Fog computing is a decentralized computing paradigm that acts as an intermediate layer between local end-devices (such as IoT sensors) and centralized cloud data centres (Gupta & Bharti, 2023). Introduced by Cisco, it extends cloud computing capabilities to the edge of the network, enabling data to be processed, analyzed, and stored closer to the source (Atlam et al., 2018; Al-Shareeda et al., 2024). By performing computations locally rather than sending all raw data to the cloud, fog computing significantly reduces communication latency and network congestion (Dastjerdi et al., 2016). It is not a replacement for cloud computing but rather a complementary architecture that handles time-sensitive, mission-critical tasks locally, while leaving compute-intensive and delay-tolerant tasks to the cloud (Jeyaraman et al., 2024; Atlam et al., 2018).
Key Characteristics
Fog computing is defined by several core architectural features that distinguish it from centralized cloud models:
- Geographical Distribution: Unlike the centralized nature of cloud data centres, fog nodes are distributed across a wide geographical area, closer to the data source (Gupta & Bharti, 2023).
- Low Latency: By minimizing the distance data must travel, fog computing supports real-time applications where every millisecond is critical, such as industrial automation or autonomous systems (Dastjerdi et al., 2016).
- Location Awareness: Fog nodes can infer their own location and track end-user devices, enabling context-aware services (Gupta & Bharti, 2023).
- Heterogeneity: The architecture supports a wide variety of devices, including wireless routers, gateways, switches, and edge servers (Atlam et al., 2018; Dolui & Datta, 2017).
- Bandwidth Efficiency: By filtering, preprocessing, and trimming data at the edge, only essential insights are transmitted to the cloud, significantly reducing unnecessary bandwidth consumption (Atlam et al., 2018).
Fog vs. Edge Computing
While the terms are often used interchangeably, there is a technical distinction:
- Edge Computing: Generally refers to processing data directly on the device or at the very edge of the network (e.g., inside an IoT sensor or camera).
- Fog Computing: Represents a broader, more structured architecture that includes an intermediary layer of fog nodes (gateways, routers) that perform processing before data reaches the cloud (Dolui & Datta, 2017).
Layers of Fog Architecture
The typical layered architecture of fog computing includes:
- Physical/Virtualization Layer: Comprises sensors, actuators, and heterogeneous devices that collect raw data (Atlam et al., 2018).
- Monitoring Layer: Tracks the availability, energy consumption, and performance status of fog nodes (Atlam et al., 2018).
- Preprocessing Layer: Performs data filtering, aggregation, and trimming to extract meaningful information (Atlam et al., 2018).
- Temporary Storage Layer: Holds processed data locally before it is either acted upon or transmitted to the cloud (Atlam et al., 2018).
- Security Layer: Manages encryption, decryption, and integrity measures to protect data (Atlam et al., 2018).
- Transport Layer: Facilitates the secure transmission of preprocessed data to the central cloud for long-term storage or deeper analytics (Atlam et al., 2018).
Applications
- Healthcare: Enables real-time monitoring of vitals during surgery or immediate diagnostics for time-sensitive conditions like stroke detection, where latency could have fatal consequences (Jeyaraman et al., 2024).
- Smart Factories (Industry 4.0): Facilitates predictive maintenance and robotics control on the factory floor by processing machine sensor data in real time (Al-Shareeda et al., 2024).
- Smart Cities: Supports intelligent traffic management, smart grids, and surveillance systems that require immediate local decision-making (Al-Shareeda et al., 2024).