Network Diffusion Algorithms and Simulators in IoT and Space IoT: A Systematic Review

1. Introduction

As the Internet of Things (IoT) continues to expand rapidly across various industries, gaining insights into how information spreads within these networks has become increasingly important [1]. Network diffusion algorithms and simulators serve as powerful tools for analyzing the propagation of data, malware, and other forms of communication within IoT systems [2,3]. In the context of Space IoT, where device heterogeneity, dynamic topologies, and large-scale networks pose unique challenges, the demand for robust and scalable diffusion models is even more pronounced.

Recent advancements in IoT applications have pushed the boundaries of what these technologies can achieve, particularly in domains involving drones and satellites. Drones are increasingly used for tasks such as surveillance, delivery, and environmental monitoring, requiring real-time communication and coordination within IoT networks. Satellite-based IoT systems are enhancing connectivity in areas where terrestrial networks are unavailable, such as remote regions, oceans, and deserts. By leveraging Low Earth Orbit (LEO) satellites, IoT devices can maintain consistent communication over vast distances, enabling applications in environmental monitoring, disaster management, and global logistics. These advancements address critical challenges in the IoT, including ensuring reliable communication over long distances, optimizing energy usage, and managing complex network topologies.

This paper presents a systematic literature review (SLR) focusing on network diffusion algorithms and simulators utilized in IoT and Space IoT studies. By examining research published between 2013 and 2024, this review identifies key diffusion algorithms, including epidemic, cascading, and threshold-based models, and evaluates the performance of widely used network simulators, such as NS-3, Cooja, and OMNeT++ [4,5,6]. Additionally, the metrics used to assess these algorithms and simulators, such as scalability, energy efficiency, latency, and packet delivery ratio, are explored.

The objective is to provide researchers and practitioners with a comprehensive analysis of the tools and methodologies used in studying network diffusion in IoT environments. By highlighting the strengths and limitations of various algorithms and simulators, this review aims to guide the selection of appropriate tools for simulating and evaluating IoT and Space IoT systems.

This paper makes several significant contributions. It presents a comprehensive systematic literature review of network diffusion algorithms, simulators, and topologies within IoT and Space IoT, encompassing research from 2013 to 2024. Unlike previous surveys, this study is the first to integrate all three aspects—diffusion algorithms, network simulators, and network topologies—into a single, cohesive framework. By adhering to the PRISMA methodology, it ensures a rigorous and transparent selection process, addressing gaps left by earlier reviews. The study categorizes and examines key diffusion models, including epidemic, cascading, and threshold-based approaches, assessing their effectiveness across diverse IoT and Space IoT environments. Additionally, it evaluates widely used network simulators, such as NS-3, Cooja, and OMNeT++, analyzing their capabilities, scalability, and suitability for different network scenarios. Beyond this, the review identifies essential performance metrics—including scalability, energy efficiency, latency, and packet delivery ratio—providing valuable insights into the strengths and limitations of various diffusion models and simulators. Finally, the study outlines open challenges and future research directions, underscoring the need for more adaptable and resource-efficient models to meet the evolving demands of IoT and Space IoT systems.

2. Methodology

2.1. Research Question Formulation

A crucial step in conducting an SLR is to formulate clear and focused research questions that guide the entire review process. The following research questions were defined to address specific aspects of diffusion algorithms, network simulators, and their performance in IoT and Space IoT configurations:

Research Question 1: General What are the key diffusion algorithms and network simulators utilized in the literature to study network diffusion in IoT and Space IoT configurations, and how do these algorithms perform in various scenarios?
Research Question 2: Network Simulators Which network simulators are commonly employed in the literature to simulate and study network diffusion in IoT and Space IoT configurations, and what are their features and limitations?
Research Question 3: Network Diffusion Algorithms What are the key characteristics, strengths, and weaknesses of network diffusion algorithms used in studies focusing on network diffusion in IoT and Space IoT configurations?
Research Question 4: Network Architecture How does the choice of network architecture impact the performance of network diffusion algorithms in IoT and Space IoT configurations?
Research Question 5: Network Diffusion Performance Evaluation What performance metrics are commonly used to evaluate the effectiveness of network diffusion algorithms in IoT and Space IoT configurations, and how do these algorithms perform under various scenarios?
Research Question 6: Performance evaluation of the network simulator How do existing network simulators perform when simulating network diffusion in IoT and Space IoT configurations, and what are the key metrics used to evaluate their performance?

These research questions are designed to guide this literature search and review, ensuring that the existing literature relevant to the study’s objectives is comprehensively analyzed.

2.2. Search Strategy

2.2.1. Identification of Databases

For this SLR, Google Scholar and Scopus were selected as primary databases due to their extensive coverage and relevance to the research topics of network diffusion, the IoT, and Space IoT.

2.2.2. Keywords and Phrases

After careful consideration and filtering, the research narrowed down its focus to key terms and concepts essential for the investigation. The following keywords and phrases formed the foundation of the search strategy:

Network Diffusion;
IoT;
Diffusion Algorithms;
Propagation Models;
Simulation;
Network Simulators.

2.2.3. Combinations Using Boolean Operators

Search strings combining relevant keywords and Boolean operators were developed to ensure a comprehensive coverage of the relevant literature:

Combination 1: ("network diffusion" OR "information propagation") AND ("IoT" OR "Internet of Things") AND ("diffusion algorithms" OR "propagation models") AND ("simulation" OR "network simulators");
Combination 2: (("Network" AND "Diffusion") OR ("Network" AND "Propagation") OR ("Network" AND "Spreading") OR ("Network" AND "Simulator") OR ("IoT" AND "Diffusion") OR ("IoT" AND "Propagation") OR ("IoT" AND "Spreading") OR ("IoT" AND "Simulation")) AND NOT Deep learning NOT Neural network.

2.2.4. Pilot Searches

Pilot searches were conducted to assess the relevance and specificity of the search strings. The search strategy was refined on the basis of the initial results to ensure optimal retrieval of the relevant literature.

2.2.5. Papers Using API

Various tools were tested to retrieve relevant articles for systematic review of the literature. Ultimately, Harzing’s Publish or Perish tool [7] was selected for its ability to retrieve and analyze academic citations. This tool provided valuable information on the citation metrics, including the number of articles, total citations, and h-index.

2.3. Inclusion and Exclusion Criteria

Inclusion and exclusion criteria are essential to ensure the selection of studies that align with research objectives while maintaining precision and relevance.

2.3.1. Objective

The objective of establishing inclusion and exclusion criteria is to create a clear framework for selecting studies that meet defined standards, thereby enhancing the quality and focus of the systematic literature review.

2.3.2. Inclusion Criteria

Academic Sources: These included academic sources from reputable journals, conference proceedings, and academic publishers to ensure the reliability of the information. While peer-reviewed papers were prioritized, some preprints from ArXiv were also considered. These were included based on their relevance, contribution from established researchers, and citation impact.
Publication Date: This included studies published within the last 12 years (from 2013 to 2024) to capture recent advancements in the field.
Top-Ranked Papers: These included the first papers ranked from 1 to 50 on the Publish or Perish software (8.17.4863) using the specified combination.
Language: This included studies published in English to align with the research team’s language proficiency.

2.3.3. Exclusion Criteria

Non-Academic Sources: We excluded non-academic sources, such as blogs or commercial publications, to maintain the reliability of the information.
Outdated Studies: We excluded studies published before 2013 to focus on recent developments and trends in network diffusion algorithms.
Non-Top-Ranked Papers: We excluded papers ranked after the top 50 on the Publish or Perish software (8.17.4863) using the specified combination.
Language: We excluded studies published in languages other than English, as the research team is proficient in English.

These criteria were established to ensure the selection of studies that meet defined standards and contribute to the objectives of the systematic literature review.

2.4. Study Selection

The study selection process happened in three main steps, each aimed at narrowing down the list of papers to focus on the most relevant ones for the analysis. The process followed the PRISMA flow diagram, which is a well-established method for conducting systematic literature reviews. As shown in the diagram, the process involved three main steps: identification, screening, and inclusion of studies based on specific criteria. The flow diagram outlines the stages from the identification of records through database searches (Google Scholar and Scopus) to the final selection of studies included in the review. Figure 1 presents the detailed breakdown of these steps [8].

2.4.1. Step 1: Finding and Initial Check

Using Harzing’s Publish or Perish tool, more than 12,000 papers were gathered based on keyword searches. This step facilitated the identification of a broad selection of papers as an initial dataset for this review.

2.4.2. Step 2: Applying Criteria and Checking Titles/Abstracts

To ensure a focus on the most relevant and high-quality papers, specific selection criteria were established. These included the publication date, language, and the paper’s relevance to the research questions. A thorough screening of the titles and abstracts was conducted to assess their alignment with the study’s objectives.

This manual selection process was carried out without automated tools, relying on careful reading and judgment to ensure accuracy. By applying this approach, the list of papers was refined into a more targeted selection for in-depth review.

2.4.3. Details of Criteria and Screening

2.4.4. Step 3: Reading Full Papers and Checking Quality

In the final step, the full papers from the selected studies were reviewed to determine their relevance and reliability for the study.

2.4.5. Final Set of Papers for Analysis

This approach ensured the selection of the most relevant papers that contribute effectively to this research.

3. Survey Comparison

The following questions were used to compare various surveys (Table 1 shows the detailed comparison):

Q1: Does this survey include network simulators?
Q2: Does this survey include diffusion algorithms?
Q3: Does this survey include categories of diffusion algorithms?
Q4: Does this survey include network topologies?
Q5: Are the algorithms, simulators, and topologies covered until 2024?
Q6: Does this survey evaluate the performance of algorithms, simulators, or topologies using specific metrics?

Table 1 provides a comparison of the surveys in terms of these six questions. It highlights the inclusion of network simulators, diffusion algorithms, algorithm categories, network topologies, and coverage until 2024, as well as performance evaluation using specific metrics.

However, none of the previous surveys provided a complete answer to all six questions. As shown in Table 1, each survey either lacks coverage in one or more aspects or does not extend its analysis up to 2024, leaving gaps in providing a comprehensive overview of diffusion algorithms and simulators.

Table 1.
Comparison of surveys on diffusion algorithms and simulators.

N	Year	Q1	Q2	Q3	Q4	Q5	Q6
[9]	2013	✗	✓	✓	✓	✗	✗
[10]	2014	✓	✗	✗	✓	✗	✓
[11]	2014	✗	✗	✓	✓	✗	✓
[12]	2015	✗	✗	✗	✗	✗	✗
[13]	2016	✗	✓	✗	✓	✗	✗
[14]	2017	✓	✓	✓	✓	✗	✓
[15]	2017	✗	✓	✓	✓	✗	✓
[16]	2019	✓	✗	✗	✗	✗	✓
[17]	2020	✓	✗	✗	✗	✗	✓
[18]	2022	✓	✗	✗	✗	✗	✓
[19]	2023	✓	✗	✗	✓	✓	✓
[20]	2024	✓	✗	✗	✓	✓	✓
This paper	2025	✓	✓	✓	✓	✓	✓

4. Analysis

4.1. Analysis of Research Question 2 and Research Question 6: Network Simulators in IoT and Space IoT Configurations and Their Performance Evaluation

Through a comprehensive review of the selected literature from 2013 to 2024, several network simulators were identified as prevalent tools for simulating and studying network diffusion in IoT and Space IoT environments. These simulators vary in terms of capabilities, scalability, extensibility, and suitability for different IoT scenarios. Additionally, their performance was evaluated based on key metrics such as scalability, energy consumption, latency, and packet delivery ratio (PDR).

Table 2 and Table 3 present an overview of the network simulators, emulators, and extensions used in various studies. These tables categorize the simulators based on the following:

The year of publication—Identifying trends in network simulation research.
Simulator or extension name—Listing either the standalone simulator or a new module/extension built upon an existing tool.
Base simulation platform—Showing whether a simulator is standalone or built as an extension of existing tools such as NS-3, OMNeT++, or Cooja.

The inclusion of these tables provides insight into the evolution of IoT network simulation, helping researchers select the most suitable simulation tools based on their specific research requirements. The wide range of simulators listed demonstrates the growing need for customizable, scalable, and domain-specific network modeling solutions in both traditional IoT and Space IoT configurations.

The most commonly employed network simulators include the following:

NS-3;
Cooja (with Contiki OS);
OMNeT++.

Table 2.
Simulators, emulators, and extensions used in IoT and Space IoT (Part 1).

N	Year	New: Emulator, Simulator, or Module	Base Simulator
[21]	2013	High-level sensor simulator
[22]	2013	IoT-LAB
[23]	2013	SMSim
[24]	2013	New RFID module	NS-2
[25]	2013	EstiNet
[26]	2013	Nano-Sim	NS-3
[27]	2013		NS-3
[28]	2014	Extension	DEUS, NS-3, Cooja
[29]	2014	Extension	Cooja, Contiki
[30]	2014	Extension	NS-3
[31]	2014	Extension	Qualnet simulation platform
[32]	2014	SimuLTE	OMNeT++
[9]	2014	Multi
[33]	2015	Semantic Sensor Simulator
[34]	2015	NetSim
[6]	2015		Cooja, Contiki
[35]	2015	MmWave Module	NS-3
[36]	2015		SenseSim
[37]	2016	CupCarbon
[38]	2016	MobIoTSim
[4]	2016	Extension	NS-3
[39]	2016	ASSIST
[40]	2017	iFogSim
[41]	2017	Simulation Engine
[42]	2017	Algo Extension	Cytoscape
[14]	2017	Multi
[43]	2017	IOTSim	Cloudsim
[44]	2017	LoRaSIM
[45]	2017	ECMO simulation system
[46]	2017		COOJA, Contiki
[47]	2017	NoS simulator	OMNet++
[48]	2018	Extension	NS-3
[49]	2018	Extension	LoRaSim
[50]	2018	Extension	OMNet++ // Matlab
[51]	2018	Extension	NS-3
[52]	2018	IoT smart grid testbed simulator
[53]	2018		OPNET
[54]	2018	Extension	Cooja, Contiki
[55]	2018		QualNet 6.1
[56]	2018		CloudSim // Cooja Contiki
[57]	2018	CupCarbon
[5]	2018		OMNET++
[58]	2018		NS-2
[59]	2018	FlyNetSim	NS-3 // ArduPilot
[60]	2019	YAFS
[61]	2019	Extension	NS-3
[62]	2019	LoRa-MAB
[63]	2019		VLE
[64]	2019	IoTNetSim
[16]	2019		Multi

Table 3.
Simulators, emulators, and extensions used in IoT and Space IoT (Part 2).

N	Year	New: Emulator, Simulator, or Module	Base Simulator
[65]	2019		NS-3
[66]	2019	CyprIoT framework
[67]	2020		CupCarbon
[17]	2020		Multi
[68]	2020	RelIoT	NS-3
[69]	2020		Cooja, Contiki
[70]	2020		Cooja, Contiki
[71]	2020	Extension	NS-3
[72]	2020	STEP-ONE	ONE simulator
[73]	2020	DTMSim
[74]	2020
[75]	2020	Extension	OMNet++ // FogNetSim++
[76]	2020	Dockemu	NS-3
[77]	2020	xFogSim
[78]	2021	LoRaWANSim
[79]	2021	ROS-NetSim	NS-3 // Gazebo
[80]	2021	Extension	Cooja, Contiki
[81]	2021	IoTSim-Osmosis
[82]	2022	NS-3SAAS	NS-3
[83]	2022
[84]	2022		Cooja, Contiki
[85]	2022	Large-scale NB-IoT
[18]	2022	Simulator for blockchain with IoT
[86]	2022	Faulty IoT network simulator
[87]	2022	ADIperf framework
[88]	2022	FLORASAT	OMNeT++
[89]	2023		Cooja, Contiki
[90]	2023	Extension	NS-3
[91]	2023	FLoRa	OMNeT++
[92]	2023	StackNet	NS-3
[93]	2023	Kaala 2.0
[94]	2023	PIot
[95]	2023	NS-3 Docker Emulator	NS-3
[96]	2023		NS-2.35
[97]	2024	Lr-WPAN, 6LoWPAN	NS-3
[98]	2024	NDT	NS-3
[99]	2024	GothX	Gotham
[100]	2024	Extension	ZigBee network simulator
[101]	2024		GI-ITM
[102]	2024		Custom simulator (Python-based)
[103]	2024	MAPE	AFTER
[104]	2024	SWI-FEED	NS-3
[1]	2024	UDP module extension	Cooja, Contiki
[105]	2024		NetSim

4.1.1. NS-3

Overview

NS-3 is a widely used discrete-event network simulator designed for both research and educational applications. As an open-source platform, it enables the simulation of internet systems, making it ideal for various network scenarios, including IoT and Space IoT configurations [106].

Features

Extensibility: NS-3 supports a modular architecture, allowing easy integration and extension of new protocols and models.
Realism: Provides realistic simulation models for various network protocols and technologies.
Support for IoT Protocols: Includes modules and extensions for simulating IoT-specific protocols such as LoRaWAN and NB-IoT.
Visualization Tools: Supports integration with external visualization tools, including NetAnim [107] for 2D simulation and NetSimulizer [108] for 3D simulation. These tools enhance the analysis and representation of simulation results, providing detailed insights into network behavior and performance.
Scalability: Capable of simulating large-scale network scenarios, typically involving thousands to tens of thousands of nodes and plus, high traffic loads, and diverse communication protocols. The simulation complexity increases based on node interactions, mobility models, and protocol layers, making NS-3 suitable for evaluating extensive IoT and Space IoT deployments.
Support for Multiple Programming Languages: Primarily supports C++ and provides Python bindings for ease of use.
Support for Wireless Technologies: NS-3 provides built-in support for wireless communication models, including Wi-Fi, LTE, and 5G, with ongoing research into 6G integration. It allows for the testing of routing algorithms, noise models, fault tolerance mechanisms, and packet drop rates, making it ideal for large-scale IoT simulations.

Limitations

Steep Learning Curve: Requires substantial understanding of network protocols and programming proficiency, which may be challenging for beginners.
Resource-Intensive: Large-scale simulations in NS-3 can be computationally demanding, especially when modeling networks with over 10,000 nodes, high event densities, and complex protocol stacks. These simulations typically require processing times ranging from a few minutes to several hours, depending on the scenario’s complexity. For example, simulating a 6144-host network can take 11 to 27 h to advance just five seconds of simulated time [109]. Similarly, a 5G network simulation with 5700 nodes and 19 towers required 10.4 h to simulate 1.5 s of real-time activity [110]. Memory usage is also significant, with simulations demanding at least 8 GB RAM for moderate scenarios and exceeding 32 GB RAM for larger-scale simulations [111]. High-performance computing clusters with 128 GB to 1 TB RAM per node are often used to handle such large-scale models efficiently.
Limited Built-in Support for Some IoT Scenarios: May require additional customization and module development to accurately simulate certain IoT-specific scenarios.
Complex Configuration: Setting up simulations can be complex due to detailed configuration requirements.

Notable Studies Using NS-3

Ref. [30]: Simulated and compared different architectures’ performance in terms of delay, throughput, and control overhead.
Ref. [4]: Extended NS-3 with modules to simulate LoRa systems, enabling performance analysis in various scenarios.
Ref. [68]: Developed a reliability simulation framework using NS-3 to evaluate power, performance, temperature, and reliability aspects.
Ref. [95]: Integrated Docker with NS-3 to simulate realistic IoT botnet scenarios.

Performance Metrics

Scalability: NS-3 is widely recognized for its scalability, making it suitable for large-scale IoT applications like LoRa networks and UAV simulations.
Latency and Propagation Delay: It is often used to evaluate latency in large-scale networks, including fog and edge computing environments.
Packet Delivery Ratio (PDR): NS-3 is frequently used to evaluate the PDR in various IoT environments, especially when testing different communication protocols.

4.1.2. Cooja (With Contiki OS)

Overview

Cooja is a network simulator tailored for wireless sensor networks (WSNs) and IoT applications. It operates alongside the Contiki operating system, offering a versatile and scalable platform for simulating low-power and lossy networks [112].

Features

Hardware Emulation: Allows for the emulation of real hardware platforms, enabling accurate and realistic simulations.
Multi-level Simulation: Supports simulation at different abstraction levels, ranging from high-level network simulations, where entire communication stacks and protocol behaviors are modeled without detailed hardware interactions, to low-level hardware interactions, where individual sensor nodes, CPU instructions, and radio transceivers are emulated for precise performance analysis.
Energy Consumption Analysis: Provides tools to monitor and analyze energy consumption, which are crucial for IoT applications focused on energy efficiency.
Support for Various Radio Models: Includes multiple radio propagation models to simulate diverse environmental conditions.
Easy Integration of New Protocols: Facilitates the addition and testing of new network protocols within the simulation environment.
Graphical User Interface: Offers a user-friendly GUI for setting up and monitoring simulations, making it accessible to users with varying levels of expertise.
Wireless Communication and Fault Tolerance: Cooja is primarily designed for wireless sensor networks (WSNs) and low-power IoT applications. While it does not natively support LTE and 5G, extensions can be added. It includes configurable noise models, routing algorithms, and fault tolerance mechanisms for evaluating network reliability.

Requirements Addressed

NS-3 is equipped to handle diverse network simulation requirements, making it suitable for IoT and Space IoT applications. It supports wireless communication models, including Wi-Fi, LTE, and 5G, and ongoing research explores 6G integration. NS-3 also provides modules for evaluating routing algorithms, noise models, fault tolerance mechanisms, and packet drop rates. Its detailed physical (PHY) and medium access control (MAC) layer models allow for precise performance evaluation under various network conditions.

Limitations

Scalability Issues: May struggle with simulating very large-scale networks due to performance constraints.
Limited Support for High-Level Network Protocols: Focuses primarily on low-power networks and may require extensions to support heterogeneous IoT environments with diverse communication protocols (e.g., 5G, LPWAN, and edge computing), large-scale deployments exceeding 10,000 nodes, and highly dynamic network conditions such as mobile and space IoT scenarios.
Dependency on Contiki OS: Tight integration with Contiki OS may limit flexibility when experimenting with other operating systems or platforms.
Performance Overheads: Cooja’s performance degrades significantly when simulating large-scale networks, particularly as the number of nodes increases beyond several thousand, due to memory constraints and CPU-intensive real-time debugging. While small-scale simulations (under 500 nodes) generally run efficiently, studies indicate that execution time and memory consumption grow exponentially with network size [113]. Simulations exceeding several thousand nodes often require optimized memory management techniques, such as reducing log verbosity and adjusting simulation granularity, to mitigate excessive delays [114]. The execution time is further impacted by complex radio propagation models, as Cooja’s event-driven simulation approach tracks each node’s state changes at high granularity. Additionally, being Java-based, Cooja may introduce performance overheads related to Java Virtual Machine (JVM) memory management and garbage collection, potentially leading to unpredictable execution slowdowns. Researchers have reported that simulations with 10,000+ nodes experience severe slowdowns and may require modifications or high-performance computing environments for stability [115]. This highlights the need for scalability enhancements when using Cooja for large-scale IoT deployments.

Notable Studies Using Cooja

Ref. [6]: WSN-Maintain enhances Cooja by optimizing sensor coverage, extending network lifetime via redundant node management.
Ref. [54]: Simulated security attacks and defense mechanisms in IoT networks.
Ref. [70]: Evaluated the performance of IoT protocols under various network densities.
Ref. [84]: Analyzed DDoS attack scenarios and potential mitigation strategies in IoT networks.

Performance Metrics

Energy Consumption: Cooja focuses heavily on energy efficiency, making it a go-to tool for simulating low-power IoT networks.
Latency: The simulator is often evaluated based on latency in real-time applications, although it struggles with scalability when simulating larger networks.
Packet Delivery Ratio (PDR): The PDR is another critical performance metric used in Cooja, especially in resource-constrained IoT environments.

4.1.3. OMNeT++

Overview

OMNeT++ is a component-based modular simulation framework written in C++ that is primarily used to develop network simulators [88]. Its adaptability and extensibility make it a popular choice for modeling various network types, including IoT systems.

Features

Modular Architecture: Facilitates easy extension and customization through reusable modules.
Comprehensive Frameworks: Supports various frameworks like INET and Castalia for simulating different network protocols and environments.
Graphical Simulation Environment: Provides a robust GUI for designing, executing, and visualizing simulations.
Support for Complex Network Scenarios: Capable of simulating large and complex networks with diverse protocols and configurations.
Integration with Other Tools: Allows integration with external tools and libraries for enhanced simulation capabilities.
Active Community Support: Backed by a strong community providing support, tutorials, and extensions.
Wireless and Routing Algorithm Support: OMNeT++ supports multiple networking standards, including Wi-Fi, LTE, and 5G, via extensions such as INET. It enables detailed routing algorithm simulations, fault tolerance analysis, and network behavior modeling under various noise conditions.

Limitations

Complexity: The extensive features and configurations can be overwhelming for new users.
Performance Concerns: Simulating networks exceeding 10,000 nodes may lead to performance bottlenecks due to high memory consumption, increased event scheduling overhead, and computational complexity of detailed protocol modeling.
Steep Learning Curve: Requires a good understanding of C++ and network simulation concepts.

Notable Studies Using OMNeT++

Ref. [47]: Integrated OMNeT++ with Matlab for detailed network and system co-simulation.
Ref. [5]: Simulated and analyzed specific protocols for RFID-based IoT networks.
Ref. [88]: Developed FLoRaSaT using OMNeT++ to simulate satellite-based IoT networks with realistic orbital mechanics.

Performance Metrics

Scalability and Flexibility: OMNeT++ is highly flexible, capable of modeling complex networks, and is often used for multi-layered simulations.
Latency and Packet Loss: OMNeT++ simulations often focus on measuring latency and packet loss, particularly in complex, large-scale network environments.
Energy Consumption: While OMNeT++ is not specialized for energy monitoring, it can be used to evaluate energy-related metrics in extended simulations.

4.1.4. Comparative Summary

Table 4 provides a comparative analysis of NS-3, Cooja, and OMNeT++, summarizing their key features, limitations, and applicability for IoT and Space IoT simulations.

4.1.5. Conclusions

In summary, NS-3, Cooja, and OMNeT++ serve as powerful tools for simulating network diffusion in IoT and Space IoT environments, each offering distinct advantages and trade-offs. As illustrated in Figure 2, NS-3 emerged as the most widely adopted simulator, followed by Cooja and OMNeT++. NS-3 is particularly well suited for large-scale simulations, offering high realism and extensibility. Cooja excels in low-power IoT simulations, making it ideal for resource-constrained environments, while OMNeT++ stands out with its modular architecture, enabling the simulation of complex network scenarios. Choosing the appropriate simulator depends on specific research and application needs, with scalability, extensibility, and ease of use being key factors to consider. We summarize the simulators as follows:

NS-3 is recommended for detailed protocol analysis and large-scale network scenarios.
Cooja is ideal for low-power IoT applications and hardware emulation.
OMNeT++ is suitable for complex simulations involving diverse network protocols and configurations.

The selection of a simulator plays a crucial role in shaping the outcomes of IoT research, making it essential to evaluate the specific requirements of the study before choosing an appropriate simulation tool [116,117].

Beyond widely used simulators such as NS-3, Cooja, and OMNeT++, several JAVA-based network simulators have played a significant role in past research. GRUBIX and SHOX, in particular, were frequently utilized from 2010 to 2020 for simulating wireless sensor networks and large-scale IoT deployments [118]. Although these simulators were not retrieved through the keyword-based search methodology applied in this review, their relevance in earlier studies justifies their mention as noteworthy alternatives for diffusion algorithm analysis.

Figure 2.
Distribution of network simulators: This chart represents the number of publications that utilized each simulator based on the included references in this paper. Ns-3 appears as the most widely used simulator, followed by Cooja (Contiki) and OMNeT++.

4.2. Analysis of Research Question 3 and 5: Network Diffusion Algorithms and Performance Evaluation

4.2.1. Introduction

Network diffusion algorithms are essential for understanding the spread of information, data, or malware across IoT and Space IoT configurations. This section explores three major categories of network diffusion algorithms: epidemic models, cascading models, and threshold models. These categories are based on the classification provided in the paper [119]. The key performance metrics used to assess the efficiency of these algorithms are further examined.

Table 5, Table 6 and Table 7 present a structured classification of network diffusion algorithms based on their underlying models, application domains, and key performance characteristics (Abbreviation for mapping table can be shown in Table 8 and Table 9). These tables focus on the following:

Diffusion Algorithm Type—Categorizes studies using Epidemic (EM), Cascade (CM), Threshold (TM), or Other (Oth) models.
Application Domain—Highlights where these models are applied, such as epidemic modeling, social networks, and wireless sensor networks.
Network Considerations—The impact of network topology, heterogeneity, dynamicity, and multi-layered structures on diffusion performance. These aspects are discussed in detail in Section 4.3.

Table 5.
Survey of network topologies, algorithm categories, and related features (Part 1); full names corresponding to abbreviations available in Table 8 and Table 9.

n	Year	Top	Het	App	Wgt	Dep	Dyn	Lay	L	Dir	Algo	Cat
[120]	2013	RT	Hm	BC			Dyn	SL	NL		MHBC	CM
[10]	2013			SN			Both		Both	Both	Many
[121]	2013	RW	Het	SN	Both	Int	Dyn			Both	SIR-SIS-LTM	EM-TM
[122]	2013	SF-RT	Inhm	SN		Int	Dyn			Undir	SIHR	EM
[123]	2013	RT	Het	SN				ML			SIR	EM
[124]	2013		Het-Inhet	SN		Int				Undir	SIR	EM
[125]	2013	LS		SN			Both				LR	Oth
[126]	2013	LS		SN	Wgt		Dyn		L	Dir	LRT	Oth
[127]	2013										SI1I2S	EM
[128]	2014					Int	Dyn	ML			SIS-UAU	EM
[129]	2014	SW		Gbl		Int				Undir	LTM	TM
[130]	2014				Wgt	Int	Dyn	Both				Oth
[131]	2014	RT-SF	Both	PHE		Int	Dyn	ML			SIRV-SIR	EM
[132]	2014	NM			Both		Dyn			Dir	RW	Oth
[133]	2014	RT-SF	Het	Virus				ML		Undir	SI1SI2S-SIS	EM
[134]	2014	SF-SW	Hm-Inhm	SN		Int	S			Undir	SIRaRu	EM
[135]	2014	RT-SF-SW		SCM		Int	Dyn		L		MTG-SIR	EM
[136]	2014	Reg-RT-SF	Het	SN			S			Undir	SIR	EM
[137]	2014	SW	Hm	SN		Int					SICR	EM
[138]	2014			SN		Ind	Dyn				Many	CM-EM-Oth
[139]	2014	SF	Het	Health		Int	Dyn	SL		Undir	SIQRS	EM
[140]	2014		Hm	Rum							2SI2R	EM
[141]	2015		Het	WSN	Wgt		Dyn	ML			LMS	Oth
[142]	2015	RT-SF-SW	Het	SN		Int	Dyn	ML		Undir	Many	EM-CM-TM
[143]	2015	Art-Real	Both	SN		Int	Dyn				SEIR	EM
[144]	2015			SN						Dir	ICM-LTM	CM-TM
[145]	2015		Het	SN		Int		ML
[146]	2015		Het	Health	Both	Int	Dyn	ML		Undir	LACS	EM
[147]	2015		Both	SN		Int
[148]	2015			SN		Int	Dyn			Dir	COE	Oth
[149]	2015		Het	SN			Dyn				SIR	EM
[150]	2015			L-S Net				ML			2l	EM
[151]	2015	Lat		WSN			Dyn	SL			Bsd SIR	EM
[152]	2015	RT-SW	Het	Agt Sim		Int	Dyn			Undir	Agt M	Oth
[153]	2015	SF	Het	SN							ISIR	EM
[154]	2015			SN			Dyn

Table 6.
Survey of network topologies, algorithm categories, and related features (Part 2); full names corresponding to abbreviations available in Table 8 and Table 9.

n	Year	Top	Het	App	Wgt	Dep	Dyn	Lay	L	Dir	Algo	Cat
[13]	2016	Lat-Star-SW	Both	Gbl		Int	Both	Both	Both	Both	SIS-SIR	EM
[119]	2016		Het	Gbl		Int	Both					Many
[155]	2016	RT				Int				Undir	HCM-HCM *	EM
[156]	2016	Mesh		WSN			Dyn
[157]	2016	RW		SN						Dir	IC-LT	CM-TM
[158]	2016	RT	Het	Ep			Dyn	SL		Dir	Mv	Oth
[159]	2016			WSN
[160]	2016			SN						Dir	TTV	Oth
[161]	2016	SF	Both	Mobile Net		Int	S				WSIS	EM
[162]	2016		Het	SN			Dyn			Dir		Oth
[163]	2016			SN	Wgt		Dyn	SL		Dir		Oth
[164]	2016			Petro Eng		Int					MDDM	Oth
[165]	2016	RT		Sens Net			Dyn			Undir	LMS	Oth
[166]	2017	Lat	Both	Ranking		Int	Both	ML		Both	RW	Oth
[167]	2017			Health	Wgt	Int	Dyn			Both	RWR-RW	Oth
[168]	2017			SN	Wgt	Int				Dir		Oth
[15]	2017		Both	SN	Wgt	Both	Dyn	Both		Both	Many	Many
[169]	2017	Tree-SW	Het	Forensic	Wgt		Both			Dir	SIS	EM
[170]	2017	SF	Het	SN			S			Dir	SEIR	EM
[171]	2017		Hm	SN			Dyn				I2SR	EM
[172]	2017		Het	WSN			Dyn				MM	Oth
[173]	2017		Hm	Bio						Undir	SIVRS	EM
[174]	2018	SF-SW	Both	Ep		Int	Dyn	ML			SIS	EM
[175]	2018		Het	Rum	Wgt	Int	Dyn	Both	L	Both	Many	Many
[176]	2018		Het	Ep		Int		ML			SIR	EM
[177]	2018	R-D-SW-SF	Het	R-D	Wgt							Oth
[178]	2018	SF-SW	Hm	SN		Int		ML		Undir	DSIR-C-DSIR	EM
[179]	2018		Het	Ep	Wgt	Int	Dyn	SL			SIR-SIS	EM
[180]	2018			IoT							GRA-AHP	Oth
[181]	2018		Het	Bio	Wgt		Dyn	ML		Dir	rSIS	EM
[182]	2018	Tree	Het	WSN			Dyn				GIN	Oth
[183]	2018		Het	Ep	Both		Dyn			Undir	SIR-SIS	EM
[184]	2018			WSN							SEIRV	EM
[185]	2018			SN	Wgt			SL		Dir	DLATrust	Oth
[186]	2018		Het	SN							I2S2R	EM
[187]	2018		Het	Bio				ML		Dir	SI1SI2S	EM
[188]	2018	RW		SN	Wgt						LP-LPA	Oth
[189]	2019	SW-SF		SN	Wgt		Dyn			Dir		Oth
[190]	2019			Ep	Unwgt	Int		ML		Undir	SIR	EM
[191]	2019			WSN							APSA	Oth

Table 7.
Survey of network topologies, algorithm categories, and related features (Part 3); full names corresponding to abbreviations available in Table 8 and Table 9.

n	Year	Top	Het	App	Wgt	Dep	Dyn	Lay	L	Dir	Algo	Cat
[192]	2019	Large				Int					LPANNI	Oth
[193]	2019	FC		SN			S
[194]	2019		Hm	SN							AB	EM
[195]	2019	SW-SF	Het	SN			Dyn			Undir	ILSR	EM
[196]	2019			SN		Int						Oth
[197]	2020	SF-SW-RT	Hm	SC		Int	Dyn			Undir		Oth
[198]	2020			SN			Dyn			Dir		Oth
[199]	2020			SN							MOEA/D-ADACO	Oth
[200]	2020			SN							SAIR	EM
[201]	2020			Ep			Dyn	ML		Dir	Co-Evo Model	EM
[202]	2020			WSN			Dyn				Attack Game	Oth
[203]	2020		Het	SN		Ind				Dir	ICM-LTM	CM-TM
[204]	2020			Dif	Wgt		Dyn			Undir	DLMS	Oth
[205]	2021	RT-SF	Het	Ep			Dyn	ML	L	Undir	S-SIS	EM
[206]	2021	SF	Het	SN							CSRT-SIR-SEIR	EM
[207]	2021		Het	SN			Dyn			Dir	P2V	Oth
[208]	2021			Rum							SIS	EM
[209]	2021			SN	Wgt					Dir		Oth
[210]	2021		Het	SN	Wgt	Int		ML		Dir	MF-Model	Oth
[211]	2021				Unwgt				L	Undir	I M	Oth
[212]	2022			SN	Wgt					Dir		Oth
[213]	2022		Het	RP	Wgt	Int	Dyn	ML		Dir		Oth
[214]	2022			WSN								Oth
[215]	2022	RT		BC							ER	Oth
[216]	2022	SW-SF-2D Mesh		SN							Sznajd-SIR-SICR	EM
[217]	2022			SN	Wgt		Dyn			Dir		Oth
[218]	2022			WSN							Many	Many
[219]	2022	RT	Het	SN	Wgt		Dyn	SL		Undir	SAR	EM
[220]	2022	RT	Het	Ep		Int	Dyn			Both	SIR	EM
[221]	2023			SN						Dir		Oth
[222]	2023			SN	Wgt	Int	Dyn	SL			ICM-LTM	CM-TM
[223]	2023		Het	Ep		Int	Dyn	ML		Undir	SIS-SIR	EM
[224]	2023		Hm	Gbl	Wgt	Int	Both	ML		Dir		Oth
[225]	2023		Het	Ep		Int	Dyn	ML		Undir	UAU-SIS	EM
[226]	2023	SF-SW	Het	Ep		Int	Dyn	ML		Undir	UAU	EM
[97]	2024	Mesh	Het	WSN			Dyn	ML
[98]	2024	Mesh	Het	WSN		Int	Dyn	ML
[102]	2024	Mesh	Het	WSN	Wgt	Int	Dyn	ML		Undir
[1]	2024	Tree	Het	WSN	Wgt	Int	Dyn	SL		Dir
[105]	2024	Tree	Hm	WSN	Wgt	Int	Dyn	SL	NL	Dir

Table 8.
Abbreviation mapping table (part 1).

Category	Abbreviation	Full Name	Category	Abbreviation	Full Name
Top	RT	Random Topology	Algo	MHBC	Multi-Hop Broadcast
Top	RW	Random Walk	Algo	SIR	Susceptible-Infected-Recovered
Top	SF	Scale-Free	Algo	SIS	Susceptible-Infected-Susceptible
Top	SW	Small World	Algo	LTM	Linear Threshold Model
Top	Lat	Lattice	Algo	SIHR	Susceptible-Infected-Hospitalized-Recovered
Top	Mesh	Mesh Network	Algo	LR	Logistic Regression
Top		Not Applicable	Algo	LRT	Likelihood Ratio Test
Het	Hm	Homogeneous	Algo	SI1I2S	Susceptible-Infected 1-Infected 2-Susceptible
Het	Inhm	Inhomogeneous	Algo	UAU	Urban Area Understanding
Het	Het	Heterogeneous	Algo	MTG	Maximum Traffic Generation
Het	Both	Both Homogeneous and Heterogeneous	Algo	SEIR	Susceptible-Exposed-Infected-Recovered
Het		Not Applicable	Algo	SIRaRu	SIR with Rumor Spreading
App	SN	Social Networks	Algo	LMS	Least Mean Squares
App	BTC	Bitcoin Network	Algo	ICM	Independent Cascade Model
App	PHE	Public Health Emergency	Algo	LACS	Local Adaptive Clustering Service
App	Gbl	Global	Algo	COE	Co-Evolution Model
App	Virus	Virus Spread	Algo	HCM	Hierarchical Clustering Model
App	Health	Health-Related Applications	Algo	TTV	Time-Travel Visualization
App	Rum	Rumor Spreading	Algo	WSIS	Weighted SIS
App	Ep	Epidemic	Algo	MDDM	Multi-Domain Decision Making
App	Bio	Biological Network	Algo	AB	Agent-Based Model
App	WSN	Wireless Sensor Network	Algo	GIN	Graph Isomorphism Network
App	IoT	Internet of Things	Algo	rSIS	Revised SIS
App	SC	Supply Chain	Algo	2SI2R	Two-Susceptible-Infected-Recovered Model
App	Ranking	Network Ranking	Algo	DLATrust	Deep Learning Algorithm for Trust
App	Petro Eng	Petroleum Engineering	Algo	P2V	Propagation2Vec
App	Dif	Diffusion	Algo	MF-Model	Mean Field Model
Wgt	Wgt	Weighted	Algo	SAR	State Augmented Regularization
Wgt	Unwgt	Unweighted	Algo	Sznajd	Sznajd Model
Dep	Int	Interdependent	Algo	DLMS	Deep Learning Model for SIR
Dep	Ind	Independent	Algo	Co-Evo Model	Co-Evolution Model
Dep		Not Applicable	Algo	UAU	Urban Area Understanding
Dyn	Dyn	Dynamic	Algo	Many	Multiple Algorithms
Dyn	S	Static	Algo		Not Applicable
Dyn	Both	Both Dynamic and Static	Cat	CM	Cascade Model

By analyzing these tables, researchers can determine which diffusion models are most effective for different IoT and Space IoT applications. For instance, epidemic models (such as SIR and SIS) dominate large-scale networks due to their rapid spread, while threshold models (LTM) are widely used in structured and community-based IoT environments such as smart cities.

Table 9.
Abbreviation mapping table (part 2).

Category	Abbreviation	Full Name	Category	Abbreviation	Full Name
Dyn		Not Applicable	Cat	EM	Epidemic Model
Lay	SL	Single-Layer	Cat	TM	Threshold Model
Lay	ML	Multi-Layer	Cat	Oth	Other
Lay	Both	Both Single- and Multi-Layer	Cat	Many	Multiple Categories
Lay		Not Applicable
L	L	Loopy
Dir	Dir	Directed
Dir	Undir	Undirected
Dir	Both	Both Directed and Undirected

4.2.2. Epidemic Models

Epidemic models, such as the SIR (Susceptible-Infectious-Recovered) and SIS (Susceptible-Infectious-Susceptible) [223] models, are commonly used to study how information or malware spreads through IoT networks. These models treat devices as nodes, with the diffusion process mimicking how a disease spreads in a population.

Strengths:

Flexibility: These models can adapt to various network topologies like scale-free and small-world networks, which are common in IoT configurations.
Scalability: Epidemic models are effective in large-scale IoT environments, making them ideal for scenarios with many interconnected devices.

Weaknesses:

Simplification: These models often assume uniform infection and recovery rates, which may not accurately reflect the diversity of device states in the IoT.
Homogeneity: The assumption of homogeneous networks may not adequately capture the complexity and dynamic behavior of real-world IoT environments.

Performance Metrics:

Propagation Delay: Epidemic models generally introduce delays as they spread information across larger networks, especially if the network topology is complex.
Resource Utilization: Epidemic models tend to use more resources (e.g., CPU, memory, etc.) as nodes continuously reinfect each other, which can lead to higher energy consumption in resource-constrained IoT devices.

Example: The SIHR model demonstrated that in scale-free IoT networks, information spreads rapidly but can be difficult to control due to the absence of a clear spreading threshold.

4.2.3. Cascading Models

Cascading models, like the Independent Cascade (IC) model, describe the spread of information in a step-by-step process where activated nodes attempt to influence their neighbors. This method is particularly relevant to IoT networks that require hierarchical or multi-hop communication between devices.

Strengths:

Detailed Propagation: The IC model captures the sequential nature of diffusion and can track the influence of individual nodes, like routers or gateways.
Real-world Applicability: These models can represent the critical role of certain devices, such as smart hubs or gateways, that control the spread of data in IoT systems.

Weaknesses:

Computational Complexity: In large IoT configurations, the IC model becomes computationally intensive, particularly as the network scales.
Fixed Activation Probability: The assumption of a constant activation probability can oversimplify dynamic IoT networks, where conditions vary in real time.

Performance Metrics:

Network Load: Cascading models tend to generate moderate traffic by selectively activating nodes, which helps reduce network load.
Reachability: Cascading models often cover less ground compared to epidemic models but can efficiently propagate data in smaller IoT systems.

Example: The IC model has been successfully applied in data diffusion studies where central nodes (e.g., routers) play a critical role. However, the fixed activation probabilities limit its performance in dynamic IoT environments.

4.2.4. Threshold Models

Threshold models, such as the Linear Threshold Model (LTM), operate on the principle that a node becomes active when the influence from its neighbors exceeds a pre-set threshold. These models are suitable for IoT configurations that require coordinated behavior from clusters of devices, such as in smart homes or industrial IoT applications.

Strengths:

Community Dynamics: The LTM is effective at modeling diffusion in clustered networks, where devices in a particular area act together.
Efficient Activation: Once a small number of key nodes are activated, the diffusion process spreads quickly across the network.

Weaknesses:

Binary Activation: The model simplifies device states by assuming nodes are either active or inactive, which may not capture the varying states of IoT devices.
Constant Thresholds: The assumption that thresholds remain fixed over time does not account for the dynamic nature of IoT environments.

Performance Metrics:

Diffusion Reachability: The LTM can be limited in its reachability if the initial active nodes are not sufficient to trigger widespread activation.
Robustness: The LTM performs well in networks with resilient community structures but may struggle in highly dynamic environments.

Example: In IoT networks, the LTM has been used to simulate the spread of firmware updates across device clusters, showing quick diffusion once critical nodes receive updates. However, its binary nature limits flexibility in dynamic networks.

4.2.5. Other Models

In addition to epidemic, cascading, and threshold models, there are other approaches to studying network diffusion in IoT configurations, but they are less commonly applied in IoT contexts.

4.2.6. Conclusions

In IoT and Space IoT configurations, the choice of diffusion algorithm—whether it be epidemic, cascading, or threshold—impacts the performance of the network in different ways. Each model has its strengths and weaknesses depending on the network topology, size, and dynamicity. Evaluating these algorithms against key performance metrics such as propagation delay, network load, diffusion reachability, resource utilization, and robustness allows researchers to select the most suitable model for their specific IoT environment.

Epidemic models offer high reachability but at the cost of increased resource utilization. Cascading models provide more efficient propagation but face challenges in computational complexity. Threshold models are ideal for coordinated diffusion in clustered networks but may be limited by their binary activation approach.

By understanding the characteristics and performance of these algorithms, researchers can optimize network diffusion for specific IoT applications, ensuring efficient and scalable diffusion processes in diverse network configurations.

Figure 3 shows the distribution of the algorithm categories used, highlighting the dominance of epidemic models and other algorithms.

4.3. Analysis of Research Question 4: Network Architecture

4.3.1. Introduction

Network architecture plays a crucial role in shaping the performance of network diffusion algorithms, particularly in IoT and Space IoT configurations. Key factors such as the network’s structure, topology, device heterogeneity, and the dynamic nature of these environments heavily influence how information spreads throughout the network. This section explores the effects of various network topologies, heterogeneity, dynamicity, and layered network structures on the diffusion processes.

As illustrated in Figure 4, the distribution of network topologies is shown through a pie chart, with Small World and Random Topology comprising the largest segments. For a detailed analysis of network topologies, heterogeneity, dynamicity, and layered structures, refer to Table 5, Table 6 and Table 7, which provide a comprehensive overview of these critical factors. These tables categorize networks based on their topology: whether they are homogeneous or heterogeneous, dynamic or static, weighted or unweighted, and whether they are interdependent or independent. The classification helps in understanding how different topological structures influence information propagation, resilience to failures, and the efficiency of network diffusion processes in IoT and Space IoT configurations. The insights gained from these tables allow researchers to evaluate which topologies and configurations are best suited for various IoT applications, such as sensor networks, epidemic modeling, or social network-based diffusion.

4.3.2. Network Topologies and Their Impact on Diffusion

The network topology significantly influences the diffusion of information in IoT and Space IoT environments. Table 10 summarizes key topologies and their relevance to diffusion dynamics.

Table 10 highlights the direct implications of network structures on diffusion efficiency. For instance, scale-free networks facilitate rapid diffusion via hubs but are highly sensitive to targeted disruptions, which is critical for IoT infrastructures like smart grids. Similarly, mesh networks provide redundant paths, ensuring fault tolerance in industrial IoT environments but at the cost of increased network overhead.

Understanding these interactions helps in designing efficient diffusion strategies tailored to IoT and Space IoT environments, where network dynamics and communication constraints must be considered.

4.3.3. Network Heterogeneity

Heterogeneity refers to the diversity in devices’ capabilities, such as differences in processing power, storage, and communication range, which directly impacts network diffusion algorithms.

Strengths:

Weaknesses:

4.3.4. Dynamicity of Networks

Dynamicity refers to the changes in node connectivity over time, which is common in mobile IoT devices or networks where nodes frequently join, leave, or change status.

Static Networks: Topologies remain unchanged, making it simpler to implement diffusion algorithms, though they do not accurately represent most real-world IoT networks.

Dynamic Networks: Devices frequently change positions or status. Diffusion algorithms must be flexible and adaptive to handle these changes, which increases complexity but improves realism.

Example: In dynamic sensor networks, where nodes are mobile, algorithms need to adapt to topology updates, improving efficiency but complicating the diffusion process if changes occur too frequently.

4.3.5. Layered and Multi-Layered Networks

Layered networks represent different communication levels (e.g., physical and application layers), while multi-layered networks capture interactions between different layers, allowing for more complex modeling of IoT systems.

Single Layer

Diffusion occurs within a single type of connection (e.g., device-to-device communication), which simplifies modeling but may not fully represent real-world IoT networks.

Multi-Layer Networks

Consider interactions between multiple types of connections (e.g., physical and virtual communications). These networks offer more realistic simulations but require more complex algorithms and higher computational resources.

4.3.6. Other Network Characteristics

Weighted Networks

Edges between nodes have weights representing the strength or capacity of connections, making them relevant in the IoT where certain connections (e.g., high-bandwidth links) are more reliable.

Directed Networks

Communication is unidirectional, which is common in IoT systems where sensors collect data and transmit them to a central server.

5. Conclusions

This paper presents a comprehensive review of network diffusion algorithms and simulators, with a focus on their application in IoT and Space IoT systems. The study explored the performance of widely used diffusion models, including epidemic, cascading, and threshold-based algorithms, which are central to the dissemination of information, data, and even malware within these networks. Understanding their strengths and limitations is critical for designing robust, scalable, and secure communication frameworks, particularly in the demanding context of Space IoT.

Additionally, a range of network simulators, such as NS-3, Cooja, and OMNeT++, is employed to simulate and evaluate the behavior of diffusion algorithms in IoT networks. These simulators are instrumental in helping researchers understand and address the complexities of IoT networks, particularly with regard to large-scale deployments, device heterogeneity, and dynamic network topologies. This review focused on identifying key metrics for evaluating diffusion algorithms and simulators, including scalability, energy consumption, latency, and packet delivery ratio, which are essential for maintaining efficient and reliable communication across diverse IoT systems.

The findings highlight that while existing diffusion models and simulators have demonstrated effectiveness in traditional IoT scenarios, the unique characteristics of Space IoT demand more adaptable and resource-efficient solutions. Space IoT systems, with their complex topologies, constrained resources, and real-time communication needs, present distinct challenges. Future research should prioritize the development of advanced diffusion algorithms and simulation tools tailored to address these challenges, paving the way for more resilient and efficient Space IoT networks.

While this paper primarily addressed diffusion algorithms and simulators in IoT systems, future research should expand to their applications in emerging domains such as Space IoT and modular robots. Space IoT systems, comprising satellites, drones, and other extraterrestrial IoT devices, present unique challenges due to extreme topologies, constrained resources, and high latency. Similarly, modular robots, which can autonomously reconfigure for diverse tasks, introduce a dynamic and adaptive network environment that requires efficient and robust communication protocols.

Future work should include a comprehensive benchmarking study of diffusion algorithms tailored for both modular robots and drones. These systems require efficient communication protocols to ensure coordinated operations, real-time decision making, and seamless interaction in unpredictable environments. Testing these algorithms using the network simulators discussed in this paper like Ns-3 will provide valuable insights into modeling the communication behaviors of both modular robots and drones.

In addition to benchmarking, a key area of interest is the ability of these systems to adapt based on changes in the environment or potential security threats, such as hacks. In highly dynamic settings, where the topology or external conditions shift frequently, modular robots and drones must be capable of adjusting their communication and coordination strategies to maintain optimal performance. Research into how diffusion algorithms can evolve in real-time to handle these challenges will be crucial to ensuring the robustness of future IoT applications.

Collaboration between Space IoT and terrestrial IoT systems, including modular robots, offers another promising research direction. For instance, studying the interaction between satellites, drones, and modular robotic units in integrated networks could provide insights into scalability, interoperability, and resilience. Comparative analysis of diffusion algorithm performance across these domains will help identify common challenges and specialized solutions.

By pursuing these directions, future research can contribute to significant advancements in areas such as satellite-based communication, interplanetary exploration, disaster management, and autonomous industrial operations. This work will not only enhance the understanding of diffusion algorithms but also pave the way for innovative solutions in the rapidly evolving IoT landscape.

Source link

Charbel Mattar www.mdpi.com

1. Introduction

2. Methodology

2.1. Research Question Formulation

2.2. Search Strategy

2.2.1. Identification of Databases

2.2.2. Keywords and Phrases

2.2.3. Combinations Using Boolean Operators

2.2.4. Pilot Searches

2.2.5. Papers Using API

2.3. Inclusion and Exclusion Criteria

2.3.1. Objective

2.3.2. Inclusion Criteria

2.3.3. Exclusion Criteria

2.4. Study Selection

2.4.1. Step 1: Finding and Initial Check

2.4.2. Step 2: Applying Criteria and Checking Titles/Abstracts

2.4.3. Details of Criteria and Screening

2.4.4. Step 3: Reading Full Papers and Checking Quality

2.4.5. Final Set of Papers for Analysis

3. Survey Comparison

4. Analysis

4.1. Analysis of Research Question 2 and Research Question 6: Network Simulators in IoT and Space IoT Configurations and Their Performance Evaluation

4.1.1. NS-3

Overview

Features

Limitations

Notable Studies Using NS-3

Performance Metrics

4.1.2. Cooja (With Contiki OS)

Overview

Features

Requirements Addressed

Limitations

Notable Studies Using Cooja

Performance Metrics

4.1.3. OMNeT++

Overview

Features

Limitations

Notable Studies Using OMNeT++

Performance Metrics

4.1.4. Comparative Summary

4.1.5. Conclusions

4.2. Analysis of Research Question 3 and 5: Network Diffusion Algorithms and Performance Evaluation

4.2.1. Introduction

4.2.2. Epidemic Models

4.2.3. Cascading Models

4.2.4. Threshold Models

4.2.5. Other Models

4.2.6. Conclusions

4.3. Analysis of Research Question 4: Network Architecture

4.3.1. Introduction

4.3.2. Network Topologies and Their Impact on Diffusion

4.3.3. Network Heterogeneity

4.3.4. Dynamicity of Networks

4.3.5. Layered and Multi-Layered Networks

Single Layer

Multi-Layer Networks

4.3.6. Other Network Characteristics

Weighted Networks

Directed Networks

5. Conclusions

Related Posts

4 Fab Uses for Leftover Paint

Scanning Electron Microscopy Reveals New Ultrastructural Features in Metacercariae of Clinostomum cutaneum (Digenea: Clinostomidae) Infecting Oreochromis niloticus (Actinopterygii: Cichlidae) in Kenya

Robust Portfolio Selection Under Model Ambiguity Using Deep Learning

Greenberg