Executive Summary

Edge Computing has evolved as a critical extension to cloud computing by bringing computation closer to data sources, thus reducing latency and bandwidth use. Early research established the principles of moving compute tasks to the edge of the network to improve performance in distributed systems. Recent advancements include federated learning for distributed training of AI models at the edge, task scheduling algorithms, energy efficiency improvements, and containerization for deployment flexibility. Current challenges involve managing heterogeneity in devices, ensuring security and privacy, and optimizing resource allocation under uncertain environments. The comprehensive implementation of edge computing also depends on advancements in hardware design such as RISC-V SoCs and AI accelerators, which address power and space constraints of edge devices. The future of edge computing looks toward fully integrating AI with edge computing frameworks to achieve better system autonomy, predictability, and reliability.

Research History

Research in Edge Computing began with works identifying its necessity due to the explosion in the volume of data produced by IoT devices. Foundational papers presented architectures for Edge Computing, delineating its advantages over traditional cloud paradigms. These papers include seminal works like **"The Fog Computing Paradigm: Scenarios and Security Issues"* (Bonomi et al., 2012)* and **"Towards an Understanding of Facets and Exemplars of Big Data Applications"* (Satyanarayanan et al., 2015)*, which have laid the groundwork for further research in the domain.

Recent Advancements

Recent advancements focus on integrating AI and machine learning into edge environments, as evidenced by **"Federated Learning on Stochastic Neural Networks"* (Tang et al., 2025)* and **"SLED: A Speculative LLM Decoding Framework for Efficient Edge Serving"* (Li et al., 2025). These advances aim to deploy intelligent systems in a distributed manner that respects the resource constraints of edge devices while promoting enhanced computational capabilities and efficiencies. Another significant progress is the *"Design and Implementation of a RISC-V SoC with Custom DSP Accelerators for Edge Computing"* (Yadav, 2025)* which enhances hardware designs for edge devices.

Current Challenges

A primary challenge in Edge Computing is the heterogeneity of devices and networks. Papers addressing these include **"Multi-dimensional Autoscaling of Processing Services: A Comparison of Agent-based Methods"* (Sedlak et al., 2025)* and **"HASFL: Heterogeneity-aware Split Federated Learning over Edge Computing Systems"* (Lin et al., 2025), which introduce frameworks to cope with diverse capabilities and facilitate learning in such environments. Security also persists as a challenge, with works like *"Toward a Lightweight, Scalable, and Parallel Secure Encryption Engine"* (Karakchi et al., 2025)* seeking to improve data privacy and integrity.

Conclusions

Edge Computing is progressing towards a more heterogeneous and AI-integrated environment. Though foundational principles remain pertinent, the field is rapidly evolving with techniques to manage device diversity, security, and AI model deployment challenges. There is a notable progression towards smarter edge systems, informed by cutting-edge algorithms and hardware developments. Interdisciplinary collaboration is key, and as edge computing matures, it is expected to significantly impact IoT, autonomous systems, and beyond. Future research should focus on addressing the scalability, security, and interoperability challenges to achieve seamless integration across various edge computing paradigms.