Wolf Pack Algorithm for UAV Swarm Control: a Review and Opportunities for Modification

Abstract

Introduction. UAV swarm control is a complex problem that requires efficient real-time optimization under uncertainty, limited resources and dynamic environmental constraints. Centralized approaches to task assignment and coordination often suffer from poor scalability, communication delays and vulnerability to failures of individual nodes, which motivates the use of swarm-intelligence-based methods capable of generating near-optimal solutions without a complete analytical model of the environment. Among such methods, the Wolf Pack Algorithm (WPA) is a promising metaheuristic inspired by cooperative wolf hunting. It organizes the population through role differentiation and phase-based search, combining exploration and exploitation via scouting, calling and besieging behaviors.

The aim of this review is to analyze the Wolf Pack Algorithm and assess its potential for solving UAV swarm control problems. The paper summarizes the biological foundations and mathematical model of WPA, presents a generalized formulation of cooperative reconnaissance and task allocation problems for UAV swarms and describes the main phases of the algorithm together with its population update mechanism. Particular attention is given to the use of WPA in two major application areas: UAV task assignment and trajectory/path planning, including three-dimensional flight scenarios in constrained environments. The review also systematizes known improvements of WPA, including discrete adaptations, hybrid PSO/GAbased variants, elite strategies, probability-threshold mechanisms and multi-population parallel schemes designed for real-time requirements. In addition, the paper analyzes practical limitations of applying the Wolf Pack Algorithm in real UAV swarms, including energy consumption, computational complexity, communication bandwidth, swarm resilience and the need for partial task redistribution in dynamic environments.

Methods. The study is based on a literature review and theoretical analysis of the Wolf Pack Algorithm and its recent modifications. The structure of WPA is examined through its objective-function formulation, role-based population model and iterative update rules for scouting, calling and besieging phases. The paper also considers a generalized mathematical formulation of multi-UAV cooperative reconnaissance, where task assignment, route planning, risk, time, energy, communication and resource constraints are integrated into a multi-criteria optimization framework. Recent publications are analyzed to identify how WPA is adapted to discrete search spaces, how convergence is accelerated through hybridization and parallelization and how elite and probabilistic mechanisms improve search robustness in UAV swarm applications. Additional attention is paid to resource-aware adaptation, fault tolerance, event-triggered partial replanning and simulation-based validation as necessary conditions for transferring WPA from theoretical optimization models to engineering-oriented UAV swarm systems.

Results. The review shows that WPA provides a flexible framework for solving complex UAV swarm problems in both discrete and continuous search spaces. In task-assignment scenarios, modified versions such as Multi-Discrete WPA (MDWPA) and PSO-GA-DWPA improve solution accuracy, robustness and convergence speed in large-scale combinatorial settings. In cooperative reconnaissance problems, elite strategies and probability-threshold mechanisms improve the exploitation of highquality solutions while reducing the risk of premature convergence and local trapping. In trajectory-planning tasks, including 3D path planning in complex terrains, improved WPA variants demonstrate higher efficiency than the basic algorithm by incorporating mutation strategies, adaptive step adjustment, chaotic initialization and enhanced local-search mechanisms. At the same time, the standard WPA still has limitations, particularly slow local optimization in later iterations, the possibility of stagnation near local optima and a potential mismatch between algorithmic complexity and the limited onboard resources of UAV platforms. These shortcomings are addressed in the literature through adaptive operators, hybrid search components, multi-population parallel structures, resource-aware formulations and partial replanning mechanisms that shorten convergence time, preserve population diversity and support swarm resilience under changing mission conditions. Overall, the reviewed studies indicate that WPA and its improved variants are promising for UAV swarm control tasks requiring decentralized coordination, high-dimensional optimization, dynamic task redistribution and adaptation to uncertainty.

Conclusions. The Wolf Pack Algorithm is a promising metaheuristic approach for UAV swarm control because it combines biologically inspired collective behavior with a formal global-optimization framework. Its role-based population structure and phase-based search allow it to balance global exploration and local exploitation, which is especially valuable in large-scale, constrained, and uncertain UAV missions. The reviewed literature indicates that WPA and its modifications are applicable to two key classes of problems: task assignment and trajectory planning. Discrete, hybrid, elite and multi-population variants improve the accuracy, stability and convergence speed of the basic algorithm, making WPA a flexible platform for the development of adaptive swarm-intelligence methods in autonomous UAV systems. However, the practical applicability of the Wolf Pack Algorithm in real UAV swarms cannot be assessed only by the quality of the obtained optimization solution. Energy limitations, computational cost, communication constraints, loss of individual UAVs, and the need for partial event-triggered replanning remain critical factors for engineering implementation. Therefore, the further development of the Wolf Pack Algorithm should include resource-aware modifications, mechanisms for partial task redistribution, and simulation-based validation in robotic environments such as ROS/Gazebo, Webots, AirSim, or RotorS. Such validation is necessary to determine whether WPA-based approaches can maintain acceptable solution quality under limited energy, limited communication, UAV failures, newly emerging tasks and real-time replanning requirements.