Advances and Challenges in Software Architecture for IoT-based Smart Computing in Environmental Applications: A Review

Abstract

The environment around us is changing faster than its traditional monitoring systems were ever designed to handle. Over the last decade, the Internet of Things (IoT) has quietly moved from a research curiosity to the backbone of how to watch over air, water, soil, forests, and cities. Yet the sensors themselves are only half the story. What actually decides whether an environmental IoT system survives in the real world is its software architecture - the way data moves from a muddy field sensor to a decision-maker's screen. This review takes a close look at how software architectures for IoT-based smart computing have evolved in environmental applications. Explaining the dominant architectural styles, from classic layered designs to service-oriented, microservice, and event-driven approaches, and examines how edge, fog, and cloud computing have reshaped where intelligence actually lives in these systems. Then, discussed about recent advances, including lightweight machine learning at the edge, digital twins of natural systems, and low-power wide-area networking, before turning to the stubborn challenges that keep coming up in the literature: interoperability between vendor ecosystems, energy constraints in remote deployments, security of unattended devices, data quality, and the general lack of architectural standardization. Further, a close with a set of research directions that deserve far more attention than they currently receive. This article aims to give researchers and practitioners a realistic picture of where the field stands - what genuinely works, what is still fragile, and where the next round of effort should go.