Adaptive Fog Computing Architecture Based on IoT Device Mobility

and Location Awareness

Narek Naltakyan

Institute of Information and Telecommunication Technologies and Electronics,

National Polytechnic University of Armenia, Armenia

Keywords: Fog Computing, IoT Mobility, Location Awareness, Adaptive Architecture, Dynamic Orchestration,

Edge Computing.

Abstract: In the modern world, we are increasingly seeing the involvement of IoT devices in our daily lives, whether in

the form of sensors, smart homes, smart cars, or several other devices and equipment. Fog computing is a

concept that is intended to stimulate the development of such systems. However, a problem arises when the

operation of a given device or equipment depends on the location. This paper considers a new adaptive fog

computing architecture that dynamically manages IoT device mobility through intelligent location awareness

and geographic area orchestration. In the proposed system, devices are divided into two parts according to

location: static and dynamic, which allows optimizing resource allocation and service approach. The

architecture features include dynamic geographic area division, adaptive scaling of fog nodes, and the use of

a smart location service that can work both in the cloud and in the fog, depending on the number and density

of user requests. The system automatically scales infrastructure based on real-time demand and the geographic

distribution of devices, while maintaining quality of service.

1 INTRODUCTION

With the rapid development of technology, IoT

devices have become widespread in areas such as

smart homes and cities, autonomous vehicles,

factories, and health monitoring systems. Current

forecasts indicate that by the end of 2025, 75 billion

IoT devices will be connected, creating an

unprecedented volume of data that requires real-time

processing and context-aware responses. Traditional

cloud-centric architectures face serious challenges in

supporting this scale of distributed, mobile, and

latency-sensitive applications. Fog computing as a

paradigm was created to address these challenges by

bringing computing resources closer to data sources

and end-users. However, existing implementation

schemes assume network topologies and static device

locations, failing to adapt to the dynamic nature of

modern IoT device deployments, where device

locations are variable and cannot be considered

constant, changes in network conditions, and, in

addition, differences in service requirements over

time and geography.

1.1 Problem Statement

When using fog computing architectures, we face

several important issues when it comes to mobile IoT

devices:

 Static Resource Allocation: Existing systems

pre-allocate resources based on patterns, rather

than dynamically adapting to the geographic

location of devices and changes in their

location.

 Location Agnostic Service Management: In

existing architectural approaches, all devices

are considered static in terms of location, i.e.,

they do not change location over time.

 Limited Scalability Mechanisms: Another

issue is the lack of mechanisms for dynamically

scaling the infrastructure based on real-time

demands and geographic service distribution.

 Inefficient Mobile Device Handling: In

existing systems, there is no separate

distinction between static and dynamic IoT

devices in terms of location, which leads to

additional costs and insufficient support for

high mobility applications.

500

Naltakyan, N.

Adaptive Fog Computing Architecture Based on IoT Device Mobility and Location Awareness.

DOI: 10.5220/0013783900004000

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2025) - Volume 2: KEOD and KMIS, pages

500-505

ISBN: 978-989-758-769-6; ISSN: 2184-3228

1.2 Research Objectives

This research aims to develop an adaptive fog

computing architecture with the following primary

objectives:

 Design an intelligent location-aware fog

orchestration system that dynamically manages

fog nodes based on real-time device

distribution and demand patterns.

 Implement different strategies depending on

the type of IoT device deployment or location,

which will allow you to optimize resource

utilization while maintaining service quality

across diverse mobility scenarios.

 Implement infrastructure scaling mechanisms,

which should be adaptive and adjust the

geographic coverage of fog nodes based on the

number of requests and the required resources.

Figure 1: Hierarchical Adaptive Fog Computing

Architecture.

 Cloud Layer: Provides global coordination

and backup services.

 Location Service: Intelligent orchestration and

device mapping.

 Fog Layer: Distributed processing nodes with

adaptive coverage.

 IoT Layer: Heterogeneous devices with

varying mobility patterns.

1.3 Key Contributions

This scientific paper includes several significant

contributions to fog computing research and practice:

Novel Architectural Framework: The

architectural model discussed in this article is a

hierarchical location-aware fog computing

architecture that intelligently orchestrates resources

based on IoT device mobility patterns and geographic

services distribution.

Dynamic Area Management Algorithm: We have

developed an algorithm that adaptively changes the

geographical area under the responsibility of a fog

node based on real-time requirements. In this

architecture, 3 algorithms for distributing

responsibilities are proposed:

1. Using a load balancer and dividing requests

into two or more nodes.

2. Increasing the scope of responsibility if the

number of requests is less than X.

3. Adding a new fog node and dividing the

geographical area under the responsibility of

the nodes if the count of requests is greater

than X.

Differentiated Mobility Services: A service

differentiation strategy has been developed, where the

key argument is the type of IoT device location - static

or dynamic - that optimizes the system performance,

taking into account the specific requirements of the

devices.

Comprehensive Performance Validation: Results

of extensive tests are provided, which show a

significant performance improvement over existing

static fog computing approaches across multiple

evaluation metrics and deployment scenarios.

2 RELATED WORKS

2.1 Fog Computing Fundamentals

Fog computing builds on cloud computing by moving

it to the network's edge. This enables distributed

computing, meaning that data processing, storage,

and network services are performed closer to IoT

devices and end users, rather than just on centralized

cloud servers. Bonomi and his colleagues were the

first to introduce the idea of fog computing as an

approach aimed at solving common problems

associated with cloud systems: latency, network

bandwidth, and reliability. The OpenFog Consortium,

now part of the Industrial Internet Consortium, has

created standardized models that describe fog

computing as a system architecture that is distributed

across multiple layers of cloud, fog computing, and

IoT devices.

Recent progress in fog technologies has

emphasized optimizing how resources are used,

service deployment, and combining with new

technologies such as 5G networks and artificial

intelligence. However, much of the current research

is based on stable and immobile networks and doesn’t

Adaptive Fog Computing Architecture Based on IoT Device Mobility and Location Awareness

501

address the changing challenges posed by mobile IoT

devices.

2.2 Mobility-Aware Fog Computing

Limited research has addressed mobility

considerations in fog computing architectures.

Qayyum et al. proposed a mobility-aware hierarchical

fog framework for Industrial IoT that considers

device mobility in fog node placement decisions.

Their work focuses primarily on industrial

environments with predictable mobility patterns

rather than general-purpose mobile IoT scenarios.

Mahmud et al. developed a mobility-aware fog

computing framework that uses prediction algorithms

to anticipate device movement and proactively

allocate resources. While this approach shows

promise, it lacks comprehensive geographic area

management and does not address dynamic

infrastructure scaling based on mobility patterns.

Singh et al. conducted a comprehensive survey of

mobility-aware fog computing in dynamic networks,

identifying key challenges including mobile node

utilization, service continuity, and resource

optimization. Their analysis reveals significant gaps

in current approaches, particularly in adaptive

infrastructure management and location-aware

service orchestration.

2.3 Location-Aware Computing

Systems

In the world of information technology, location

awareness has been widely studied in the context of

mobile computing, from purely context-aware

applications to location-based services. But while

traditional location-aware systems rely on the

location of individual devices, they do not focus on

the systematic geographic optimization of distributed

resources.

Kalaria et al. have developed adaptive context-

aware access control systems that use location

information to make security decisions in IoT

environments. However, even in this case, there is no

consideration of resource management or dynamic

infrastructure scaling, which can be problematic

when there are a large number of users and/or data

flows.

The concept of Location-Aware Phase Frames,

studied by Ahmad et al., proposed a lightweight

Location-Aware Phase Frame (LAFF) for QoS

optimization. However, the work focuses on the

selection of phase nodes based on the user location,

but lacks dynamic area management and adaptive

scaling mechanisms.

2.4 Dynamic Resource Discovery vs.

Adaptive Geographic

Orchestration

There are significant differences between our

proposed architecture and existing dynamic resource

discovery (DRD) approaches in the field of fog

computing. Although both address resource

management challenges, they operate at

fundamentally different architectural levels and

address different problem domains.

Dynamic Resource Discovery Scope: Existing DRD

research focuses on runtime discovery and allocation

of computational resources within fixed

infrastructure topologies. Xu et al. developed DRAM

for load balancing through "static resource allocation

and dynamic service migration" to existing fog nodes.

Similarly, FogBus2 implements dynamic resource

discovery mechanisms to assist new entities in joining

established fog systems. These approaches work

reactively within predetermined infrastructure

boundaries.

Architectural Variation: Our research addresses

fundamental, broader problem areas. Our proposed

architecture addresses “How should I optimally

organize the geographical coverage” and “When

should I expand or contract the service areas”, while

DRD answers “Which fog nodes are available” and

“How should I distribute the load among the available

nodes”.

Complementary Technologies: Our proposed

architecture and DRD extensions can be used together

to enable DRD to identify existing fog nodes, while

our architecture can use dynamic resource discovery

as a component technology within a broader

geographic orchestration framework. Our location

service can use DRD mechanisms to identify existing

fog nodes, while our area management algorithms

determine optimal geographic boundaries and scaling

decisions.

Novel Geographic Intelligence: Our research

introduces geographic area management as a primary

optimization dimension, which existing DRD

approaches do not address. Current resource

discovery focuses on computational resource

allocation without considering the spatial

organization of fog infrastructure or the geographic

optimization of service coverage areas.

KMIS 2025 - 17th International Conference on Knowledge Management and Information Systems

502

3 SYSTEM ARCHITECTURE

Figure 2: Dynamic Geographic Area Coverage with Fog

Nodes and IoT Devices.

Key Features:

• F1: Medium density area with balanced

static/mobile devices.

• F2: High-density area with contracted

coverage for efficiency.

• F3: Low-density area with extended

coverage for broad service.

• F4: Very high-density area with minimal

coverage, ready for scaling.

3.1 Architectural Overview

Our adaptive fog computing architecture implements

a three-tier system: IoT devices<=>fog

infrastructure<=>cloud resources. An orchestration is

performed through an intelligent location service that

provides load balancing, area assignment among fog

nodes, and dynamic resource management.

The main innovation of the architecture is the

dynamic allocation of responsibility areas across

geographic areas to fog nodes based on real-time

device distribution, demand patterns, and

infrastructure capacity. This enables a fog

environment that adapts to real-time requirements

while maintaining quality of service guarantees

across different IoT deployment scenarios.

Key Architectural Components:

1. Location Service Layer: Provides

centralized intelligence for device-to-fog

mapping, geographic area management, and

infrastructure scaling decisions

2. Fog Orchestration Layer: Manages

distributed fog nodes, handles service

placement, and coordinates resource

allocation across geographic areas

3. Device Management Layer: Implements

differentiated services for static and

dynamic IoT devices with optimized

communication protocols

4. Adaptive Scaling Controller: Monitors

system performance and triggers

infrastructure scaling events based on

demand patterns and service quality metrics

3.2 Location Service Architecture

The location service is the cornerstone of the above

architecture, enabling intelligent zoning and rezoning

over time. This service provides communication

between IoT devices and the fog infrastructure,

informing IoT devices over time whether their

responsible fog node has changed or not, while

continuously optimizing the zoning.

Adaptive Service Deployment: The location service

itself applies an adaptive deployment strategy based

on the number and density of requests. It is initially

deployed in the cloud and runs from the cloud for

low-density deployments, but migrates to distributed

fog nodes as requests grow.

Geolocation Management: The service maintains

real-time state information, including device

locations, fog node capabilities, current area

boundaries, and historical demand patterns. This

enables decisions to be made for area adjustments and

scaling events.

Query Optimization: For dynamic devices that

require frequent location updates, the service includes

a query optimization algorithm that attempts to

perform predictive caching based on historical data,

direction, and speed of movement.

3.3 Location Service Architecture

The dynamic area management system represents a

novel contribution that enables fog nodes to

automatically adjust their geographic coverage areas

based on real-time demand characteristics and service

quality requirements.

Area Sizing Algorithm: The coverage area of each

fog node is dynamically scaled. When there are few

requests, it is expanded to a specified maximum size

that is allowed for that fog node, and is reduced when

the number of requests exceeds the specified size.

This algorithm balances the load on the fog nodes and

optimizes resource utilization.

Adaptive Fog Computing Architecture Based on IoT Device Mobility and Location Awareness

503

Figure 3: Area Adaptation Algorithm Flow.

Minimum Area Constraints: The system has a

minimum area size constraint, when the area of

responsibility of a given fog node shrinks to a defined

minimum, but the number of requests does not

decrease, then another fog node and load balancer are

added, added, and both fog nodes are responsible for

the given area.

Border Coordination: Neighboring fog nodes

coordinate area boundary adjustments to avoid gaps

in coverage and service continuity

3.4 Differentiated Device Management

The architecture is based on device classification and

management strategies that allow optimizing system

performance by segmenting devices according to

location characteristics.

Static Device Protocol: Devices that do not change

their location during operation will be called static in

our architecture, such as sensors, fixed infrastructure,

and permanent installations.

Dynamic Device Protocol: Devices that change their

location during operation will be called dynamic in

this architecture. These are: vehicles, wearable

devices, portable sensors. I also find it noteworthy to

note that the frequency of requests increases

depending on the speed of the device's movement,

since the device needs to "know" which node is

responsible for a given area.

Predictive Assignment: For devices with predictable

paths, such as public transit, delivery vehicles,

commuters, the system uses predictive fog node

assignment to minimize handover delay and improve

service continuity.

4 DYNAMIC ORCHESTRATION

ALGORITHMS

4.1 Geographic Area Optimization

Algorithm

The geographic area optimization algorithm

continuously modifies the coverage areas of fog

nodes to balance quality of service, efficient resource

use, and infrastructure efficiency. It operates in real

time based on metrics such as request rate, response

latency, and fog node load levels.

Mathematical Formulation:

For each fog node i, the optimal area size A_i is

determined by:

A_i = f(R_i, C_i, L_i, Q_target)

Where:

• R_i = request frequency in area i

• C_i = computational capacity of fog node i

• L_i = average latency from area boundary to

fog node i

• Q_target = target service quality threshold

Area Adjustment Decision Logic:

IF (R_i > threshold_high AND A_i > A_min):

REDUCE area size by scaling factor α

ELIF (R_i < threshold_low AND utilization_i

< utilization_target):

EXPAND area size by scaling factor β

ELIF (A_i ≤ A_min AND R_i >

threshold_critical):

TRIGGER fog node addition

4.2 Location Service Placement

Algorithm

The location service placement algorithm selects

where to deploy location service components based

on client distribution, request patterns, and network

topology characteristics.

Cloud vs. Fog Placement Decision:

IF (client_density < threshold_fog AND

query_frequency < threshold_freq):

DEPLOY location service in cloud

ELSE:

KMIS 2025 - 17th International Conference on Knowledge Management and Information Systems

504

DEPLOY location service in fog

infrastructure IMPLEMENT regional

service distribution

Figure 4: Area optimization algorithm.

4.3 Device Assignment Optimization

Static Device Assignment:

fog_node = SELECT fog_node_i WHERE:

device_location ∈ coverage_area_i AND

capacity_available_i > service_requirements

AND latency(device, fog_node_i) <

latency_threshold

Dynamic Device Assignment:

fog_node = SELECT fog_node_i WHERE:

predicted_location ∈ coverage_area_i AND

handoff_probability < handoff_threshold

AND service_continuity_score >

continuity_threshold

REFERENCES

Ahmad, N., Hussain, M., Riaz, N., Subhani, F., Haider, S.,

Alamri, K. S., & Alasmary, W. (2021). Enhancing SLA

compliance in Cloud computing through efficient

resource provisioning. Future Generation Computer

Systems, 118, 365-380.

Bonomi, F., Milito, R., Zhu, J., & Addepalli, S. (2012). Fog

computing and its role in the internet of things. In

Proceedings of the first edition of the MCC workshop

on Mobile cloud computing (pp. 13-16). ACM.

Kalaria, R., Gupta, A., Singh, R., & Sharma, P. (2020).

Adaptive context-aware access control framework for

IoT environments. IEEE Access, 8, 42462-42475.

Mahmud, R., Kotagiri, R., & Buyya, R. (2018). Fog

computing: A taxonomy, survey and future directions.

In Internet of everything (pp. 103-130). Springer.

Qayyum, T., Malik, A. W., Khattak, M. A. K., Khalid, O.,

& Khan, S. U. (2018). FogNetSim++: A toolkit for

modeling and simulation of distributed fog

environment. IEEE Access, 6, 63570-63583.

Singh, S., Sharma, P. K., Yoon, B., Shojafar, M., Cho, G.

H., & Ra, I. H. (2020). Convergence of blockchain and

artificial intelligence in IoT network for the sustainable

smart city. Sustainable Cities and Society, 63, 102364.

Xu, R., Wang, Y., Cheng, Y., Zhu, Y., Xie, Y., Sani, A. S.,

& Yuan, D. (2021). Improved particle swarm

optimization based workflow scheduling in cloud-fog

environment. In International Conference on Business

Process Management (pp. 337-347). Springer.

Adaptive Fog Computing Architecture Based on IoT Device Mobility and Location Awareness

505