Saving energy in Wireless Sensor Networks (WSNs), is critical in different applications, such as environment monitoring, keeping human awareness and etc. Many studies have investigated energy consumption and improved the WSN lifetime longevity by reducing the energy consumption. Still, proposed approaches overlook the nodes’ distribution role in energy model and routing protocol, which is a key factor in a WSN. In this work, we propose a novel approach; namely GDECA; which assumes nodes’ distributions are mixtures of Gaussian distribution, as an assumption applied in real world. So GDECA rely on a distribution estimation borrowed from Machine Learning (ML) to fit the Gaussian Mixture Model (GMM) to the nodes and calculate the parameters for these distributions. Next, the estimated parameters are employed in Cluster Head CH selection policy. Besides, sinks routing is determined based on nodes distribution. Results showed the improvement close to 40–50% in energy consumption. As another outcome, GDECA keeps all the nodes active until end of the simulation. Observations also demonstrate that sinks path calculation using this approach is optimum, and randomly changing number of sinks increases energy consumption.

在无线传感器网络（WSN）中节省能源，对于诸如环境监测，保持人类意识等不同应用至关重要。许多研究已经调查了能耗，并通过降低能耗来提高WSN的使用寿命。仍然，所提出的方法忽略了节点在能量模型和路由协议中的分布角色，这是WSN中的关键因素。在这项工作中，我们提出了一种新颖的方法。即GDECA ; 它假设节点的分布是高斯分布的混合，这是在现实世界中应用的一个假设。所以GDECA依赖于从机器学习（ML）借用的分布估计，以将高斯混合模型（GMM）拟合到节点并计算这些分布的参数。接下来，在簇头CH选择策略中采用估计的参数。此外，宿路由是基于节点分布确定的。结果表明，能耗降低了近40–50％。另一个结果是，GDECA保持所有节点处于活动状态，直到模拟结束。观察结果还表明，使用此方法的汇槽路径计算是最佳的，并且汇槽数量的随机更改会增加能耗。