The internet of things is a worldwide technological development in communications. Low Power and Lossy Networks (LLN) are a fundamental part of the internet of things with numerous monitoring and controlling applications. This network has many challenges as well, due to limited hardware and communication resources, which causes problems in applications such as routing, connections, data transfer, and aggregation between nodes. The IETF group has provided a routing model for LLN networks, which expands IPv6 protocol based on Routing Protocol (RPL). The pro-posed decision system DDSLA-RPL creates a list of limited k member optimal parents based on qualitatively effective parameters such as hop, link quality, SNR rate, and ETX energy consumption, by informing child nodes of their connection link to available parents. In the routing section, a decision system approach based on learning automata has been proposed to dynamically determine and update the weight of influential parameters in routing. The effective parameters in the routing phase of DDSLA-RPL include battery depletion index, connection delay, and node queuing and throughput. The results of the simulation show that the proposed method outperforms other methods by about 30, 17, 20, 18, and 24 percent in mean longevity and energy efficiency, graph sustainability, operational power and latency, packet delivery rate test, and finally number of control messages test, respectively.