Closer to the Cloud or Closer to the Ground? Fog vs Edge Computing

 

Closer to the Cloud or Closer to the Ground?

Fog vs Edge Computing

 

Introduction:

Imagine a self-driving car that takes half a second too long to detect a pedestrian, or a factory sensor that reports a critical fault after the machine has already failed. In a world where billions of IoT devices generate data every second, waiting for that data to travel all the way to a centralized cloud server and back is no longer good enough [2], [3].

Figure 1: Scenario showing why low-latency processing matters.

Image Source: AI-generated image

This is where Edge computing and Fog computing step in – two computing paradigms built to bring processing power closer to where the data is created.

While both aim to reduce distance between data generation and data processing, they take different paths to get there. Edge computing pushes computations directly to the device or sensor level, while fog computing creates an intermediate layer between the edge and the cloud, distributing processing across a wider network [2], [4].

This results in two architectures that look similar on the surface but differ in how they handle performance, latency, energy consumption and reliability.

We'll break down how Fog and Edge computing perform against each other across critical parameters showing where each one shines and help you understand which one is the right fit — not just in theory, but in real-world application.

v Edge computing:
Edge computing is a distributed computing model that brings data processing as close as possible to the source of data generation – the “edge” of the network- instead of relying on a centralized, distant data centre or cloud. In simple terms, instead of sending raw data all the way to the cloud for analysis, the computation happens right on or near the device itself: a sensor, a smartphone or a security camera [1], [2], [7].

ü Working of Edge Computing system:

 When a device generates data, an edge computing setup processes that data locally, in real time, using compute resources embedded in or near the devices. Only the essential, processed information is then sent to the cloud for long-term storage or deeper analysis [2], [3].

ü Key Characteristics:

·       Real time decision making: Ideal for applications where even a slight delay can cause failure [3], [6].

·       Reduced Bandwidth usage: Less data needs to travel to the cloud, easing network congestion [2], [3].

·       Proximity: Processing happens at or very near the data source (within milliseconds of travel time) [8].

ü Examples of edge computing:

1.      autonomous vehicles processing sensor data to make split-second driving decisions [6].

2.      Smart security cameras detecting motion or faces without cloud dependency [3].

Figure 2: Edge computing

Image source: AI-generated image

vFog Computing

Fog computing is a distributed computing model that acts as an intermediate layer between edge devices and the cloud [2], [4], [9]. Rather than sending all data to a distant cloud server, or processing everything right at the device level, fog computing introduces local network nodes- routers, gateways or dedicated fog servers that sit closer to the ground to handle processing, storage and analysis for a group of connected devices [4], [5].

ü Working of Fog Computing:

Data from multiple edge devices is first sent to a nearby fog node instead of traveling all the way to the cloud [4], [5].

This fog node aggregates, filters, and processes the data from several sources at once, then forwards only the necessary insights or summaries to the cloud for deeper, long-term analysis [4], [9].

Key Characteristics:

·       Intermediate Layer: Positioned between edge devices and the cloud, not directly on the device itself [2], [4].

·       Centralized local processing: Can handle and coordinate data from multiple devices simultaneously

ü  Example:

1.      Smart city infrastructure managing traffic lights across multiple intersections.

2.      Industrial IoT systems coordinating data from an entire factory floor, not just a single machine.

 

 Figure 3: Fog computing [10]

Image source: https://www.geeksforgeeks.org/computer-networks/fog-computing

 

Comparison Between Edge Computing and Fog Computing with respect to IoT Architecture with an example of Smart City IoT project.

Case: SmartUrbia Smart City:

SmartUrbia is a city project designed to improve the quality of life by using real-time data analytics and automation. The project works with four integrated Internet of Things systems. Each of these four IoT systems has a role. They all use Edge and Fog computing together to function well. SmartUrbia is a hypothetical scenario created by the authors to illustrate the comparison; smart-city IoT use cases of this kind are discussed in [3] and [6].

System 1: Smart Traffic Management System

Cameras and sensors are placed at every traffic junction. They keep track of how much traffic flows on roads and pedestrian paths. The cameras have built-in processing chips. These chips allow the cameras to analyse data right where it is collected. SmartUrbia also uses Fog nodes at the district level. These nodes process data from more than ten traffic junctions in one district. They summarize the traffic data before sending it to the cloud for review.

System 2: Smart Surveillance & Public Safety System

A network of CCTV cameras and motion sensors has been set up in areas across the city. The cameras and motion sensors have processing units inside them. These units let the devices detect and react to events as they happen. For public safety, SmartUrbia has installed CCTV monitors in local control rooms. These monitors show footage from several cameras at once. The Fog nodes at each control room check the camera footage and look for security threats, helping staff respond quickly.

System 3: Smart Energy Grid System

The city uses smart meters to track and control the supply and demand of electricity. These smart meters have processing chips built in. This allows them to make decisions at the point where electricity is used. Fog nodes located at substations support the meters by analysing electricity use data. In case of a fault the Fog nodes reroute the power to keep the system running. The substation powers all homes and businesses within its area of responsibility.

System 4: Smart Waste Management System

Fill-level sensors are placed inside garbage bins. These sensors track how full each bin is across the city. The sensors send real-time updates about waste levels to Fog nodes at neighborhood collection points. Waste management workers use this data to plan collection routes based on need. Because of this collection times are not fixed. Bins fill up at different rates, and the system adapts to that.

 

Comparison:

Parameter	System	Edge Computing	Fog Computing
Location	Smart Traffic Management	Processing chip embedded directly in the intersection camera itself	A fog node positioned at a district hub, between the intersections and the city cloud
	Smart Surveillance & Public Safety	Processing unit built into the CCTV camera at the park or street	A fog node stationed at a local control room overseeing multiple cameras in the area
	Smart Energy Grid	Processing chip inside the smart meter installed in each household	A fog node located at a substation, managing multiple households in that zone
	Smart Waste Management	Sensor chip inside each individual garbage bin	A fog node placed at a neighborhood collection point, managing multiple bins
Performance	Smart Traffic Management	Each intersection camera detects vehicles and adjusts individually	A district fog node coordinates green-light synchronization across more than 10 intersections
	Smart Surveillance & Public Safety	A single camera detects motion/intrusion on-site	A fog node cross-references footage across multiple cameras in a park to confirm threat
	Smart Energy Grid	A smart meter detects a local voltage fluctuation in one household	A fog node balances load across an entire residential area during peak hours
	Smart Waste Management	A bin sensor detects its own fill level	A fog node cross-references all bin fill levels in a neighborhood to optimize pickup routes
Latency	Smart Traffic Management	A pedestrian sensor initiates signal change immediately	Coordinating signal timing across an entire street takes a few extra seconds to complete
	Smart Surveillance & Public Safety	A camera flags a break-in as soon as it detects motion	The fog node takes extra seconds to correlate footage between multiple cameras to confirm break-in
	Smart Energy Grid	A smart meter reacts immediately to a local power surge	A fog node takes extra seconds to rebalance load across multiple buildings in the zone
	Smart Waste Management	Near-instant response is not a priority for this system	A fog node compiles bin data periodically (every few hours) – latency is acceptable
Energy Consumption	Smart Traffic Management	Each camera's in-built processor uses continuous power to analyze video 24/7	The fog node (with stable power source) takes heavier lifting off individual cameras
	Smart Surveillance & Public Safety	Battery-backed cameras drain power constantly when doing on-device motion detection	The fog node handles heavier lifting of correlating footage across cameras; individual cameras conserve battery
	Smart Energy Grid	Smart meters use extra energy to process consumption data locally	The fog nodes handle heavier lifting of load-balancing calculations; individual meters conserve energy
	Smart Waste Management	Bin sensors would drain battery constantly if they processed complex routing algorithms themselves	Bin sensors stay in ultra-low power mode; only fog node does heavier lifting of route planning
Reliability	Smart Traffic Management	An individual intersection keeps functioning safely even if it loses connection to the city	If one fog node fails, a neighboring fog node can temporarily take over traffic coordination for that zone
	Smart Surveillance & Public Safety	A camera keeps recording and detecting motion locally even during network outage	If a fog node fails, cameras can temporarily operate in edge mode (independent functionality)
	Smart Energy Grid	A smart meter can still detect and report a local fault even without cloud connection	If one fog node fails, load-balancing for that zone may be delayed, but other zones remain unaffected
	Smart Waste Management	A bin sensor still records fill level locally even if it loses connection	If a fog node fails, waste pickup optimization pauses temporarily but will resume once connection is re-established

Note: The SmartUrbia rows above are illustrative examples based on the hypothetical scenario described earlier. The general characteristics they demonstrate are drawn from the sources cited: edge computing at or near the device [1], [2], [3], [8] and fog computing at an intermediate layer [4], [5], [9]. Energy savings on devices depend on the deployment and are not guaranteed.

 

***

 

 

 

References

[1]         Amazon Web Services (n.d.) What is edge computing? Available at: https://aws.amazon.com/what-is/edge-computing/ (Accessed: 28 September 2026).

[2]         IBM (2026) What is edge computing? Available at: https://www.ibm.com/think/topics/edge-computing (Accessed: 28 September 2026).

[3]         Flinders, M. and Smalley, I. (2025) Edge computing for IoT. IBM. Available at: https://www.ibm.com/think/topics/iot-edge-computing (Accessed: 28 September 2026).

[4]         Iorga, M., Feldman, L., Barton, R., Martin, M.J., Goren, N. and Mahmoudi, C. (2018) Fog computing conceptual model. NIST Special Publication 500-325. Gaithersburg, MD: National Institute of Standards and Technology. Available at: https://doi.org/10.6028/NIST.SP.500-325 (Accessed: 28 September 2026).

[5]         National Institute of Standards and Technology (2018) Fog computing conceptual model. Available at: https://www.nist.gov/publications/fog-computing-conceptual-model (Accessed: 28 September 2026).

[6]         Powell, P. and Smalley, I. (2025) Edge computing: top use cases. IBM. Available at: https://www.ibm.com/think/topics/edge-computing-use-cases (Accessed: 28 September 2026).

[7]         Darling, G. (2025) Overview of edge computing. IBM Developer, 8 July. Available at: https://developer.ibm.com/articles/what-is-edge-computing/ (Accessed: 28 September 2026).

[8]         Amazon Web Services (n.d.) AWS for the edge. Available at: https://aws.amazon.com/edge/ (Accessed: 28 September 2026).

[9]         National Institute of Standards and Technology (2018) NIST releases report on fog computing for Internet of Things devices. Available at: https://csrc.nist.gov/news/2018/fog-computing-for-internet-of-things-devices (Accessed: 28 September 2026).

[10]       GeeksforGeeks (n.d.) Fog computing. Available at: https://www.geeksforgeeks.org/computer-networks/fog-computing (Accessed: 28 September 2026).

 

 

 

 

Authored by Team Loading… (first year students of B.Eng(Hons) Software Engineering Vidyalankar institute of International Education (VIIE)).

Anirudh Shetty.

Atharva Kurle.

Kunjal Jaiswal.

Samiksha Surve.

Yuvraj Parab.

 

Under the guidance of: Dr. Amit K Nerurkar.(Associate Professor, Vidyalakar Institute of Technology)