Securing device-to-device (D2D) communications has become crucial to ensuring dependable and trusted interactions with the proliferation of ultra-dense 5G communication scenarios. This paper suggests a novel method called Deep Q Blockchain Networks (DQBN) to address the need for an enhanced security framework. The development of the suggested model was motivated by the identification of the shortcomings of current 5G blockchain models in efficiently enhancing communication security. The DQBN model makes use of distributed Deep Q Learning to enhance the security of D2D communications by optimizing the choice of miner nodes. To determine the best miner nodes, it uses an analysis of the performance levels of temporal block mining, temporal throughput, temporal energy, spatial distance, residual energy, and temporal delay. The model also uses Q Learning to optimize the length of the blockchain, which enhances Quality of Service (QoS) in terms of latency, throughput, Packet Delivery Ratio (PDR), and energy usage levels. The proposed DQBN model outperforms current 5G blockchain models in terms of performance. According to the results of the experiments, there was a 5.9% decrease in delay, a 4.5% increase in throughput, a 2.4% improvement in PDR, and an 8.5% decrease in energy consumption. These developments help the model become more resilient to different kinds of attacks, such as Distributed Denial of Service (DDoS), Man-in-the-Middle (MITM), Sybil, and Spoofing attacks. In terms of security for D2D communications in extremely dense 5G communication scenarios, the proposed DQBN model represents a significant advancement. Its use of distributed Deep Q Learning, along with the optimization of miner nodes and blockchain length, leads to appreciable gains in QoS and attack resilience, outperforming current 5G blockchain models.