.
Harjot Kaur Gill, Jagdeep Singh, A. Girdhar· International Conference on...· 0 citations
The rapid growth of artificial intelligence (AI), machine learning, and large-scale digital services is placing unprecedented demand on cloud infrastructure, making scalable and sustainable compute increasingly important. While advances in processors and accelerators continue to improve computational capability, architectural coordination can unlock efficiency gains that hardware-generation improvements alone may not fully capture, particularly across increasingly heterogeneous compute environments (Armbrust et al., 2010). This paper will examine virtualization, workload scheduling, the provisioning of heterogeneous resources and infrastructure lifecycle management's effects on infrastructure efficiency at hyperscale. The paper extends the NIST cloud service definition (Mell & Grance, 2011) and most recent research on energy efficient resource management (Khan et al., 2022, Ilager et al., 2021) to propose a Sustainable Compute Lifecycle Framework comprised of five inter-connected phases: Platform Planning, Platform Deployment and Optimization, Platform Operations, Hardware Modernization, and Hardware Retirement and Resource Reclamation. The paper also explores new trends such as Compute Express Link (CXL) for memory pooling and disaggregation (Das Sharma et al., 2024;Chen et al., 2024) and the use of AI for infrastructure scheduling (Sanjalawe et al.,2025). The key message is that there is a new central design coordination of the cloud architecture that enables sustainable growth of the cloud, not merely increasing incremental capacity on hardware.
Priyadarshni Shanmugavadivelu· International journal of com...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
