Collaborative IoT Cloud and Edge Computing Solutions for Smart Cities

Abstract

Smart-city Internet of Things (IoT) applications combine large device populations, heterogeneous workloads, intermittent wireless links, and demanding latency targets. Central cloud computing offers elastic capacity for computation-intensive analytics, whereas multi-access edge computing places compute and storage near data sources to reduce response time and backhaul traffic. Neither layer is universally optimal. This paper develops an expanded collaborative edge-cloud framework in which each task can be executed locally, at an edge server, or in a central cloud. A mathematical model captures communication delay, execution delay, device energy, monetary cost, server capacity, deadline compliance, and privacy restrictions. The resulting mixed-integer optimization problem is coupled with a distributed Deep-learning-Driven Task-Offloading (DDTO) policy that approximates high-quality decisions using locally observed system states and coordinated resource allocation. The manuscript also distinguishes proposed methodology from reported evidence: the frequently cited performance gain of up to 63% belongs to the experimental study by Wu et al. and is used here as prior evidence rather than as a new result. Security, interoperability, fairness, resilience, and governance are integrated into the design discussion. The framework provides a reproducible basis for future simulation or deployment studies in traffic control, video analytics, environmental sensing, emergency response, and smart energy management.