Handover Decisions for Ultra-Dense Networks in Smart Cities: A Survey

We sincerely thank Reviewer 1 for their constructive comments and valuable suggestions, which have greatly helped us improve the quality, clarity, and contribution of our survey paper. We have carefully revised the manuscript in response to all comments. The detailed responses and actions taken are provided below, with corresponding changes highlighted in the revised manuscript.

Comment 1:
The organizational structure should be made more clear, e.g., add an organizational figure.

Response 1:
Thank you for the insightful suggestion. To enhance clarity, we have added a new figure titled "Figure 1. Organizational Structure of the Survey" (Line 180). This figure outlines the logical structure of the manuscript, including key sections: Background, Related Works, Comparative Summary, Research Challenges, and Future Directions.

“Figure 1 presents the structural roadmap of the survey, illustrating the flow from foundational concepts to in-depth literature synthesis and emerging challenges in HO management for 6G networks.”

Comment 2:
The review of each reference can be made more useful throughout the whole paper. For example, some analysis, advantages/benefits, and limitations/shortcomings can be added. Moreover, the connection/relationship with other references can be explained.

Response 2:
Thank you for the helpful suggestion. To make the literature review more insightful, we updated Table 1 to clearly show the main strengths, limitations, and methods used in each study. The revised format makes it easier for readers to see how the different works compare and connect in terms of their approach and contribution. We believe this update makes the overall analysis more readable and meaningful (Line 654).
Also, Section 3 Related Works was divided into four subsections for more clarity and thematic coherence.

Comment 3:
Most importantly, “lessons learned” should be included after the literature review of each main part to give the summary and some insights for readers.

Response 3:
Thank you for this valuable recommendation. We have now added 3.4 Summary and Research Gaps in Handover Methods (Line 645). This section provides a synthesized summary of insights from the reviewed works, identifies common challenges, and highlights promising trends such as hybrid intelligence and context-aware HO strategies.

Line 656 “To provide a clearer understanding of the current landscape in handover (HO) decision-making for ultra-dense 6G networks, the surveyed literature has been categorized into three main approaches: AI-based, fuzzy logic-based, and hybrid frameworks. This classification highlights methodological diversity while helping to identify critical limitations that restrict the practical deployment of many existing models.

AI-Based Approaches utilize machine learning (ML), deep learning (DL), and reinforcement learning (RL) to enable adaptive and predictive HO decisions. These methods are effective in modeling complex mobility behavior and enhancing QoS. However, their applicability in real-time systems is constrained by limited interpretability, the need for large training datasets, and significant computational overhead, which is particularly problematic in energy-sensitive or latency-critical environments.

Fuzzy Logic-Based Approaches offer interpretable, rule-based reasoning that handles uncertainty well in HO scenarios. They are lightweight and transparent, making them suitable for constrained devices. Nonetheless, such models lack adaptability in highly dynamic settings and often require manual tuning, limiting their scalability.

Hybrid Approaches, which integrate AI techniques with fuzzy logic systems, aim to balance learning capability with interpretability. These models show promise in adapting to real-time variations while maintaining clear decision logic. However, they introduce increased complexity, integration overhead, and a lack of validation in real-world 6G deployments.

Despite advances in handover efficiency across all categories, many models still lack energy-awareness, which is critical for ultra-dense smart city scenarios involving thousands of mobile and IoT devices. Without considering energy constraints, these algorithms may lead to unnecessary signaling and battery depletion. Additionally, secure and real-time deployment capabilities remain underexplored. Most models are tested only in simulation environments and rarely account for latency, dynamic interference, or authentication mechanisms during HO.

To address these challenges, future work must prioritize the development of lightweight, secure, and context-aware HO algorithms. This includes integrating federated learning, energy-aware optimization, and blockchain-based mechanisms to enable privacy-preserving, tamper-proof, and interoperable handover solutions for 6G networks.

Comment 4:
In the literature review, comparison and discussion, more informative figures can be added to improve the overall readability.

Response 4:
We appreciate the reviewer’s suggestion regarding the addition of more informative figures. While we agree that visual representation is valuable for enhancing readability, we have carefully structured the manuscript to include key figures and tables that already serve to illustrate the comparison of existing handover techniques (Table 1). These visual elements were designed to synthesize complex information in a concise and accessible format.

Comment 5:
More future directions should be explored, such as the integration with large models. Besides, each aspects can be made more specific. More related surveys can be added and cited such as "(2025) Large Model Based Agents: State-of-the-Art, Cooperation Paradigms, Security and Privacy, and Future Trends" in IEEE Communications Surveys & Tutorials, DOI: 10.1109/COMST.2025.3576176. 

Response 5:
Thank you for this excellent suggestion. We have added in conclusion this paragraph (Line 782)

In parallel, emerging large-scale models such as Large Language Models (LLMs) and multi-agent systems hold promises for intelligent mobility management [62]. Their ability to generalize across diverse mobility contexts and support autonomous decision-making in heterogeneous environments opens new directions for adaptive HO policies. Federated or cooperative learning architectures may also allow for privacy-preserving optimization without centralized data reliance.

Comments 1:  From Table 1:  i) Both of the classes of fuzzy logic and ANN by Sanusi et al. (2020) [41] and Aibinu et al. (2017) [43] are same, ii) References are not enough. 

Response 1: Thank you for pointing this out.

i)                We acknowledge that both Sanusi et al. (2020) [41] and Aibinu et al. (2017) [43] utilize similar hybrid fuzzy logic and artificial neural network (ANN) architectures. However, the two studies differ in terms of their application context and system configuration. Sanusi et al. focus on mobility prediction in heterogeneous networks using a neuro-fuzzy model optimized for urban scenarios, while Aibinu et al. apply a fuzzy-ANN approach to optimize handover decisions under varying velocity and load conditions.

ii)  We agree with this comment. Therefore, we have added 6 more references.

Comments 2: Significance of this review is not clearly described. How this survey is different than the existing work? Evolution of the network generation as well as their application maybe described with some diagram.

Response 2: Thank you for this important comment. In the revised manuscript, we have clarified the unique contribution of our survey in the updated Introduction and Motivation sections. Specifically, we emphasize that this work provides a focused review of handover decision-making strategies in ultra-dense network environments, with particular attention to AI, fuzzy logic, and hybrid approaches relevant to 6G smart city deployments. Unlike previous surveys, we examine not only the technical mechanisms but also practical limitations such as energy efficiency, real-time deployment, and security, which are often overlooked. For the convenience of the reader we added in line 164 [The significance of this survey lies in its focus on adaptive and intelligent handover strategies tailored for ultra-dense, heterogeneous 6G networks in smart city environments. Unlike prior reviews that broadly cover mobility management or traditional handover algorithms, this work highlights emerging AI-fuzzy and hybrid mechanisms, while also addressing overlooked aspects such as energy efficiency, security, and deployment scalability.]

Comments 3:  This survey doesn't discuss the critical gaps by systematically categorizing state-of-the-art HO approaches into AI-based, fuzzy logic based, and hybrid frameworks.

Response 3: Thank you for highlighting this important point. We added in Line 656

“[To provide a clearer understanding of the current landscape in handover decision-making for ultra-dense 6G networks, the surveyed literature has been categorized into three main approaches: AI-based, fuzzy logic-based, and hybrid frameworks. This classification highlights methodological diversity while helping to identify critical limitations that restrict the practical deployment of many existing models.

AI-Based Approaches utilize machine learning, deep learning, and reinforcement learning to enable adaptive and predictive HO decisions. These methods are effective in modeling complex mobility behavior and enhancing QoS. However, their applicability in real-time systems is constrained by limited interpretability, the need for large training datasets, and significant computational overhead, which is particularly problematic in energy-sensitive or latency-critical environments.

Fuzzy Logic-Based Approaches offer interpretable, rule-based reasoning that handles uncertainty well in HO scenarios. They are lightweight and transparent, making them suitable for constrained devices. Nonetheless, such models lack adaptability in highly dynamic settings and often require manual tuning, limiting their scalability.

Hybrid Approaches, which integrate AI techniques with fuzzy logic systems, aim to balance learning capability with interpretability. These models show promise in adapting to real-time variations while maintaining clear decision logic. However, they introduce increased complexity, integration overhead, and a lack of validation in real-world 6G deployments.

Despite advances in handover efficiency across all categories, many models still lack energy-awareness, which is critical for ultra-dense smart city scenarios involving thousands of mobile and IoT devices. Without considering energy constraints, these algorithms may lead to unnecessary signaling and battery depletion. Additionally, secure and real-time deployment capabilities remain underexplored. Most models are tested only in simulation environments and rarely account for latency, dynamic interference, or authentication mechanisms during HO.

To address these challenges, future work must prioritize the development of lightweight, secure, and context-aware HO algorithms. This includes integrating federated learning, energy-aware optimization, and blockchain-based mechanisms to enable privacy-preserving, tamper-proof, and interoperable handover solutions for 6G networks.]”

Comments 4: The objective as well as the idea is not clear becouse there is no flow of the litrature. Authors may add more sections. Each section should transition smoothly, helping readers to understand the progression of ideas.

Response 4 Thank you for pointing this out.

We added in line 138

Also this paper aims to bridge that gap by systematically reviewing intelligent handover techniques and identifying key limitations and future directions.

Also Section 3 Related Works was divided into four subsections for more clarity and thematic coherence.

Comments 5:  add some more figures and table  analyze the literature critically, to identify trends, gaps, and inconsistencies in the existing works.. 

Response 5: Thank you for the suggestion. While we appreciate the recommendation to include additional figures, we believe that the current set of illustrations and the newly revised comparative table sufficiently capture the key findings and methodological distinctions among the reviewed works. To strengthen the critical analysis, we have updated Table 1 to present a more structured comparison of the literature, highlighting core methods, benefits, limitations, and technology categories. Furthermore, Section 3 has been reorganized into four thematic subsections - covering traditional, AI-based, hybrid approaches and Summary of Works in Handover Methods —which improves clarity and facilitates a more focused identification of trends, gaps, and inconsistencies in the existing research. We hope these revisions meet the intent of your comment while preserving the clarity and balance of the manuscript.

Comments 6: Litrature is also not enough

Response 6: Agree. We have added 5 more references.

Comments 1:  This is a survey paper, the authors claim this work provides a systematic and thorough survey, however, what’s the methodology for selecting the papers that were reviewed. 

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, I/we have added 1 paragraph line 307

“[The literature was collected through a systematic search across reputable databases including IEEE Xplore, ScienceDirect, SpringerLink, and MDPI, using keywords such as “handover decision”, “6G mobility”. We included publications from 2015 to 2025 that focus on handover strategies in 5G/6G networks, particularly in ultra-dense or smart city scenarios, while excluding works unrelated to mobility, lacking technical contributions, or centered solely on legacy technologies.]”

Comments 2: Some statements are not precise, e.g., line 43-45, “In contrast to its previous incarnations, 6G networks accept newer technologies such as artificial intelligence (AI), machine learning (ML), edge computing, and terahertz (THz) communications for improving the efficiency, reliability, and experience of networks.” 5G network should also accept AI & ML

Response 2: Thanks. We acknowledge the imprecise wording and have revised the sentence for clarity. The updated version (line 44) now reads: “[While 5G networks have already begun incorporating AI and ML for network optimization, 6G is expected to deepen this integration by embedding these technologies as fundamental building blocks for autonomous, real-time, and context-aware network control.]”

Comments 3:  Some sentences are not well referenced, e.g., Line 72,  “while there are several review papers on handover mechanisms.

Response 3: Thank you for this observation. We have added appropriate references to support this statement in line 73.

Comments 4: Table 1 needs to be reorganised, too many text, not concise, and the organisation of the contents lacks system.

Response 4: We appreciate this constructive feedback. Table 1 has been reformatted to improve its clarity. We believe this update makes the overall analysis more readable and meaningful (Line 654).

Also, Section 3 Related Works was divided into four subsections for more clarity and thematic coherence.