DESIGN OF FAIRNESS AWARE DEEP REINFORCEMENT LEARNING-BASED GF-NOMA FOR INTERNET OF THINGS
Project no. 21820 · Status: Kapanmış · 2020
Summary
<p>Next generation networks have special areas related with the Internet</p><p>of Things (IoT) to improve the performance of cellular networks in terms</p><p>of throughput. Grant-free non-orthogonal multiple access (GF-NOMA) seems</p><p>a feasible solution, letting machine type communication (MTC) devices</p><p>transmits their packets by using the superimposing signals with different power</p><p>levels over the same time and frequency resources. However, the main drawback</p><p>of GF-NOMA is the high probability of collisions while selection of power level</p><p>of MTC devices.</p><p>In 6G the intelligence should be met to multiple access to solve the GFNOMA</p><p>issues that should be between the two extremes of randomness and</p><p>fully coordinated medium access. Deep-Q-Network (DQN) has become a very</p><p>hot research topic in recent years that let the MTC devices to make a smart</p><p>decision in an intelligent way to improve the throughput.</p><p>A novel learning framework for power domain GF-NOMA is proposed in</p><p>this thesis. The goal of learning framework is to maximize the throughput</p><p>considering fairness in power consumption which provides long-life to the</p><p>IoT network. The learning algorithm push the MTC devices to exchange the</p><p>resources between each other over time. The results show that the proposed</p><p>method outperform the NOMA scheme with random selection in terms of</p><p>throughput and increase the fairness index when the DQN with selfish behavior</p><p>is employed.</p><p>Keywords: Internet of Things, Random Access, GF-NOMA, Machine Learning.</p>