Routing and scheduling techniques in NOMA-based multiple access

In wireless communications, user nodes in networks require techniques to transfer a high load of packets to each other in a limited time frame. Non-orthogonal Multiple Access (NOMA) and Successive Interference Cancellation (SIC) are such techniques. In addition to that, Q-learning is a new method of machine learning that can be used to analyze the success communication planning and transfer of packets. According to the simulation results, integrating Q-learning has increased the number of transferred packets the most. Scheduling and multi-path routing for system output optimization has been studied for wireless networks. According to the literature reviews, our model differs from others by applying NOMA (Non-orthogonal Multiple Access) at the transmitters and SIC (Successive Interference Cancellation) at the receviers. We assume that the transmission powers are constant. NOMA (one transmitter, two receivers), SIC (two transmitters, one receiver) and single links (one transmitter and one receiver) are formed. Links have been scheduled to optimize the weighted sum of end to end flow rate. The improvement in performance has been tested by the integration of NOMA and SIC to the system. Numerical results show that NOMA and SIC improves the system wieghted sum rate by 25-30%. A case where the source of all flows is the same has been investigated. In this scenario, NOMA has provided 65-70% improvement in performance whereas SIC has shown no effect on the performance. In a distributed ad hoc network where SIC is integrated, joint optimal routing and minimum length scheduling is studied. Optimal solution finds minimum amount of time slots required to schedule transmissions at a given flow rate in an SIC network. The performance of the optimal solution with SIC is compared to the one without SIC. According to the results, the optimal solution using SIC shows a significiant improvement compared to the one that does not use SIC.