Hybrid Bio-Inspired Algorithm Improves Vehicular Network Routing Stability and Data Delivery

A newly published simulation study in the journal Discover Internet of Things demonstrates that combining two distinct nature-inspired optimization models significantly improves communication stability and data delivery across vehicular ad hoc networks. Developed by researchers Ataollah Sattari, Ali Ghaffari, and Abbas Mirzaei, the novel framework—designated as CRAHO—integrates the global exploration mechanics of the Crow Search Algorithm with the localized dynamic exploitation strategies of the Harris Hawks Optimization algorithm. Vehicular ad hoc networks, commonly abbreviated as VANETs, form an indispensable pillar of modern intelligent transportation systems by facilitating automated vehicle-to-vehicle, vehicle-to-infrastructure, and vehicle-to-everything message exchanges designed to increase roadway safety and streamline traffic efficiency. However, because connected vehicles travel at high speeds across unpredictable urban pathways, network topologies fluctuate continuously, causing frequent wireless link disconnections, severe communication bottlenecks, and substantial transmission delays that classical routing mechanisms struggle to resolve.

To mitigate the extreme instability inherent in mobile vehicular topologies, network engineers frequently partition vehicles into localized groups known as clusters. Within this structural hierarchy, designated representative vehicles, known as cluster heads, collect data packets from surrounding cluster members and forward them along multi-hop routes across the network. While clustering reduces overall routing complexity and controls communication overhead, identifying optimal cluster heads and coordinating multi-hop paths among them are classified mathematically as non-deterministic polynomial-time hard, or NP-hard, optimization challenges. In conventional vehicular network routing, standalone meta-heuristic algorithms such as genetic algorithms, particle swarm optimization, or single-stage grey wolf optimizers have been deployed to search for near-optimal routing paths. Nevertheless, the authors emphasize that existing single-stage models routinely fail to balance the competing priorities of broad exploration across the entire network topology and rapid exploitation of promising pathways, often stagnating prematurely in local optima when vehicular mobility patterns shift rapidly.

The fundamental operational architecture of the CRAHO framework resolves these historical limitations by establishing a decoupled, sequential two-stage optimization pipeline that divides clustering and routing into separate algorithmic responsibilities. In the first phase, the Crow Search Algorithm is executed globally across all active vehicles in a given geographical region to elect stable cluster heads. Modeled on the food-caching and thieving behaviors of flocking crows, this algorithm leverages an awareness probability parameter and dynamic flight lengths to search the topological landscape without prematurely collapsing into localized configurations. Candidate cluster heads are evaluated through a normalized multi-objective fitness function that systematically minimizes intra-cluster distance and relative velocity differences while simultaneously maximizing node degree. This design ensures that elected vehicle leaders share similar travel vectors with their neighboring peers, thereby preventing premature link breakage and reducing the frequency of costly network re-clustering events.

Once the Crow Search Algorithm has partitioned the vehicular nodes into cohesive clusters, the network topology undergoes a logical abstraction that substantially reduces the computational search space. Rather than forcing the routing protocol to evaluate every individual vehicle operating across the roadway, the second phase restricts route discovery exclusively to the virtual backbone formed by the elected cluster heads and target roadside units or base stations. At this handover juncture, the Harris Hawks Optimization algorithm is activated to establish the most reliable and efficient inter-cluster transmission paths. Inspired by the cooperative surprise-pounce hunting strategies of brown hawks—specifically their transitions between soft encirclement, hard encirclement, and progressive rapid dives governed by the calculated escaping energy of their prey—the algorithm exploits path configurations to discover low-latency, energy-efficient multi-hop trajectories across the cluster-head backbone.

A critical mathematical contribution detailed in the study is the specialized integration between the two algorithms to guarantee continuous algorithmic progress. In standard implementations of the Harris Hawks algorithm, the optimization process relies on a linearly decreasing escape energy variable; however, in fast-changing vehicular environments, this mechanism frequently succumbs to premature convergence as search agents become trapped in suboptimal local minima. Within CRAHO, when the variance of the cluster-head fitness gradient approaches zero, the embedded awareness probability mechanism derived from the Crow Search Algorithm operates as a non-linear stochastic trigger. If an optimization agent detects diminishing returns caused by sudden vehicle movements, the probabilistic condition forces a global exploratory jump. This structural coupling maintains necessary population diversity throughout the optimization search, preserving a total time complexity of O(Tmax × N × (D + log N)).

To establish optimal operational parameters for the multi-objective fitness functions, the research team conducted systematic sensitivity analyses and grid-search parameter tuning during the preliminary simulation phases. For the clustering stage governed by the Crow Search Algorithm, the normalized weighting parameters were thoroughly evaluated across thirty independent trials in a fifty-vehicle test topology. Because relative velocity directly determines whether neighboring vehicles maintain persistent wireless connectivity, the authors assigned the primary priority weight of zero point four five to relative vehicular velocity. Average intra-cluster distance received a weight of zero point three five to constrain propagation delays and transmission energy consumption, while normalized node degree received a weight of zero point two zero to act as a stabilizing tie-breaker. Empirical results confirmed that this mobility-centric distribution increased average cluster head duration by twelve point four percent relative to equal-weight baseline allocations.

The routing phase implemented through the Harris Hawks Optimization algorithm similarly relies on a balanced evaluation of multi-hop network attributes. The routing objective function incorporates end-to-end transmission latency, overall energy consumption across forwarding nodes, and wireless link stability. Consequently, the authors adopted a balanced parameter distribution across delay, energy, and link stability metrics. This configuration prevents search agents from selecting vulnerable intermediate hops, ensuring that inter-cluster data transmission routes remain structurally viable long enough for packets to traverse the multi-hop network safely.

The quantitative performance of the CRAHO protocol was evaluated using a rigorous joint simulation environment combining MATLAB for network protocol execution and algorithmic iterations with the Simulation of Urban MObility platform, known as SUMO, for realistic vehicular mobility generation. The simulation modeled an urban Manhattan grid environment measuring two thousand meters by two thousand meters, implementing the Krauss car-following model to accurately reproduce vehicle acceleration, deceleration, and intersection queuing dynamics. Communication parameters adhered to the IEEE 802.11p wireless access in vehicular environments standard, utilizing a two-megabit-per-second channel bandwidth, sixty-four-bit packet payloads, and a Nakagami-m fading distribution to represent urban multipath signal interference. Vehicle speeds were dynamically distributed across a broad range from ten meters per second to forty meters per second across densities scaling from fifty to two hundred and fifty vehicles.

Across all evaluated performance benchmarks, the proposed CRAHO framework demonstrated statistically significant superiority, corroborated by narrow ninety-five percent confidence intervals derived from thirty independent randomized simulation runs. In comparative trials against modern baseline protocols—specifically the energy-efficient clustering protocols EEWCA and EECBR, as well as classical reactive routing via AODV and shortest-path calculation through Dijkstra’s algorithm—CRAHO sustained an average packet delivery ratio of ninety-one point three five percent, peaking near ninety-eight percent in moderately dense vehicular arrangements. Furthermore, CRAHO improved network throughput by eleven point one nine percent over EEWCA and thirteen point four five percent over EECBR, while reducing average end-to-end packet delay to forty-three point one eight milliseconds. By distributing transmission loads across stable intermediate nodes, the protocol also extended the network lifetime metric, measured by the time elapsed until the first node depletes its battery energy, by approximately eighteen percent and twenty-four percent over the respective baseline systems.

Despite these strong quantitative results, the researchers delineate several operational boundaries and technological limitations that define clear priorities for subsequent inquiry. Because current evaluations were conducted within numerical simulation software, future validation requires hardware-in-the-loop testing and co-simulation through dedicated platforms such as VEINS to account for physical signal obstruction, radio frequency interference, and MAC-layer contention under real-world conditions. Furthermore, because onboard vehicular processing units possess bounded computational resources, deploying dual-metaheuristic frameworks will require algorithmic pruning or specialized hardware acceleration. The authors outline planned extensions that include integrating deep reinforcement learning for predictive route recalculation prior to physical link failure, testing the architecture across cellular vehicle-to-everything fifth-generation and sixth-generation communication frameworks, and incorporating trust-management mechanisms to defend the routing backbone against malicious network attacks.

Subject of Research: Clustering-based routing optimization in vehicular ad hoc networks using hybrid meta-heuristics

Article Title: A hybrid crow search and Harris Hawks optimization approach for clustering based routing in vehicular ad hoc networks

Article References: Sattari, A., Ghaffari, A., & Mirzaei, A. (2026). A hybrid crow search and Harris Hawks optimization approach for clustering based routing in vehicular ad hoc networks. Discover Internet of Things, 6(1), Article 136. https://doi.org/10.1007/s43926-026-00483-8

Image Credits: AI Generated

DOI: 10.1007/s43926-026-00483-8

Keywords: vehicular ad hoc networks, clustering routing, Crow Search Algorithm, Harris Hawks Optimization, CRAHO protocol, packet delivery ratio, network throughput, end-to-end delay, cluster head selection, intelligent transportation systems, SUMO simulation, IEEE 802.11p, meta-heuristic optimization, link stability, swarm intelligence, dynamic topology