Mobile Ad-hoc Networks (MANETs) consist of mobile nodes that usually have limited energy resources. MANET routing protocols should consider the dynamics and energy constraints of the network, and this makes them an optimization problem. Various optimization-based MANET routing protocols have been proposed in literature and each of them consider different metrics and try to cope with specific problems. In this paper, a novel heterogeneous MANET routing protocol called “learning Automata and Genetic based Ad hoc On-Demand Distance Vector” (AGEN-AODV) is proposed, in which routes are rated based on energy, stability, traffic, and hop-count criteria. The Genetic Algorithm (GA) in conjunction with Learning Automata (LA) is used to select the optimal route. The LA runs concurrent to the GA and initializes, adjusts and optimizes its coefficients based on the network feedback, preventing the GA from divergence or sub-optimal convergence. Compared with related works, the throughput, packet delivery ratio (PDR), delay, network lifetime, and energy consumption are improved by at least 4%, 8%, 8%, 13%, and 30% respectively.

移动自组织网络 (MANET) 由通常具有有限能源资源的移动节点组成。MANET 路由协议应该考虑网络的动态和能量约束，这使得它们成为一个优化问题。文献中已经提出了各种基于优化的 MANET 路由协议，它们中的每一个都考虑了不同的度量并试图解决特定的问题。在本文中，提出了一种新的异构 MANET 路由协议，称为“学习自动机和基于遗传的 Ad hoc 按需距离向量”（AGEN-AODV），其中路由基于能量、稳定性、流量和跳数进行评级标准。遗传算法 (GA) 结合学习自动机 (LA) 用于选择最佳路线。LA 与 GA 并发运行并初始化，根据网络反馈调整和优化其系数，防止 GA 发散或次优收敛。与相关工作相比，吞吐量、数据包传递率（PDR）、延迟、网络寿命和能耗分别提高了至少 4%、8%、8%、13% 和 30%。