在第五代行動通訊技術中，D2D 通訊 (Device-to-Device communications) 因能減輕基地台的負荷並顯著提升頻譜效率而受到重視，更被視為能有效滿足三大應用場景之一的巨量物聯網通訊 (mMTC) 之關鍵技術。在 D2D 通訊中，鄰近的裝置不須透過基地台的協助即可直接通訊，並被允許與 CUE (cellular user equipment) 共用相同的頻譜資源，藉此提升頻譜效率以及服務裝置數量。然而，共用頻譜資源會使得裝置之間彼此互相干擾，因此頻譜資源分配以及傳輸功率控制顯然成為 D2D 通訊中相當重要的研究議題。本篇論文將探討該如何在無線通訊系統中設計一資源分配演算法，使得系統能源效率能夠最大化，並確保所有裝置都能滿足最低傳輸速率之要求。由於上述問題是一個 NP-hard 問題，難以得到最佳解。為了有效地解決該問題，我們提出一種迭代演算法，該演算法利用一系列的凸優化技巧將原始問題近似成凸優化問題 (convex optimization problem)，並在每次迭代中，透過一些現有的優化工具來幫助我們解決近似問題，從而逐步提高系統能源效率。基於上述演算法獲得的局部最佳資源分配結果，我們進一步訓練了具有空間金字塔池化層 (spatial pyramid pooling layer) 的卷積神經網絡 (CNN)，藉此降低迭代演算法之計算複雜度。實驗結果表明，我們提出之具有空間金字塔池化層的卷積神經網絡架構不僅在系統能源效率以及系統傳輸速率上優於其他種神經網路，甚至達到與迭代演算法差不多的效能，並且能大幅降低該演算法所需的計算時間。

In this thesis, we introduce the application of Device-to-Device (D2D) communication into the scenario of massive Machine Type Communication (mMTC). More specifically, we formulate the channel allocation and power control problem aiming at maximizing the system energy efficiency under the constraints of minimum rate requirements and power budget limitations of cellular users and D2D pairs. However, the formulated resource allocation problem is NP-hard, which is difficult to obtain the optimal solution directly. To solve the problem efficiently, we propose an iterative algorithm that utilizes convex approximation techniques to approximate the original problem as a geometric programming problem. In each iteration, we solve the approximated problem by some off-the-shelf optimization tools (e.g., CVX) to maximize the system energy efficiency progressively. Based on the sub-optimal resource allocation results, a convolutional neural network (CNN) with spatial pyramid pooling layer is constructed to obtain the decisions on resource allocation, thereby reducing the computational time. The simulation results demonstrate that the proposed CNN outperforms the other neural networks in terms of system energy efficiency and system sum rate, even achieves similar performance as the iterative algorithm with ultra-low CPU runtime.

[1] F. Hussain, M. Y. Hassan, M. S. Hossen, and S. Choudhury, “An optimal resource allocation algorithm for d2d communication underlaying cellular networks,” in 2017 14th IEEE Annual Consumer Communications Networking Conference (CCNC), 2017, pp. 867–872.
[2] M. Liu, L. Zhang, and Y. You, “Joint power and channel allocation for underlay d2d communications with proportional fairness,” in 2019 15th International Wireless Communications Mobile Computing Conference (IWCMC), 2019, pp. 1333–1338.
[3] S. Liu, Y. Wu, L. Li, X. Liu, and W. Xu, “A two-stage energy-efficient approach for joint power control and channel allocation in d2d communication,” IEEE Access, vol. 7, pp. 16 940–16 951, 2019.
[4] F. Idris, J. Tang, and D. K. C. So, “Resource and energy efficient device to device communications in downlink cellular system,” in 2018 IEEE Wireless Communications and Networking Conference (WCNC), 2018, pp. 1–6.
[5] Q. Mao, F. Hu, and Q. Hao, “Deep learning for intelligent wireless networks: A comprehensive survey,” IEEE Communications Surveys Tutorials, vol. 20,no. 4, pp. 2595–2621, 2018.
[6] D. Xu, X. Chen, C. Wu, S. Zhang, S. Xu, and S. Cao, “Energy-efficient subchannel and power allocation for hetnets based on convolutional neural network,” in 2019 IEEE 89th Vehicular Technology Conference (VTC2019-Spring), 2019, pp. 1–5.
[7] H. Sun, D. Zhai, Z. Zhang, J. Du, and Z. Ding, “Channel allocation and power control for device-to-device communications underlaying cellular networks incorporated with non-orthogonal multiple access,” IEEE Access, vol. 7,pp. 168 593–168 605, 2019.
[8] X. Liu, “Optimisation of the duplex d2d network: A deep learning approach,” IET Networks, vol. 9, no. 3, pp. 139–144, 2020.
[9] K. K. Nguyen, T. Q. Duong, N. A. Vien, N. Le-Khac, and M. Nguyen, “Non-cooperative energy efficient power allocation game in d2d communication: A multi-agent deep reinforcement learning approach,” IEEE Access, vol. 7,pp. 100 480–100 490, 2019.
[10] A. Asheralieva and Y. Miyanaga, “An autonomous learning-based algorithm for joint channel and power level selection by d2d pairs in heterogeneous cellular networks,” IEEE Transactions on Communications, vol. 64, no. 9,pp. 3996–4012, 2016.
[11] S. Nie, Z. Fan, M. Zhao, X. Gu, and L. Zhang, “Q-learning based power control algorithm for d2d communication,” in 2016 IEEE 27th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), 2016, pp. 1–6.
[12] Z. Fan, X. Gu, S. Nie, and M. Chen, “D2d power control based on supervised and unsupervised learning,” in 2017 3rd IEEE International Conference on Computer and Communications (ICCC), 2017, pp. 558–563.
[13] W. Dinkelbach, “On nonlinear fractional programming, Management science, vol. 13, no. 7, pp. 492–498, Mar. 1967.
[14] S. Boyd, S.-J. Kim, L. Vandenberghe, and A. Hassibi, “A tutorial on geometric programming,” Optimization and engineering, vol. 8, no. 1, p. 67, 2007.
[15] S. Boyd, S. P. Boyd, and L. Vandenberghe, Convex optimization. Cambridge university press, 2004.
[16] C.-Y. Chi, W.-C. Li, and C.-H. Lin, Convex optimization for signal processing and communications: from fundamentals to applications. CRC press, 2017.
[17] A. Khan, A. Sohail, U. Zahoora, and A. S. Qureshi, “A survey of the recent architectures of deep convolutional neural networks,” Artificial Intelligence Review, pp. 1–62, 2019.
[18] K. He, X. Zhang, S. Ren, and J. Sun, “Spatial pyramid pooling in deep convolutional networks for visual recognition,” IEEE transactions on pattern analysis and machine intelligence, vol. 37, no. 9, pp. 1904–1916, 2015.
[19] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014.
[20] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” in Advances in neural information processing systems, 2012, pp. 1097–1105.