Caching Placement and Resource Allocation for Cache-Enabling UAV NOMA Networks

Abstract

This article investigates the cache-enabling unmanned aerial vehicle (UAV)\ncellular networks with massive access capability supported by non-orthogonal\nmultiple access (NOMA). The delivery of a large volume of multimedia contents\nfor ground users is assisted by a mobile UAV base station, which caches some\npopular contents for wireless backhaul link traffic offloading. In\ncache-enabling UAV NOMA networks, the caching placement of content caching\nphase and radio resource allocation of content delivery phase are crucial for\nnetwork performance. To cope with the dynamic UAV locations and content\nrequests in practical scenarios, we formulate the long-term caching placement\nand resource allocation optimization problem for content delivery delay\nminimization as a Markov decision process (MDP). The UAV acts as an agent to\ntake actions for caching placement and resource allocation, which includes the\nuser scheduling of content requests and the power allocation of NOMA users. In\norder to tackle the MDP, we propose a Q-learning based caching placement and\nresource allocation algorithm, where the UAV learns and selects action with\n\\emph{soft ${\\varepsilon}$-greedy} strategy to search for the optimal match\nbetween actions and states. Since the action-state table size of Q-learning\ngrows with the number of states in the dynamic networks, we propose a function\napproximation based algorithm with combination of stochastic gradient descent\nand deep neural networks, which is suitable for large-scale networks. Finally,\nthe numerical results show that the proposed algorithms provide considerable\nperformance compared to benchmark algorithms, and obtain a trade-off between\nnetwork performance and calculation complexity.\n