MANET (mobile ad hoc network) comprises of a set of wireless mobile node connected in a self-healing and self-configured network devoid of any fixed infrastructure.The cognitive radio (CR) in the MANET system has been developed for addressing the spectrum congestion issue. However, the existing techniques failed to solves the complex convex problem of channel assignment. To address this issue, the proposed protocol is designed. In the proposed scheme, both primary user (PU) and secondary user (SU) are clustered initially based on the graph theory process. After that, in each cluster, formation of robust spatial Gabriel graph (RS-GG) takes place at which the neighboring nodes are predicted by estimating the weighted end-to-end delay. Once the multi path decision-making condition is satisfied, the route path is established, and the communication takes place based on QoS constraint. This can lead to the enhancement of PDR, network connectivity maintenance and network lifetime. The performance analysis of the proposed methods is carried out for PDR, control overheads, energy consumption, and End-to-End delay and the analysis is compared with existing protocols to prove the effectiveness of proposed design. The simulation outcomes illustrate that the suggested strategy performs well and improves the data transmission.