Wolf Pack Algorithm

The Wolf Pack Algorithm (WPA) is a nature-inspired metaheuristic optimization algorithm, inspired by the social behavior and hunting strategies of wolves, particularly how they hunt in packs.

Real-World Wolf Pack Behavior

1. The alpha wolf leads the pack, representing the best-known hunting path — the others watch and learn from its direction. This leadership provides a reference point for others to follow or challenge.
2. Some wolves break away as scouts, venturing into uncharted areas to search for better hunting grounds. Their goal is to discover signs of prey that the alpha might have missed.
3. When a scout finds promising prey, it howls to alert the rest of the pack. This communication allows the group to redirect their effort toward the new discovery.
4. The pack responds by converging toward the caller, reevaluating the new location against the alpha’s path. This reinforces the best strategy, combining individual finds with group intelligence.
5. As the wolves close in on the prey, they surround it and coordinate their movements.This siege behavior ensures the prey (solution) is effectively captured by narrowing down options.
6. If the hunt fails or prey escapes, the wolves remember and adapt, updating their tactics and dispersing again. This leads back to exploration, where new scouting efforts begin based on learned experiences.

Steps of WPA

Initialization

• Generate a population of wolves (candidate solutions), each with a position in the search space.
• Identify the alpha wolf (best solution found so far).

Scouting (Exploration)

• Wolves explore the environment individually.
• Wolves use a random walk to discover potential prey (i.e., better solutions).
• If a wolf finds a better solution than the alpha, the alpha updates.

Calling (Information Sharing)

• When a wolf finds a good solution, it "howls" to signal others to move toward it.
• This increases convergence by focusing the search.

Siege (Exploitation)

• Wolves surround the prey (solution) and refine their positions to get even closer.
• The siege phase improves local search and convergence.

Updating

• The population is updated, replacing weaker wolves.
• Repeat scouting and sieging until termination criteria are met.

Mathematical Modeling

Let:

1. Initialization

2. Scouting Phase

3. Calling Phase (Communication & Attraction)

4. Siege Phase (Exploitation)

5. Adaptation (Escape Local Optima)

Repeat scouting → calling → siege → adaptation for up to Tmax iterations.

Applications

• Engineering Design Optimization: WPA Can handle high-dimensional, nonlinear, and constrained design spaces.
• Electrical and power systems: Effective in solving mixed-integer and multi-objective problems in power systems.
• Machine learning and data mining: Balances exploration and exploitation well, avoiding overfitting and local optima.
• Bioinformatics: Can handle noisy, high-dimensional biological datasets and search for global optima.
• Logistics and supply chain optimization: Finds near-optimal solutions in large-scale combinatorics problems.
• Robotics & path planning: Mimics natural group intelligence for path finding and coordination.
• Cyber security: Works well in discrete spaces and can explore vast search landscapes.
• IoT and sensor networks: Adaptive behavior helps with dynamic environments and topology changes.

References

1. Xu, W., Wang, Y., Xu, P., Qiu, T., Yan, T., Wang, Z. (2025). Wolf Pack Algorithm: An Overview. In: Lan, X., Mei, X., Jiang, C., Zhao, F., Tian, Z. (eds) Intelligent Robotics and Applications. ICIRA 2024. Lecture Notes in Computer Science(), vol 15203. Springer, Singapore. https://doi.org/10.1007/978-981-96-0795-2_8
2. Wu, H. S., Zhang, F. M., & Wu, L. S. (2013). New swarm intelligence algorithm: Wolf Pack Algorithm. Systems Engineering and Electronics, 35(11), 2430–2438 
3. Lai et al. (2021). Solving No‑Wait Flow Shop Scheduling Problem Based on Discrete Wolf Pack Algorithm, Scientific Programming, 2021:473101