@Article{cmc.2026.085627,
AUTHOR = {Taha Bachir Ammour, Mohammed Kaddi, Mohammed Omari},
TITLE = {Longevity-Aware Crayfish Optimization (LACO) for Lifetime Maximization in Wireless Sensor Networks},
JOURNAL = {Computers, Materials \& Continua},
VOLUME = {},
YEAR = {},
NUMBER = {},
PAGES = {{pages}},
URL = {http://www.techscience.com/cmc/online/detail/28325},
ISSN = {1546-2226},
ABSTRACT = {Wireless Sensor Networks (WSNs) are the backbone of modern Internet of Things (IoT) deployments, but they share a critical bottleneck: limited battery life. Because of this, picking the right Cluster Heads (CHs) in an energy-aware way is absolutely essential for keeping the network alive. To tackle this challenge, this paper introduces the Longevity-Aware Crayfish Optimization (LACO) algorithm. It is a targeted upgrade to the standard Crayfish Optimization Algorithm (COA) that bakes explicit energy awareness directly into the clustering process. At its core, LACO relies on a three-layer Energy-State Adaptive Control (ESAC) mechanism. First, it dynamically balances exploration and exploitation based on the network’s remaining energy. Second, it uses a longevity-biased selection strategy paired with a time-decaying step size to stabilize the search process. Finally, a strict merit-based filter prevents nearly drained nodes from taking on demanding leadership roles. We put LACO to the test through extensive simulations, varying network densities from 100 to 500 nodes and adjusting initial energy levels between 0.75 and 2.0 J. The algorithm consistently outperformed classical protocols like LEACH, PEGASIS, and the baseline COA, as well as recent state-of-the-art methods. Specifically, LACO immediately proved its early-stage stability by boosting the First Node Death lifespan by 15.1 percent, before going on to increase in Half Node Death by a massive 66.48 percent and extend the Last Node Death by 9.66 percent. In high-energy scenarios, the protocol reached an impressive 9747 operational rounds, yielding an efficiency of 4783 rounds per Joule. This represents a performance leap of up to 49 percent over recent hybrid metaheuristics, with all improvements rigorously confirmed as statistically significant using the paired Wilcoxon Signed-Rank Test (<math id="mml-ieqn-1"><mi>p</mi><mo>&lt;</mo><mn>0.05</mn></math>, effect size <math id="mml-ieqn-2"><mi>r</mi><mo>≥</mo><mn>0.392</mn></math>).},
DOI = {10.32604/cmc.2026.085627}
}