DB-SAHO network-based distributed mobility management: enhancing handover performance in mobile networks

1 Introduction

The current and future 5G and 6G networks have seamless mobility support capabilities remained a great challenge (Khan et al., 2024; Islam et al., 2025; Sulaiman and Al-Raweshidy, 2025; Topazal et al., 2025). As these networks scale up to link more devices and power intensive applications like augmented reality, industrial automation and autonomous transportation, so does the demand for robust and reliable mobility management (Ojaroudi Parchin, 2025). Distributed mobility management (DMM) is an improvement over the traditional centralized systems. In older networks, user situations are managed when devices move from one area to another by a central mobility anchor. While such a setup is easy, it can lead to bottlenecks, to some limits of scalability and to single points of failure, which does not work well in crowded or high-reliability settings (Adam, 2025). DMM distributes the mobility anchor’s functionality across multiple nodes of the network. This allows to balance the load of data and signaling within the network, reduce the bottleneck, and makes the network more resilient. Better scalability also makes user mobility easier to take care of, as the number of devices and service types increase (Kalamati et al., 2025).

Even with these advantages, network-based distributed mobility management still side provocation, specifically with transition. In NB-DMM the mobility routers and a central database have to work together. Every time a node moves to a new area, signalling messages are used to update the new location of the node and other session details. Frequent signaling contributes to costs and creates more difficulty in managing. Handover events can also result in data loss/delays, which can result in high latency and potential further data loss. These problems are particularly detrimental to applications that require real-time stable connections. These problems directly impact service quality to the users, as well as, the performance of important applications. Because of this, handover mechanisms in NB-DMM systems require more improvement. Researchers are now working to reduce signaling overhead and latency, and ensure packets are delivered reliably to meet the strict demands of future mobile networks.

Existing improvements to the NB-DMM (Network-based Distributed Mobility Management) typically address these problems one at a time. Methods of abbreviating signaling, such as data aggregation (combining multiple data into a single one) or signaling compression (making messages smaller), reduce overhead but cannot assure packet reliability. Alternatively, buffering (temporarily storing data) can eliminate packet loss but adds processing delay. It is unable to reduce control signaling. As a result, current solutions lack a comprehensive strategy that can simultaneously reduce signaling costs, prevent packet loss, and ensure low latency (short delays). This kind of gap leaves space for an optimized handover strategy (Elbatal et al., 2025; Wong et al., 2025).

To fill this research gap, we propose a DB-SAHO (Dual Buffering and Signaling Aggregated Handover Optimization) as a new NB-DMM (Network-Based Distributed Mobility Management) framework. DB-SAHO combines multi-message signaling aggregation with a dual-buffering scheme. The signaling aggregation mechanism consolidates several handover-related messages into a single, optimized exchange. This process significantly reduces the signaling rate and control-plane waits. At each Mobile Access and Anchor Router (MAAR), dual buffering enables a Forward Data Buffer (FDB) to store packets from the source MAAR. A New Data Buffer (NDB) stores packets arriving at the target MAAR. These buffers hold data until the Mobile Node (MN) can receive it. This joint design ensures lossless, in-order packet delivery and reduces handover latency.

This work’s originality lies in combining multi-message signaling aggregation (grouping handover-related control messages) and double buffering (using two packet storage areas) within a PMIPv6-based DMM solution (Proxy Mobile IPv6, an IP mobility protocol).

Current NB-DMM improvements do not significantly change the flow of handover execution, as data forwarding still depends on signaling completion. As a result, packet loss and reordering remain closely related to handover latency. This limitation is addressed in this work by modifying the handover execution flow and dependency structure to retain data-plane continuity independently during handover.

This paper has the following contributions:

• The paper proposes a dual-buffer and signalling-aggregating handover algorithm, DBSAHO, for NB-DMM, which alters the flow of execution and the dependency structure of handovers.
• The paper also introduces an analytical model that focuses on the proposed dependency change by examining packet loss and reordering during handover.
• A simulation-based study shows that DB-SAHO significantly reduces signaling overhead. It also minimizes packet loss and improves latency compared to the baseline NB-DMM.

The rest of this paper is structured in the following way. Section 2 is a literature survey of mobility management and NB-DMM. Section 3 explains the methodology, system model, and the suggested framework, and gives the performance metrics. In Section 4, the results and analysis are discussed. Lastly, Section 6 presents the conclusion and gives the research directions.

2 Literature review

Research on handover performance has become increasingly vital with the advent of 5G mobile networks. This portion examines the existing handover management strategies of NB-DMM and rapid handover schemes in PMIPv6. The discussion presents important contributions, limitations, and research trends that are important to the proposed work.

Lin et al. (2011) Suggest an HF-PMIPv6 handover approach in the study, and this approach uses a decision table to identify the movement type. When the mobility type is predictive, HFPMIPv6 creates a link that connects the PMAG & N-MAG, similar to F-PMIPv6. If the movement style is reactive, HF-PMIPv6 executes handoffs immediately before establishing a tunnel. The findings reveal that HFPMIPv6 outperforms F-PMIPv6 in terms of handoff delay and delivery cost. Chuang and Lee (2011). suggested a quick handoff strategy for PMIPv6 known as FH-PMIPv6. This strategy makes use of the buffering strategy, implementing the two activities of authorization and registration simultaneously in order to reduce handoff time and provides with a two-buffer system in order to prevent messages getting out of sequence. Kim et al. (2012) proposed an improved version of the fast handover method called ePFMIPv6 for vehicular networks by establishing a channel between the SMAG and the next candidate MAG in advance. Once the notice of the vehicle handover is received, the SMAG re-distributes packets to the next MAG right away. The simulation results show that the proposed ePFMIPv6 is superior to PFMIPv6 in terms of handover delay, data loss and signaling overhead.

Park et al. (2014) proposed an efficient buffering approach to address the data loss during handover in PMIPv6. The results indicate that the packet loss is decreased by using PMIPv6 in most cases; especially, when the handover time is longer and the message throughput is higher. With the improved buffering technique, the LMA is able to forward messages at faster rates, and store fewer messages in the buffer. The authors (Giust et al., 2014) evaluated a network-based DMM technique analytically and experimentally in their research. They created and compared mathematical formulas for signaling charge, packet transmission, handoff delay, and data loss with PMIPv6. The outcomes demonstrate that, in certain instances, using a DMM technique offers resource reduction in networks through a decreased packet transmission cost. Additionally, there are still some circumstances where DMM results in more significant expenses such as when a mobile node has persistent apps, a limited cell residence period. In these circumstances, choosing a centralized mobility management system would seem to be better. Nguyen and Bonnet (2015) presented an H-DMM architecture, and this approach enables the mobile node to select a suitable prefix from those allocated by the current Mobile Access Router (MAR) and allows the primary entity to initiate a new flow. As a result, it is possible to maintain threshold levels of active prefixes, especially in scenarios with high mobility or persistent traffic flow. The numerical outcomes demonstrated that HDMM obtains the benefits of DMM compared to PMIPv6.

Balfaqih et al. (2016) introduced NB-DMM method which is based on FPMIPv6. The proposed approach gives PFMIPv6 procedure functionality with particular modifications of crucial keywords, message forms, and functionality. The results infer that the proposed technique reduces handover delay and data loss by a significant margin as compared to the existing IETF approach. Balfaqih et al. (2017a) proposed the usage of the handover initiation process and the row 802.21-MIH facilities to improve the handover of DMM in through the process named as the row IEEE 802.21 Assisted Method. The proposed solution reduces the handover delay by 72.3%, and the signaling cost exceeds the existing NB-DMM. Balfaqih et al. (2017b) presented the fast handover method, referred to as the FDMM approach. FDMM reduces the handover delay by detecting the mobility and performing the binding registration procedure at an early stage. The numerical results obtained indicate that the FDMM approach is superior compared to the NB-DMM.

Huang et al. (2018) suggested the group-based strategy which is known as BMH-DMM technique. The proposed BMH-DMM methodology can be used to save about 30% of the bandwidth of the transmission of handoff signaling information. Additionally, the proposed BMH-DMM system shows less network utilization, handoff time, data loss, and handoff blocking frequency than the individual handover technique. Sharma et al. (2018) proposed a blockchain-based approach to DMM that decentralizes mobility anchoring, reduces the need for trust management, but the overheads of its signaling and processing make it inapplicable to latency-sensitive handovers. Siddiqui et al. (2022) investigate the 5G mobility challenges and suggest that Distributed Mobility Management (DMM) could enhance scalability and reduce latency. However, it points out that the proposed solutions must undergo more testing in the real world in order to prove their effectiveness.

Ullah et al. (2023) conducted an in-depth review of handover and mobility management for 5G heterogeneous networks. The author focuses on essential issues, including growing data traffic, a broader range of devices, and delay constraints in real-time applications. Mane and Khairnar (2023) highlighted the importance of the use of buffering techniques which are important in the maintenance of data integrity. Das et al. (2025) authors have used a Markov model to compare several DMM techniques for 5G-based ITS. The study helps in selecting the most effective mobility management strategies. Table 1 shows the literature relevance to our proposal.

TABLE 1. Relevant literature of this study.

Figure 1 shows how mobility management approaches have changed over time and highlights research gaps found in earlier studies.

Although recent DMM and SDN-based mobility solutions have been proposed, this study considers baseline NB-DMM to enable a consistent, architecture-aligned comparison. The suggested DB-SAHO scheme is intended to improve the NB-DMM setup.

2.1 Literature gap and scope of the proposed

Despite the impressive advancements in network-based distributed mobility management (NB-DMM), the current literature mainly examines handover latency, packet loss, or signaling cost in isolation, either in idealized or more realistic mobility models. Although these distributed methods and recent optimization methods have reduced the need for centralized bodies, packet losses during handover and signalling overhead remain non-negligible, and out-of-order packet delivery has not been explicitly addressed in most previous papers. Further, the overall effects of signaling optimization processes and buffering processes in the same NB-DMM model have not been adequately investigated. Based on these examinations, this paper aims to explore an integrated mobility support strategy that can simultaneously reduce handover latency, mitigate packet losses, improve signaling efficiency, and address packet ordering behavior in the presence of realistic mobility characteristics. The offered DB-SAHO framework is thus placed as an integrated improvement of the NB-DMM paradigm, which will fill the specified gaps, but will not change the underlying structure of NB-DMM.

3 Methodology

The section describes the system model, mobility and traffic assumptions, workflow, and the proposed DB-SAHO algorithm with its pseudocode. It also specifies the performance measures that are applied.

3.1 System model

The proposed approach is applied in a PMIPv6-capable NB-DMM system, which comprises four main functional components, as shown in Figure 2.

3.1.1 Centralized Mobility Database (CMD)

The CMD is a logically centralized component that houses the global binding cache and synchronizes mobility management for all Mobile Access and Anchor Routers (MAARs). It facilitates signaling aggregation, minimizes redundant updates, and offers routing information during handovers.

3.1.2 Mobile Access and Anchor Routers (MAARs)

Each MAAR serves as both an access router and an anchor for a Mobile Node (MN). Two logical buffers are used at every MAAR: the Forward Data Buffer (FDB), which stores packets forwarded from the previous M000AAR temporarily. New Data Buffer (NDB), Buffers new packets received from the Correspondent Node (CN) until the MN is attached to the new MAAR.

3.1.3 Mobile node (MN)

A user device (e.g., smartphone, IoT node) traversing MAARs. It is not aware of mobility management operations because the process is network-based.

3.1.4 Correspondent node (CN)

A communication peer that shares data with the MN.

Finally, the system model requires the assumption of operation of more than one MAAR in a single CMD domain. Handover is done when the MN changes the MAAR coverage area. Handover is done, and ongoing data with charges are maintained.

3.2 Execution flow and dependency structure

DBSAHO adheres to the usual NB-DMM signaling messages and network devices. However, it changes how handover is executed by relaxing the strict coupling between control-plane signaling and data-plane packet processing. With traditional NB-DMM, packet loss and reordering are very sensitive to handover latency. This occurs because data forwarding requires signaling to complete before packets are forwarded. DBSAHO mitigates this shortcoming by allowing data-plane operations to continue throughout the handover. It stimulates dual buffering at the target MAARs and serving MAARs through ongoing signaling aggregation. Packets meant for the MN are held in the Forward Data Buffer of the serving MAAR. Packet traffic arriving at the time of transition is placed in the New Data Buffer of the target MAAR. This approach ensures packets are continuously received, even when signaling is delayed. Meanwhile, the target MAAR configures routing and buffer behavior and completes Layer-2 attachment and IP configuration for the mobile node. It does so without stopping data buffering. Signal Completion is needed before the buffers synchronize. Once this confirmation is received, packets in the forward and new data buffers are released sequentially, and regular data forwarding resumes. DBSAHO modifies the handover dependency structure without needing additional protocol messages. It allows buffer activation to be independent of signaling completion, while limiting the synchronization point to the release of packets. As a result, signaling latency is reduced during handover. This facilitates limited packet loss and a deterministic packet sequence, as shown in Figure 3.

3.3 Proposed DB-SAHO workflow

The dual-buffering with signaling-aggregated Handover Optimization (DB-SAHO) scheme is used to improve the baseline NB-DMM through combined multi-message signaling aggregation and dual-buffering technique. The proposed workflow is intended to reduce the signaling costs, to prevent packet loss, and packet reordering during mobility transitions. The whole procedure is illustrated in Figure 3 as a type of ladder and explained below.

The process starts with handover detection, i.e., the functions of the serving MAAR (e.g., MAAR4) are to monitor the MN’s link quality and movement patterns. Upon link layer indications such as RSSI drop or Layer-2 reports of mobility, MAAR4 expects that MN is approaching to another MAAR say MAAR5. This forewarning, rather than waiting for the network, assists the process to begin from forehand, therefore, avoiding the reactive handover interruptions.

After detection, there is a handover initiation phase which is an essential novelty of DB-SAHO over the baseline NB-DMM. In the traditional NB-DMM procedure, the serving MAAR stimulates a sequence of separate signalling exchanges with the Centralized Mobility Database (CMD) - i.e., binding updates, acknowledgments, context transfer, and routing updates. Multiple round-trips cause signaling overhead unnecessary while adding the handover latency. By contrast, DB-SAHO was able to bundle this signaling up into one optimized transaction. In particular MAAR4 creates a Handover Request Message (HRM) and sends the HRM to the CMD. The HRM completes the request and packages the binding update request, together with the context information of MN, i.e., QoS state and session information along with forwarding instructions for the Buffered packets and the target MAAR identifier obtained via the link layer triggers. Upon receiving the HRM the CMD performs signaling aggregation and generates an Aggregated Update Message (AUM) encapsulating all the control responses required in one. The AUM carries the binding acknowledgement which confirms new point of attachment of the MN, new routing and forwarding information, session continuity parameters such as mobility state and IP prefix retention and synchronization commands to synchronize the forward and new data buffers. As this AUM, the CMD simultaneously sends it towards MAAR4 and MAAR5. Through this mechanism, the multiple signalling steps of the NB-DMM (one after the other) are substituted by one optimised transaction, which reduces the cost of signalling, latency and keeps both MAARs in synchronization for buffer activation.

The buffer activation phase is triggered at the same time as the signaling phase. When serving MAAR4, the Forward Data Buffer (FDB) is operational to buffer all the in-transit packets bound to the MN so that they are not lost in the attachment process. At destination MAAR5, the New Data Buffer (NDB) is operational, to buffer packets in its memory received from the corresponding node while MN is still completing its attachment process. This dual-buffering strategy ensures that the old and new packets are both kept which helps to maintain the continuity of the packet sequence.

Next the CMD binding update is carried on. The CMD makes an entry in the global binding cache to name the new point of attachment of the MN at MAAR5. In contrast to the baseline NB-DMM, which requires multiple messages in order to execute the above updates, DB-SAHO takes advantage of the already created AUM, which contains all the necessary binding, routing and session continuity information. This allows MAAR5 to preprogram its routing tables as well as buffer mechanisms on the fly with no further signaling latency.

The second phase, delivery and forwarding, is the process in which packets that are held in the FDB from MAAR4 to MAAR5 are passed on. The packets were sent at the end of the active sessions prior to the handover. Simultaneously, packets that are buffered inside the NDB of MAAR5 are delivered only in case of successful attachment of the MN to the network. By engaging in synchronization of handling the FDB and NDB, the packets are sent out in the appropriate order; hence, reordering is avoided.

Finally, the handover becomes effective when the MN finishes the attachment to the MAAR5. At this point both the FDB and the NDB buffered packets are transported without any conditions and result in 0 packet loss and elimination of the out-of-order property. The CMD, MAARs, and MN then return to steady state operations with full session continuity.

Even though the proposed workflow explains the steps of a signaling process in detail, it is necessary to explain how signaling aggregation maintains the functionality of individual signaling messages. Signaling aggregation under the proposed methodology does not eliminate any signaling messages. Instead, it aggregates the information that has been encoded by several signaling messages into one aggregated message. Every field of the aggregated message is associated with a certain signaling capability, e.g., handover initiation, binding update, session context transfer, and forwarding instruction. These fields are then extracted and processed in order at the receiving MAAR to ensure that the same logical functions performed in standard signaling are maintained, with fewer signaling exchanges.

The DB-SAHO process has a number of advantages over the baseline NB-DMM. It saves signaling costs due to the aggregation of signaling messages in an optimized transaction. It also reduces the handover latency by processing signaling messages in a batched mode of operation. With dual buffering, the packet loss is minimized, and the packets are sent in an ordered manner. In addition, playing in sync with the CMD guarantees scalability and efficiency, even in a situation with many MAARs involved. Overall, these features make DB-SAHO a trustworthy approach to seamless mobility manageability in the next-generation of distributed networks. The handover procedure is described in Algorithm 1. The pseudocode section below explains how the signaling and buffering work.

Algorithm 1 Proposed DB-SAHO.

3.3.1 Pseudocode

3.3.1.1 Signaling aggregation

• // At Serving MAAR (MAAR4)
• on Handover_Detected (MN, TMAAR):
• HRM = [ MN ID, SMAAR, TMAAR, Context Info]
• sendtoCMD(HRM)
• // At CMD
• on Receive_HRM(HRM):
• AuM = [bindingupdate, routeinfo, session context, buffersync]
• sendto (TMAAR, AuM)
• sendto (SMAAR, AuM)

Typically, NB-DMM includes the use of several messages (Binding Update, Acknowledgment, Context Transfer Routing Update). In DB-SAHO, a Handover Request Message (HRM) is sent from the serving MAAR to the CMD. The CMD replies with one Aggregated update Message (AuM) that contains all the required control info to be sent to both MAARs. Doing this eliminates signaling cost and reduces latency.

3.3.1.2 Dual buffering

• // At Serving MAAR (MAAR4)
• on Activate_FDB(MN):
• store all in-transit packets in the Forward Data Buffer (FDB)
• on Receive_AUM(AUM):
• forward all packets from FDB to TMAAR
• // At Target MAAR (MAAR5)
• on Receive_AUM(AUM):
• activate New Data Buffer (NDB)
• store new packets from CN into NDB
• on MN Attach (MN):
• deliver packets from FDB (old) first
• Then deliver packets from NDB (new)
• ensure packets are in the correct sequence

The mechanisms employ two buffers. The Forward Data Buffer (FDB) at the serving MAAR stores packets while in transit so that they do not get lost. The New Data Buffer (NDB) at the target MAAR stores packets coming in from the CN during the handover. When the MN attaches to the target MAAR (e.g., MAAR5), packets are released from the FDB first, followed by packets from the NDB, ensuring no packets are lost, and nothing arrives out of order.

3.4 Mobility and traffic model

A linear transfer model is employed, wherein the mobile node traverses a straight route. This makes it easier for us to observe handover behavior and run the same test repeatedly. UDP flow with a fixed bit rate is used in this study to concentrate on mobility-induced performance behaviors.

3.5 Performance metrics

The analytical expressions show idealized behavior during the handover interval to highlight the effect of the proposed mechanism. We use simulations to evaluate how realistic network impairments affect the results.

3.5.1 Signaling cost

The overall overhead (in terms of control messages sent and received) to finish a handover. It is a function of the number of hops, the size of the signalling messages, and the frequency of mobility events. The following Equations 1, 2 represent the signaling for NB-DMM and proposed equations.

NB-DMM

S i g C o s t N B D M M = ∑ i = 0 N α . H o p M A A R − C M D + β . H o p M A A R − M A A R (1)

Proposed

S i g C o s t P r o p o s e d = γ . S i g C o s t N B D M M , 0 < γ < 1 (2)

Where:

α , β − The cost for each message between the groups.

H o p M A A R − C M D − hop counts for sending signals.

N-number of handovers.

γ-aggregation factor (the amount of reduction that happens when multiple messages are combined).

3.5.2 Handover latency

The time needed for a mobile node to disconnect from the old MAAR and then successfully reconnect to the new MAAR to receive data. Equations 3, 4 represent the handover latency for NB-DMM and the proposed.

NB-DMM

H O N B D M M = T D e t e c t + T M A A R − C M D + T S e t u p p a t h (3)

Proposed

H O P r o p o s e d = T D e t e c t + T M A A R − C M D * γ + T f o r w a r d d a t a b u f f e r (4)

Where:

T D e t e c t-Detection delay.

T M A A R − C M D − Setting a time for the MAAR and the CMD for updates.

T S e t u p p a t h-Delay in establishing a trail.

T f o r w a r d d a t a b u f f e r - delayed forwarding with a buffer.

3.5.3 Packet loss

The amount of data packets that were lost during the handover was due to the inability to forward or buffer them. Equations 5, 6 represent the Packet loss for NB-DMM and the proposed.

In General,

PacketLoss = λHO* _*disruption*

where HO_disruption: Period when packets cannot be transmitted.

NB-DMM

P a c k e t L o s s N B − D M M = λ * H O d i s r u p t i o n N B − D M M (5)

Proposed

P a c k e t L o s s p r o p o s e d = λ * H O d i s r u p t i o n P r o p o s e d (6)

The dual-buffering mechanism greatly reduces the handover latency H O P r o p o s e d during packet disruption, bringing it close to zero.

PacketLoss _*Proposed* = λ HO* _{*disruption (Proposed)*} ≈ 0

In real situations where buffer capacity is limited, and signaling delays can change, some packet loss may still happen. To address this, we can set a limit on packet loss as follows in Equation 7:Where δ _**L** is the limited residual loss that can occur in non-ideal conditions. This value is small because the data plane remains available throughout the handover process.

E f f e c t i v e P a c k e t L o s s = λ * H O d i s r u p t i o n + δ L (7)

3.5.4 Out-of-order packets

Packets are received at the mobile node in a different order than they were originally transmitted, which affects application performance. Equations 8, 9 represent the Out-of-Order Packets for NB-DMM and the proposed.

NB-DMM

O o u t , N B D M M = P f o r w a r d − P r e c e i v e d (8)

Proposed

O o u t , P r o p o s e d D B S A H O = P f o r w a r d − P r e c e i v e d (9)

With perfect coordination between forwarded and incoming packets, out-of-order delivery in the suggested scheme is insignificant. In non-ideal operating conditions, there exists the possibility of limited reordering of packets, which is stated as in Equation 10:

E f f e c t i v e O o u t = P f o r w a r d − P r e c e v i e d + δ r e o r d e r i n g (10)

The analytical model robustness bound is shown in Table 2.

TABLE 2. Analytical model robustness bound.

3.6 Implementation feasibility analysis

In this study, a simulation is used to investigate the performance of the proposed method. OMNeT++ can be used to model packet behavior in great detail, including buffering, signaling, and signal timing. Simulations can be used to analyse protocols, but it is also worth considering how these solutions can be implemented in real networks. The simulated entities in our architecture are similar to the real entities in IP-based mobile networks. Indicatively, the Mobile Access and Anchor Router (MAAR) can be configured as an access or edge router in LTE/5G networks, and the Centralized Mobility Database (CMD) can serve as a central control unit for routing and forwarding. The proposed dual-buffering mechanism is an improvement to the software at the target MAAR. It has two logical buffers: the Forward Data Buffer (FDB), which contains packets that were in flight in the last MAAR, and the New Data Buffer (NDB), which contains packets in the CMD when a handover begins. Notably, we do not alter NB-DMM signaling messages, packet headers, and the operations of the correspondent node. Buffering is based on the regular queuing and scheduling techniques already provided by the modern routers and operating systems. Buffering commences only at handover, and packets are still delivered sequentially, either in FDB or NDB order, with no additional signaling and no loss. The dual buffering applies to operations that use kernel packet queues, SDN flow control, or programmable data-plane technologies. Buffering does not incur any overhead since it occurs only during the handover period; therefore, memory and processing overhead are low and proportional to traffic and handover delay. This implies that the solution can be deployed at the network edge without any hardware or protocol changes. The design does not require any new networking features; it is also sensitive to deployment issues, although this work has yet to be implemented on a physical testbed or an emulation network. We accept that the testbed validation has not yet been done. To evaluate the robustness in real-world scenarios, we plan to test it on future SDN-based emulation platforms and to apply it in experimental testbeds. Overall, these arguments suggest that we are efficient in simulation and can transfer the findings to practice.

3.7 Security impact

The NB-DMM improvement under consideration presupposes the introduction of signalling aggregation and centralised coordination via the CMD to minimise handover latency and signalling overhead, but these mechanisms can, at the same time, raise security concerns. Signalling aggregation includes several messages containing mobility-related control information and only one exchange between MAARs and the CMD, which, in theory, may increase the impact of signalling injection or modification attacks. The proposed design controls aggregation to the control plane and prevents data packets from reaching the data plane; thus, data-plane confidentiality and integrity are not compromised. Also, signing aggregation is not presented with new signalling entities or trust relationships, since the existing NB-DMM deployments have a trusted administrative domain of operation for CMD and MAARs. The current secure transport systems, e.g., IPsec or TLS tunnels between CMD and MAARs, can thus be reused as-is; aggregation is beneficial, as it enhances signalling efficiency without raising exposure to external attackers. There might be concerns with the centralised coordination through the CMD, as there is also a risk of a single point of failure or attacks targeting it, but the CMD already exists in the baseline NB-DMM as a mobility database, and the offered approach does not impose new responsibilities on it as a critical data-plane component. Notably, the packet forwarding during handover is not tied to signalling completion in the proposed method, thereby preserving data-plane continuity even in the presence of signalling delays. In this model of execution, the security effect of delayed or lost control signalling is mitigated by operational approaches to data forwarding that condition data forwarding on the completion of control signalling. Also, the risk of replay or spoofing attacks against aggregated signalling messages is not specific to the proposed approach, nor is it unique to the underlying baseline NB-DMM signalling; existing countermeasures, like sequence numbers, timestamps, mutual authentication, etc., can be applied directly, since message semantics are not affected by aggregation. The suggested signalling aggregation and centralised coordination, in general, do not create additional attack points beyond those already in NB-DMM, and the decoupling of data-plane continuity and control-plane signalling reduces the ability of signalling-level attacks to cause packet loss and service disruption. Though the scope of this work does not include a detailed security assessment of the mechanism through formal analysis or testbed validation, the mechanism is security equivalent to the baseline NB-DMM and is more robust during handover.

4 Results and discussion

In this section, simulation results are provided. It also compares the suggested strategy with the existing schemes using the key performance measures. The experiments were conducted using the OMNeT++ 6.0.2 simulation framework, and the DB-SAHO architecture was simulated with the following modules: Central Mobility Database (CMD), five Mobile Access and Anchor Routers (MAARs), a Mobile Node (MN), and a Correspondent Node (CN). OMNeT++ is utilized to offer a controlled and repeatable evaluation environment and allows systematic comparison of mobility management schemes without hardware-based limitations. Signaling, dual buffering, and data forwarding were simulated by writing custom C++ modules. The results obtained in the simulation were registered in CSV files, and with Python 3.11, using the Pandas and Matplotlib libraries, graphs and statistical tables were developed. The given results indicate the relative performance in the case of controlled mobility and traffic conditions, instead of worst-case or highly dynamic network settings. Buffer size, aggregation factor, and predictability of handover are set to representative values for all the schemes considered. The predefined simulation settings and the values are shown in Table 3.

TABLE 3. Predefined simulation settings and the values.

Figure 4 shows the handover delay for NB-DMM and DB-SAHO during mobility. NB-DMM had an average delay of 120 m and DB-SAHO of 85 m. A reduction in latency is achieved through elimination of extraneous signaling through aggregation, and a form of buffering (on the transmission side) that eliminates retransmissions.

Figure 5 depicts the damage of packets. NB-DMM lost nearly 3.5% of its price, but DB-SAHO reduced that to only 0.8%. The dual-buffering method in DB-SAHO holds the packets securely during departure from and recovers the packets after the handover is completed. Even though the analytical model reveals that the DB-SAHO has almost zero packet loss, simulation findings reveal that there is a little residual packet loss in the real network conditions.

Figure 6 shows that NB-DMM sent incorrect packets with an average rate of 2.2%. DB-SAHO, on the other hand, nearly entirely corrected the issue with only 0.3% of packets being out of order. This was possible due to the newly added buffers and forward memory buffer, which worked hand in hand to ensure the sequence integrity when transitioning. Although the analytical model indicates negligible out-of-order packets for DB-SAHO, simulation results show a small number of reordered packets under realistic network conditions.

Figure 7 shows the cost of the signaling as a function of the total amount of control messages sent during a handover. The average amount of communications sent by NB-DMM was 480 while DB-SAHO decreased it to 360, representing a 25% decrease. The improvement is the use of multi-message grouping, which reduces the overhead, while the correctness of state changes and state changes is not affected.

Comparison analysis has clearly shown that the DB-SAHO is better than the NB-DMM (in all parameters measured, Table 4). Signaling aggregation and reduction of redundant exchanges resulted in drastically reduced handover latency. The dual buffering mechanism is the reason for the observed improvements in packet loss and out-of-order packet delivery. DB-SAHO uses FDB and NDB to isolate in-flight and new packets, respectively, unlike earlier methods that use either single or no buffering. This separation prevents early packet forwarding and buffer overflows, which lead to reduced packet loss and improved packet ordering in the results. In addition, message aggregation significantly lowered the amount of signaling overhead by up to one-fourth, thereby improving scalability in dense mobility environments.

TABLE 4. Comparative analysis.

4.1 Behaviour under worst-case and high-mobility scenarios

The mobile nodes will travel at high speed in high-mobility or worst-case conditions, leading to frequent handoffs. Traffic overload is typically associated with these situations in access networks. As a result, signalling data may be delayed during handoff, leading to discrepancies in network routers. Handover signalling is important in data forwarding in a conventional system. During a handover, packets may be lost due to heavy traffic. Consequently, the probability of lost packets and non-sequential delivery increases. In the case of two consecutive handovers, they would have a compounding effect, reducing overall performance. Compared with other approaches, the proposed approach maintains data transfer during handover even under extremely dynamic conditions. The packets that have been forwarded and those that arrive at the same time are temporarily held in two separate buffers to allow data to be transmitted without signalling. The handover delay does not cause most of the packet loss and packet reordering; instead, it is the buffer availability. The presented method is more predictable in its behaviour than conventional NB-DMM, even under very extreme conditions that occur when the buffer capacity is exceeded.

4.2 Evaluation scope concerning latest DMM developments and SDN-Assisted distribution

The recent improvements in DMM and SDN-based mobility solutions primarily focus on the control plane and reducing signaling overhead, and usually require centralized controllers or additional system components. Since these solutions are built on various structures, a direct numerical analogy would require the use of entirely different transition models. Thus, the paper compares the proposed approach with the baseline NB-DMM to clearly demonstrate the influence of the proposed handover execution in the regular architecture. The suggested approach operates in the data-plane mode, it does not modify the fundamental NB-DMM control processes, and is consistent with the improved DMM and SDN-based mobility solutions.

4.3 Qualitative robustness of the proposed method under parameter variation

The simulation employs fixed numbers for buffer size, the aggregation factor (γ), and handover predictability. We can still assess the scheme’s strength by examining how each parameter affects the handover process, not just the chosen values. It is shown in Table 5.

TABLE 5. Robustness under parameter variation.

Buffer size: Buffer size mainly affects how well the system can handle short disruptions during handover. If the required value is higher than the buffer size, packet loss will increase, especially if the buffer is smaller than the time gap of the handover. Once the buffer size reaches a minimum threshold, making it larger does not remarkably change handover latency/packet order. The signaling cost and procedures are not affected by buffer size. Instead, the buffering interval, not the exact buffer size, is what decides system performance.

Aggregation factor: The aggregation factor determines the degree of signaling aggregation and the resulting signaling overhead. Changes in γ do not affect the order of handover operations or data-plane forwarding, so they do not impact handover latency, packet loss, or out-of-order delivery. When γ is low, signaling overhead is high, and when γ is high, signaling cost drops in proportion to 1/γ. Importantly, the proposed scheme keeps its relative signaling reduction across different γ values, showing that it is not sensitive to aggregation configuration.

Handover Predictability: Predictability of handover influences the correctness of timing of activation of buffers, as opposed to the correctness of packet delivery. Buffering is proactively started in early prediction or accurate prediction, allowing very little latency associated with handover. In the case of handover latency due to delayed prediction or error-prone prediction, there will be a small increase in the handover latency, but the loss of packets and out-of-order delivery is limited by the dual-buffering process, which separates the forwarded packets and the new packets. This proves that the schemes that have been suggested do not work on the assumption of ideal prediction.

Combined variation: In real operating conditions, various parameters can go out of nominal at the same time. Absolute values of HL, PL, OOP, and signaling cost can also vary in such non-ideal situations. However, the performance gains of the proposed scheme are the same as those of the baseline approaches. This behaviour is a sign that the improvements that have been observed are based on the architectural design and execution level decoupling as opposed to parameter tuning.

5 Conclusion and future work

This research has introduced DB-SAHO, which denotes Dual Buffering and Signalling Aggregation for Handover Optimization. DB-SAHO is an advanced enhancement of the NB-DMM and will bring enhancements in continuous user connectivity during the handover operations, an improved number of users to access the wireless network, and improved mobility parameters in the next-generation of wireless networks. The combination of an efficient dual buffer system with a simplified signalling aggregation method will minimize the number of data packets dropped, significantly improve the handover process and reduce the amount of signalling per user, which will ensure a seamless and durable experience for the users of the network as well as offering support for services requiring strict timing requirements, high reliability and persistent data flow. Furthermore, DB-SAHO can meet the needs of various and highly mobile operating environments, as well as support multiple wireless technologies. There are a number of obvious advantages of DB-SAHO, but there are also some issues. The handover situations that have been considered during the assessment were predictable, and the methodology has not been tested under conditions that produce less predictable mobility or under very congested network conditions. The dual-buffering method may also need additional modifications for use with more complicated network configurations or other transition norms. Future work will investigate the proposed approach using more varieties of traffic and mobility patterns. A comprehensive quantitative and experimental security evaluation is designated as future work. A further area is to apply ML to be able to predict mobility events and optimise the allocation of resources to buffers. The further development of research in such directions will allow DB-SAHO to be more complete and practical as a solution for managing roaming in next-generation communications equipment.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Nomenclature

PMIPv6 Proxy Mobile IPv6 DMM Distributed Mobility Management DBSAHO (Proposed) Dual Buffering and Signaling Aggregation of Handover Optimization NBDMM Network-based Distributed Mobility Management MAAR Mobile Access and Anchor Routers FDB Forward Data Buffer NDB New Data Buffer MN Mobile Node FDMM Fast Distributed Mobility Management CMD Centralized Mobility Database CN Correspondent Node RSSI Received Signal Strength Indicator AUM Aggregated Update Message HRM Handover Request Message SMAAR Serving MAAR TMAAR Target MAAR SigCost NBDMM Cost of transmitting signals in NBDMM SigCost Proposed Cost of transmitting signals in the recommended method α , β Value of every transmission Hop X − Y Number of travels N The quantity of handoffs HO NBDMM Handover of Network-Based DMM HO Proposed Handover of Proposed T Detect Detection Delay T Setuppath Delay in setting up the path T forward data buffer Delay in sending the information into a buffer γ Aggregation factor PL Packet Loss λ Arrival time of packets O out , NBDMM Out-of-order packets P forward Forward Packets P received Received Pack

Footnotes

Edited by: Ghanshyam Singh, University of Johannesburg, South Africa

Reviewed by: Sanjay Dhar Roy, National Institute of Technology, Durgapur, India

Angshuman Jana, Indian Institute of Information Technology Guwahati (IIITG), India