Adaptive Context-Aware VANET Routing Protocol for Intelligent Transportation Systems

Abstract

Vehicular Ad-Hoc Networks (VANETs) play a critical role in Intelligent Transportation Systems (ITS), enabling communication between vehicles and roadside infrastructure. This paper proposes an Adaptive Context-Aware VANET Routing (ACAVR) protocol designed to handle the challenges of high mobility, dynamic topology, and variable vehicle density in urban environments. The proposed protocol integrates context-aware routing, dynamic clustering, and geographic forwarding to enhance performance under diverse traffic conditions. Simulation results demonstrate that ACAVR achieves higher throughput, improved packet delivery ratio, lower end-to-end delay, and reduced routing overhead compared to existing routing schemes. The proposed ACAVR outperforms benchmark protocols such as DyTE, RGoV, and CAEL, improving PDR by 12–18%, reducing delay by 10–15%, and increasing throughput by 15–22%.

1. Introduction

Within Intelligent Transportation Systems (ITS), Vehicular Ad-Hoc Networks (VANETs) have emerged as an indispensable technology, supporting seamless data exchange between vehicles (V2V) and between vehicles and roadside units (V2I) [1,2]. Beyond safety-related functions, VANETs enable applications such as congestion management, emergency response, and entertainment services. As the number of connected vehicles continues to rise, particularly in densely populated urban areas, the demand for high-throughput, low-latency, and dependable communication has intensified [3].

1.1. Challenges in VANET Routing

Although numerous studies have explored routing mechanisms in VANETs, the task continues to pose considerable challenges due to the network’s inherent properties:

• High Mobility: The constant motion of vehicles leads to rapid and unpredictable topology variations, which often cause route disruptions and lower packet delivery rates [4].
• Dynamic Network Topology: Because the network structure changes continuously, routing algorithms must be capable of quick reconfiguration while maintaining reliable performance [5].
• Traffic Density Variation: The contrast between congested urban areas and sparsely populated highways produces inconsistent link availability, resulting in unstable communication patterns [6].
• Scalability Requirements: As vehicular networks expand, protocols must scale effectively without adding excessive routing overhead [7].
• Safety and Real-Time Demands: Safety-focused applications require near-instantaneous message delivery, making latency reduction essential [8].

Together, these constraints underline the necessity for context-aware and resource-efficient routing approaches that can deliver robust communication across the diverse and dynamic scenarios encountered in VANETs.

1.2. Existing VANET Routing Protocols

VANET routing protocols are generally categorized into topology-based, position-based, cluster-based, and hybrid approaches [1,2,3].

Topology-based protocols rely on maintaining end-to-end paths. Examples include AODV, DSR, and OLSR. They often suffer in highly dynamic environments due to frequent link failures [9,10].

Position-based protocols utilize geographic information to make forwarding decisions, such as GPSR [11]. These protocols reduce routing overhead but may struggle in sparse networks or urban areas with obstacles.

Cluster-based protocols group vehicles into clusters, where cluster heads manage intra- and inter-cluster communication [12]. These improve scalability but can introduce additional delay due to cluster management.

Hybrid protocols combine multiple strategies, e.g., geographic clustering with context-awareness, to balance reliability, latency, and scalability [13].

While these protocols provide improvements in specific scenarios, no existing protocol simultaneously achieves high throughput, low latency, minimal overhead, and adaptability across diverse traffic densities. This gap motivates the development of the proposed ACAVR protocol.

Among the existing routing schemes, DyTE and RGoV are classified as position-based and cluster-based protocols, respectively, while CAEL is a hybrid approach. These were selected as benchmarks due to their proven scalability and relevance to dense urban scenarios. Other well-known protocols, such as GPSR, GPCR, and AODV, were excluded due to either their static assumptions or lack of adaptability under high-mobility conditions.

1.3. Motivation and Research Gap

The main research gaps in VANET routing can be summarized as follows:

• Existing protocols often fail in high-mobility urban environments, leading to broken routes and dropped packets.
• Most solutions do not adapt dynamically to vehicle density or mobility patterns, resulting in suboptimal performance.
• Scalability remains a concern, especially for large-scale networks with heterogeneous vehicle types.
• Integration of context-awareness, clustering, and geographic strategies is limited in current research.

Addressing these gaps requires a comprehensive routing approach that combines dynamic clustering, context-aware routing decisions, and geographic awareness to optimize network performance.

1.4. Proposed Solution Overview

This paper proposes the Adaptive Context-Aware VANET Routing (ACAVR) protocol, designed to:

• Dynamically adapt routing decisions based on vehicle density, mobility, and road topology.
• Minimize routing overhead while ensuring high packet delivery ratio and low end-to-end delay.
• Integrate clustering mechanisms with geographic positioning to enhance scalability and reliability.
• Demonstrate superior performance compared to DyTE (Dynamic Trilateral Enrolment) [14], RGoV (Reliable Group of Vehicles) [15], and CAEL (Compacted Area with Effective Links) [16].

The network model of VANET is illustrated in Figure 1, showing the interactions between different vehicle types and roadside units (RSUs).

The key contributions of this study can be summarized as follows: (i) the development of a novel context-aware routing framework designed to address the challenges of highly dynamic vehicular environments, (ii) the incorporation of clustering techniques with geographic forwarding to improve scalability and communication reliability, and (iii) an extensive performance assessment demonstrating that ACAVR consistently outperforms state-of-the-art VANET routing protocols. Taken together, these contributions establish ACAVR as a resilient and efficient routing solution, well-suited for next-generation ITS applications that demand reliable, high-performance vehicular communication.

1.5. Organization of the Paper

The rest of the paper is organized as follows:

• Section 2 presents a detailed survey of existing VANET routing protocols and identifies research gaps.
• Section 3 introduces the proposed ACAVR protocol, including the network model, protocol flow, and mathematical formulations.
• Section 4 presents simulation results and performance comparisons with existing protocols.
• Section 5 concludes the paper and outlines directions for future research.

2. Related Work

Vehicular Ad-Hoc Networks (VANETs) have been extensively studied due to their potential to enhance intelligent transportation systems (ITS). Early seminal works introduced the foundational concepts of vehicular communication and mobility challenges [1,2]. These studies laid the groundwork for subsequent research exploring efficient routing, scalability, and reliability in dynamic vehicular environments.

Early vehicular routing research primarily adopted existing MANET protocols such as AODV (Ad hoc On-Demand Distance Vector) [9], DSR (Dynamic Source Routing) [10], and OLSR (Optimized Link State Routing) [17]. Although these schemes achieved satisfactory performance in traditional Ad-Hoc Networks, they proved less effective in highly dynamic vehicular settings where rapid node movement frequently breaks established routes. This limitation encouraged the emergence of position-based approaches such as GPSR (Greedy Perimeter Stateless Routing) [11] and GPCR (Greedy Perimeter Coordinator Routing) [6]. These methods rely on vehicles’ geographic positions to forward packets progressively toward the destination and employ perimeter or coordinator recovery modes when encountering routing voids, thereby improving their reliability in city environments. Later refinements like GpsrJ+ [18] enhanced junction handling and reduced route detours, while geographic opportunistic forwarding strategies [12,19] further leveraged neighbor awareness to increase delivery success. In parallel, cluster-oriented routing strategies [20] arranged vehicles into semi-stable groups managed by elected cluster heads, which coordinated message exchange within and between clusters. This organization reduced routing overhead and improved stability in networks where topology changed rapidly.

More recently, research attention has shifted toward hybrid routing frameworks that fuse positional, topological, and contextual information to achieve greater adaptability. Khan et al. [8] developed an adaptive hybrid mechanism to enhance route reliability, while Sharma et al. [13] demonstrated that routing decisions can be improved by incorporating vehicle density, mobility, and road geometry through context awareness. Energy management and balanced traffic distribution have also emerged as key optimization goals [3]. Protocols including DyTE [14], RGoV [15], and CAEL [16] specifically targeted dense-traffic conditions, dynamically tuning their route selections to prevent congestion and minimize path convergence.

A notable line of work now explores the infusion of artificial intelligence (AI) and machine learning (ML) into routing logic. Dutta et al. [21], Diaa et al. [22], and Gillani et al. [23] presented distributed learning approaches that safeguard data privacy while improving the predictive accuracy of routing decisions. Such AI-supported schemes allow vehicles to forecast potential disconnections and reroute traffic preemptively, leading to improved stability in practical deployments.

With the advent of 5G and the anticipated shift toward 6G networks, several studies have explored how edge and cloud computing can reinforce vehicular communication. Ali et al. [24,25] and Ramamoorthy et al. [26] highlighted the latency reduction and computational efficiency achieved through multi-access edge computing (MEC). Similarly, Feraudo et al. [27] and Shaheen et al. [28] proposed blockchain-based models to ensure secure and auditable data exchange among vehicles. Complementing these, Gomides et al. [29] and Muthaiyan et al. [30] examined software-defined networking (SDN)–driven VANET frameworks that centralize control for improved resource management and bandwidth allocation in congested environments.

Cross-disciplinary approaches are also gaining traction. Elsayed et al. [31] investigated QoS-driven and AI-enhanced routing frameworks, while Arif et al. [32] focused on probabilistic and fuzzy-logic-based approaches for dynamic environments. Mohammad et al. [33] analyzed secure and delay-tolerant routing, presenting methods to address packet drops and malicious attacks in VANETs.

Collectively, the literature shows a clear evolution: from topology-based protocols to geographic and hybrid approaches, and more recently toward AI-driven, blockchain-enabled, and edge-assisted solutions. Despite these advances, the challenges of maintaining reliability, scalability, and efficiency in highly dynamic VANET scenarios remain open research problems. This motivates the design of new context-aware and adaptive routing protocols that can integrate the strengths of prior approaches while addressing emerging vehicular communication demands. DyTE, RGoV, and CAEL were selected due to their alignment with urban topology, cluster stability, and hybrid adaptability.

3. Methodology

This section elaborates on the proposed Adaptive Context-Aware VANET Routing (ACAVR) protocol by explaining its core components — the network model, overall protocol design (Figure 2), routing mechanism, and selected evaluation metrics. The purpose of the proposed approach is to address key limitations of urban vehicular communication, particularly the frequent topology variations, inconsistent vehicle densities, and unstable connectivity resulting from fast vehicular motion.

3.1. Network Model

In this study, a Vehicular Ad-Hoc Network (VANET) is modeled, comprising N mobile vehicles and M stationary roadside units (RSUs) deployed throughout an urban layout. Every vehicle is assumed to possess a Global Positioning System (GPS) receiver along with wireless interfaces capable of supporting both vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) interactions. Due to the dynamic mobility patterns of vehicles, the network topology continuously changes as nodes enter or leave each other’s transmission range.

The entire network is represented as a graph $G=(V,E)$, where V refers to the set of participating entities (vehicles and RSUs), and E represents the set of possible communication links. A link $(i,j)\in E$ exists if the Euclidean distance between nodes i and j is less than or equal to a predefined communication threshold $R_t$. Hence, network connectivity at any given moment depends directly on the vehicles’ relative positions and movement patterns:

$$E=\left\{(i,j)\right|d(i,j)\le R_t,i,j\in V\}$$

(1)

Here, $d(i,j)$ represents the Euclidean distance between nodes i and j.

Although the transmission range $R_t$ is assumed constant for the purpose of analysis, in realistic vehicular communication environments it fluctuates due to factors such as path loss, shadowing, and multipath fading. To improve accuracy in future versions, this model can be extended to integrate probabilistic wireless channel representations, such as Rayleigh or Nakagami-m fading, which would yield more realistic transmission dynamics.

In addition to the basic network formulation, the ACAVR framework integrates a mobility prediction feature to refine routing decisions. Each vehicle anticipates its future location through a Kalman filter, which processes GPS-based parameters such as speed and heading direction. By forecasting short-term movement trends, the system identifies neighboring nodes that are expected to remain within the communication range for a longer duration. This predictive insight enables the selection of more reliable next-hop vehicles, thereby contributing to steadier packet delivery and enhanced routing performance.

3.2. Protocol Architecture

The ACAVR framework introduces a dynamic, context-aware routing strategy that can shift between different forwarding modes according to the current traffic and connectivity conditions. When the network becomes highly congested, data packets follow a greedy forwarding scheme to exploit dense neighborhood links. In regions with moderate traffic, a cluster-oriented approach is adopted to balance routing efficiency and control overhead. Conversely, during sparse or fragmented connectivity, the protocol employs a store–carry–forward technique to maintain message delivery even across intermittent links. This smooth adaptation among routing modes strengthens the protocol’s resilience and ensures dependable communication despite continuous topology variations. The complete operation of ACAVR relies on three closely integrated functional modules:

• Context-Aware Routing Module: Analyzes instantaneous vehicle parameters such as mobility, neighborhood density, and surrounding road layout to determine optimal relay paths.
• Dynamic Clustering Module: Organizes vehicles into short-lived clusters, thereby lowering routing overhead and improving scalability. Each cluster designates a Cluster Head (CH) by evaluating metrics like link stability, speed uniformity, and relative position.
• Geographic Forwarding Module: Utilizes GPS coordinates to transmit packets toward the intended destination while minimizing unnecessary detours or congestion points.

Enhanced Adaptive Clustering and Context Evaluation: To make routing more reactive to environmental variations, the ACAVR protocol keeps assessing each vehicle’s operational status through a set of contextual indicators such as the current speed ($v_i$), inter-vehicle distance ($d_i$), and connectivity index ($c_i$). Each parameter is first scaled using the min–max normalization method so that they all share a comparable numerical range:

$$x^{\prime }={\displaystyle \frac{x-x_{min}}{x_{max}-x_{min}}}$$

This normalization prevents any single parameter from dominating the decision process. A combined context measure, termed the node’s weight $W_i$, is then obtained as:

$$W_i=\alpha v_i^{\prime }+\beta d_i^{\prime }+\gamma c_i^{\prime }$$

The coefficients $\alpha$, $\beta$, and $\gamma$ vary dynamically with network density ($\rho$) and the average link lifetime ($L_t$), satisfying:

$$\alpha +\beta +\gamma =1,\mathrm{where}\alpha =f\left(\rho \right),\beta =f\left(L_t\right)$$

When vehicle density grows, $\alpha$ becomes more significant, prioritizing speed-related responsiveness. In contrast, during sparse traffic, $\beta$ and $\gamma$ gain influence to sustain link quality and continuity. This adaptive weighting is what gives ACAVR its responsive and self-regulating character.

Enhanced Routing Decision Process: Once Cluster Heads (CHs) have been determined, the routing phase refines the forwarding path through local context sharing among neighboring vehicles. Each participant periodically broadcasts beacon packets containing updated state data, allowing surrounding nodes to identify stable next-hop candidates collectively. The ACAVR controller interprets these shared parameters to determine a route that minimizes latency while maintaining high delivery success. This distributed approach ensures reliable communication performance in both densely built city layouts and relatively open roadway conditions.

3.3. Cluster Formation and Maintenance

Clusters are organized by assigning each vehicle i a score $W_i$, calculated through the weighted sum:

$$W_i=\alpha ·S_i+\beta ·C_i+\gamma ·D_i$$

(2)

where:

-

$S_i$ = instantaneous speed of the vehicle (lower speed signifies better stability for CH candidacy)

-

$C_i$ = node connectivity, representing the number of nearby neighbors

-

$D_i$ = distance between the vehicle and the cluster’s geometric center

-

$\alpha ,\beta ,\gamma$ = tuning coefficients that fulfill $\alpha +\beta +\gamma =1$

The vehicle yielding the highest $W_i$ is selected as the Cluster Head. To accommodate frequent topology variations, cluster structures are updated at regular intervals of $T_c$ seconds.

The coefficients $\alpha$, $\beta$, and $\gamma$ were initially chosen through repeated sensitivity analyses to balance the contribution of each factor. All parameters were normalized within [0, 1] to avoid metric bias. Experimental trials showed stable performance when $\alpha =0.3$, $\beta =0.5$, and $\gamma =0.2$. During runtime, these coefficients are further fine-tuned according to localized vehicle density and mobility, ensuring adaptive and consistent routing outcomes under diverse operating scenarios.

3.4. Routing Algorithm

The ACAVR routing algorithm functions in three consecutive stages (pseudo-code provided in Algorithm 1):

• Context Evaluation: Each node determines a Context Score (CS) using the density of neighboring vehicles, velocity characteristics, and the type of road segment:

  $$C{S}_i=\delta ·{\rho }_i+ϵ·v_i+\zeta ·R_i$$

  (3)
  Here, ${\rho }_i$ is the local vehicle density, $v_i$ denotes relative speed, and $R_i$ represents the road-type factor. The coefficients $\delta$, $ϵ$, and $\zeta$ are adjustable to align with environmental conditions. To minimize frequent re-elections, each CH maintains a stability index that records its velocity variance and history of link breaks. When this stability metric drops below a set limit, a new CH is selected, effectively reducing unnecessary control traffic and maintaining steady cluster organization.
• Packet Forwarding Decision: Packets are forwarded to the neighbor with the highest CS, ensuring progress towards the destination while maintaining link stability.
• Cluster-Assisted Routing: Inter-cluster routing is performed via CHs, reducing the number of hops and control overhead. Intra-cluster packets are broadcast efficiently within the cluster.

Algorithm 1: ACAVR Routing Algorithm Pseudo-Code.

Input: Source node S, Destination node D, Vehicle set V, RSU set R Output: Forwarding path from S to D 1: Initialize clusters using weighted metric $W_i$ for each vehicle 2: For each cluster, elect Cluster Head (CH) 3: While S has packets to send: 4:    Evaluate Context Score (CS) for all neighbors 5:    Select neighbor with highest CS as next hop 6:    If next hop is CH, forward packet to CH for inter-cluster routing 7:    Else, forward packet directly to next hop 8:    Update cluster membership every $T_c$ seconds 9: End While

4. Results and Discussion

This section presents the performance evaluation of the proposed ACAVR protocol compared with existing VANET routing protocols: DyTE [14], RGoV [15], and CAEL [16].

4.1. Simulation Setup

To ensure realistic evaluation, the simulations employ the SUMO (Simulation of Urban Mobility) [34] tool integrated with NS-2 [35]. This model replicates both highway and urban traffic scenarios, enabling ACAVR to be tested under heterogeneous conditions, and each protocol is evaluated over multiple runs to ensure statistical validity. Such modeling provides more practical results than mobility models like Random Waypoint. All the key parameters with their values are tabulated in

Table 1. Simulation Parameters for ACAVR Evaluation.

Parameter	Value
Simulation area	$2000\times 2000$ m2
Number of vehicles	50–300
Vehicle speed	20–80 km/h
Transmission range	300 m
MAC protocol	IEEE 802.11p
Simulation time	900 s
Channel Bandwidth	10 MHz
Data Rate	6 Mbps
Mobility model	SUMO urban traces

.

The simulation parameters in Table 1 were selected after a comprehensive review of recent and widely accepted VANET studies that utilized NS2 and SUMO for hybrid network simulation. The values for transmission range (250 m), data rate (6 Mbps), and channel bandwidth (10 MHz) were aligned with IEEE 802.11p standards operating in the 5.9 GHz Dedicated Short-Range Communication (DSRC) band. Vehicle speeds (20–80 km/h) and density values were based on realistic urban traffic patterns derived from SUMO mobility traces, while simulation duration (900 s) ensured sufficient temporal granularity to capture network dynamics under varying traffic conditions. These parameters were validated through multiple preliminary simulation runs to ensure convergence and stability in results, ensuring both repeatability and comparability with existing benchmarks. This setup allows us to evaluate the proposed ACAVR protocol under dense, high-mobility urban traffic scenarios, comparing its performance against DyTE [14], RGoV [15], and CAEL [16].

Although IEEE 802.11bd has recently been standardized, 802.11p remains widely used for comparative benchmarking and ensures compatibility with prior VANET studies. The selected simulation parameters were aligned with existing works for fair evaluation. During the SUMO and NS2 integration, synchronization challenges were encountered in maintaining mobility trace accuracy, which were resolved by re-sampling vehicular traces every 100 ms to ensure temporal consistency.

4.2. Performance Metrics

The following standard performance metrics were considered for the evaluation of ACAVR:

• Throughput (Mbps): Amount of data successfully delivered per unit time.
• Packet Delivery Ratio (PDR): Ratio of packets received to packets sent.
• End-to-End Delay (ms): Average time taken for a packet to reach the destination.
• Routing Overhead (%): Ratio of control packets to data packets.

4.3. Throughput Analysis

Throughput is measured as the total amount of data successfully delivered per unit time. Figure 3 shows that ACAVR consistently outperforms DyTE, RGoV, and CAEL across all traffic densities.

The improvement in throughput (approximately 22%) is attributed to dynamic clustering and context-aware forwarding, which reduce packet loss and optimize route selection.

4.4. Packet Delivery Ratio (PDR)

PDR measures the ratio of packets successfully received to packets sent. Figure 4 illustrates that ACAVR achieves higher PDR under all vehicle densities.

The enhanced PDR (18% improvement) results from cluster-assisted routing that maintains connectivity even in highly mobile networks, reducing packet drops.

4.5. End-to-End Delay

End-to-end delay measures the average time for packets to reach the destination. As shown in Figure 5, ACAVR reduces latency significantly compared to DyTE, RGoV, and CAEL.

The reduction in delay (approximately 15%) is due to optimized next-hop selection based on the context score (CS), which avoids congested or unstable links.

4.6. Routing Overhead

Routing overhead is the ratio of control packets to data packets.

Table 2. Routing Overhead Comparison (%).

Protocol	Routing Overhead (%)
DyTE [14]	18.5
RGoV [15]	16.2
CAEL [16]	14.8
ACAVR (Proposed)	12.5

summarizes the overhead for all protocols.

ACAVR reduces routing overhead through cluster-assisted forwarding, minimizing unnecessary control messages while maintaining high reliability.

4.7. Overall Performance Comparison

Figure 6 presents a combined bar chart comparing throughput, PDR, delay, and overhead across all protocols. The Composite Score (ComS) used in Figure 6 was computed as a normalized weighted aggregation of major performance metrics:

$$ComS=0.4\times PDR+0.3\times {\displaystyle \frac{1}{Delay}}+0.2\times Throughput-0.1\times Overhead$$

(4)

This formulation provides a balanced comparison by rewarding delivery efficiency and penalizing latency and control overhead.

ACAVR consistently outperforms existing protocols across all performance metrics. Its adaptive routing strategy, context-aware forwarding, and cluster-based design enable it to achieve higher throughput, improved PDR, lower latency, and reduced overhead, making it suitable for dense urban VANET scenarios.

The performance criteria defined at the beginning of the study—namely Packet Delivery Ratio (PDR), Throughput, End-to-End Delay (E2E), and Routing Overhead—were selected as they comprehensively represent network efficiency, reliability, and latency behavior in VANET environments.

During the simulation and analysis stages, these parameters continued to effectively capture the behavior of the proposed ACAVR protocol under varying traffic and mobility conditions. Hence, no additional parameters were required, and none were discarded. However, to ensure robustness, a composite performance score was later introduced to summarize overall network performance across these multiple metrics, without altering the originally defined criteria.

Therefore, the evaluation framework remained valid and consistent throughout the research process.

4.8. Scalability Analysis

To assess scalability, simulations are conducted with vehicle densities ranging from 50 to 300 nodes. Figure 7 shows that ACAVR maintains performance with minimal degradation as network size increases.

Scalability is ensured by dynamic clustering, which limits the number of nodes involved in each routing decision, reducing computation and communication overhead.

4.9. Summary of Results

Table 3. Percentage Improvement of ACAVR over Existing Protocols.

Metric	DyTE	RGoV	CAEL
Throughput (%)	+22	+18	+15
PDR (%)	+18	+15	+12
End-to-End Delay (%)	−15	−12	−10
Routing Overhead (%)	−6	−3.7	−2.3

summarizes the percentage improvement of ACAVR over DyTE, RGoV, and CAEL.

The results demonstrate that ACAVR achieves superior performance across all critical metrics, validating its effectiveness in dynamic urban VANET environments. The clustering process runs in $O\left(nlogn\right)$ due to weighted CH selection, while route discovery is near-linear in neighbor size. The fallback routing adds only constant overhead, keeping ACAVR computationally efficient and scalable for large VANET deployments.

5. Conclusions

This paper proposed the Adaptive Context-Aware VANET Routing (ACAVR) protocol, designed to address the challenges of high mobility, dynamic topology, and variable traffic density in urban vehicular networks. By integrating context-aware routing, dynamic clustering, and geographic forwarding, ACAVR optimizes routing decisions to achieve high throughput, improved packet delivery ratio (PDR), low end-to-end delay, and reduced routing overhead.

Extensive simulations conducted with NS-2 and realistic SUMO-based urban mobility scenarios indicate that the proposed ACAVR protocol consistently outperforms several well-established VANET routing solutions, including DyTE, RGoV, and CAEL. Performance analysis indicates that ACAVR achieves up to 22% higher throughput, along with an 18% increase in packet delivery ratio (PDR) and a 15% decrease in average end-to-end delay, while maintaining notably lower routing overhead. Furthermore, the scalability assessment confirms that the proposed approach sustains its efficiency as the network size increases, indicating that its operation remains stable even in high-density urban traffic scenarios.

The key strength of ACAVR comes from its adaptive, context-aware framework, which enables the routing process to respond naturally to continuous topological variations in vehicular networks. Through the consistent selection of durable communication links and careful limitation of redundant control signaling, ACAVR achieves reliable data forwarding while keeping routing overhead under control. Its hybrid strategy—merging dynamic clustering with context-aware geographic forwarding—offers a practical balance between responsiveness and efficiency, making it particularly well-suited for complex, high-density urban traffic scenarios. This makes ACAVR a well-suited approach for ITS applications that require low-latency and reliable communication, such as improving road safety, preventing accidents, and providing real-time traffic updates. Future work will extend the protocol for 802.11bd-based simulations and integrate channel fading models for improved realism.