The article focuses on addressing the issue of data protection in the context of dynamic network topology and limited resources. In modern systems of autonomous unmanned platforms, the key task is to ensure reliable, secure, and energy-efficient information exchange between agents in conditions of constant changes in the swarm structure. The approaches proposed in the article include the use of lightweight cryptographic algorithms SIMON and SPECK, which provide minimal data transmission delays, low power consumption, and high resistance to attacks at the interception and modification level. The Q-learning algorithm, which allows agents to quickly adapt to changes in network topology, is discussed. Simulations conducted using the NS-3 platform demonstrated the advantage of intelligent approaches based on self-learning and cooperative decision-making methods in ensuring high system performance with minimal energy consumption and rapid adaptation to environmental changes. Security assessments confirmed the system's resilience to routing and data interception attacks, making these methods promising for further use in autonomous unmanned platforms