Latest technologies with fifth-generation eventually networks that are wireless, SWIPT, and MC-NOMA, seem to be included in this research. Hence, the MC-NOMA system is analyzed and used along with connection with moreover Pattern Division Multiple Access (PDMA) techniques such as enabling SWIPT for the joint time switching (TS) centered receivers downlink resource allocation issue. To use deep learning, a model seems to be the Deep Belief Network (DBN) is used, developed approach's comprehensive process is divided into three parts: the data preparation, the training, and the running. Further, similarly, outcomes also affirm whether the SC-NOMA underperforms in comparison with MC-NOMA along by using SCMA instead of PDMA because of the Power consumption of SWIPT-enabled systems.