在第五代通訊系統（5th Generation of mobile communication system, 5G）中有非常多的通訊方式，如多輸入多輸出通訊系統（Multi-Input Multi-Output, MIMO），亦或是不同種的波形如：Universal Filtered Multi-Carrier（UFMC）、Generalized Frequency Division Multiplexing（GFDM）、Filter Bank Multi-Carrier（FBMC），上述所提到之波形皆為5G通訊系統的候選波形。在5G通訊系統因應著高傳輸速度、高可靠度、超低延遲的要求下，隨之而來的是眾多種應用，例如物聯網、Machine Type Communication（MTC）等等。
隨著通訊用戶數量的上升，基地台變得擁塞，因此在第五代通訊系統中提出裝置對裝置通訊（Device to Device, D2D）的概念，使相同蜂巢網路下的用戶直接連線，而非藉由基地台來轉送，試圖使基地台的流量下降，達到卸載目的。然而D2D通訊中的用戶使用相同頻道，使得裝置間互相干擾。如何控制所有裝置之傳輸的功率顯得非常的重要。因此本論文將會研究利用深度神經網路的方法來控制傳輸功率使蜂巢網路（Cellular Network）內的通道容量（Channel Capacity）最大化。然而，在訓練神經網路的過程中發現神經網路只單純的提升通道容量，捨棄眾多用戶的服務品質（Quality of Service, QoS）。因此，本篇論文提出了不同的優化目標函數，取得通道容量及服務品質之間的平衡。
最後，將會在第四章做出通道容量、中斷機率、能量效率、運算時間以及覆蓋機率五種性能指標的模擬，藉由模擬結果分析並比較各種演算法之間的差異。

In the 5th Generation of mobile communication system (5G), there are many communication scenarios such as Multi-Input Multi-Output (MIMO). Or the waveforms such as Universal Filtered Multi-Carrier (UFMC), Generalized Frequency Division Multiplexing (GFDM), Filter Bank Multi-Carrier (FBMC). The waveforms mentioned above are all candidate waveforms for the 5G communication system. In response to the requirements of high transmission speed, high reliability, and ultra-low latency, 5G communication systems are accompanied by numerous applications, such as the Internet of Things, Machine Type Communication (MTC), etc.
As the number of communication users increases, base stations become congested. Therefore, a scenario called device-to-device (D2D) is proposed in the 5th generation communication system to allow users to connect directly under the same cellular network instead of forwarding through the base station. Trying to offload the traffic of the base station. However, users with D2D communication using the same channel. It causes more interference from other devices. So how to properly control the power transmitted by all users is very important.
Therefore, this paper will study the method of deep neural network to control the transmission power to maximize the channel capacity in the cellular network. However, in the process of training the neural network, it was found that the neural network only increased the channel capacity, but sacrificed the Quality of Service (QoS) of many users. For the reason, this paper proposes a new optimization objective function to find a balance between channel capacity and quality of service.
Finally, in Chapter 4, we will simulate five performance, including channel capacity, outage probability, energy efficiency (EE), execution time and coverage probability. Based on those results, we will analyze and compare the differences between various algorithms.

[1] M. Shafi et al., "5G: A Tutorial Overview of Standards, Trials, Challenges, Deployment, and Practice," IEEE Journal on Selected Areas in Communications, vol. 35, no. 6, pp. 1201-1221, 2017, doi: 10.1109/JSAC.2017.2692307.
[2] F. Jameel, Z. Hamid, F. Jabeen, S. Zeadally, and M. A. Javed, "A Survey of Device-to-Device Communications: Research Issues and Challenges," IEEE Communications Surveys & Tutorials, vol. 20, no. 3, pp. 2133-2168, 2018, doi: 10.1109/COMST.2018.2828120.
[3] A. Asadi, Q. Wang, and V. Mancuso, "A Survey on Device-to-Device Communication in Cellular Networks," IEEE Communications Surveys & Tutorials, vol. 16, no. 4, pp. 1801-1819, 2014, doi: 10.1109/COMST.2014.2319555.
[4] M. Klügel and W. Kellerer, "The Device-to-Device Reuse Maximization Problem With Power Control," IEEE Transactions on Wireless Communications, vol. 17, no. 3, pp. 1836-1848, 2018, doi: 10.1109/TWC.2017.2785818.
[5] W. Lee, M. Kim, and D. Cho, "Deep Learning Based Transmit Power Control in Underlaid Device-to-Device Communication," IEEE Systems Journal, vol. 13, no. 3, pp. 2551-2554, 2019, doi: 10.1109/JSYST.2018.2870483.
[6] B. Fang, Z. Qian, W. Zhong, W. Shao, and H. Xue, "Coordinated precoding for D2D communications underlay uplink MIMO cellular networks," in 2015 IEEE/CIC International Conference on Communications in China (ICCC), 2-4 Nov. 2015 2015, pp. 1-5, doi: 10.1109/ICCChina.2015.7448661.
[7] N. Lee, X. Lin, J. G. Andrews, and R. W. Heath, "Power Control for D2D Underlaid Cellular Networks: Modeling, Algorithms, and Analysis," IEEE Journal on Selected Areas in Communications, vol. 33, no. 1, pp. 1-13, 2015, doi: 10.1109/JSAC.2014.2369612.
[8] Q. Shi, M. Razaviyayn, Z. Luo, and C. He, "An iteratively weighted MMSE approach to distributed sum-utility maximization for a MIMO interfering broadcast channel," in 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 22-27 May 2011 2011, pp. 3060-3063, doi: 10.1109/ICASSP.2011.5946304.
[9] H. Sun, X. Chen, Q. Shi, M. Hong, X. Fu, and N. D. Sidiropoulos, "Learning to optimize: Training deep neural networks for wireless resource management," in 2017 IEEE 18th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), 3-6 July 2017 2017, pp. 1-6, doi: 10.1109/SPAWC.2017.8227766.
[10] R. Hecht-Nielsen, "Theory of the backpropagation neural network," in Neural networks for perception: Elsevier, 1992, pp. 65-93.
[11] S. Gengtian, T. Koshimizu, M. Saito, P. Zhenni, L. Jiang, and S. Shimamoto, "Power Control Based on Multi-Agent Deep Q Network for D2D Communication," in 2020 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), 19-21 Feb. 2020 2020, pp. 257-261, doi: 10.1109/ICAIIC48513.2020.9065192.
[12] F. Jiang, Z. Yuan, C. Sun, and J. Wang, "Deep Q-Learning-Based Content Caching With Update Strategy for Fog Radio Access Networks," IEEE Access, vol. 7, pp. 97505-97514, 2019, doi: 10.1109/ACCESS.2019.2927836.
[13] F. Meng, P. Chen, and L. Wu, "Power Allocation in Multi-User Cellular Networks with Deep Q Learning Approach," in ICC 2019 - 2019 IEEE International Conference on Communications (ICC), 20-24 May 2019 2019, pp. 1-6, doi: 10.1109/ICC.2019.8761431.
[14] 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), 4-8 Sept. 2016 2016, pp. 1-6, doi: 10.1109/PIMRC.2016.7794793.
[15] S. K. Panda and P. K. Jana, "An efficient task scheduling algorithm for heterogeneous multi-cloud environment," in 2014 International Conference on Advances in Computing, Communications and Informatics (ICACCI), 24-27 Sept. 2014 2014, pp. 1204-1209, doi: 10.1109/ICACCI.2014.6968253.
[16] S. K. Panda and P. K. Jana, "Efficient task scheduling algorithms for heterogeneous multi-cloud environment," The Journal of Supercomputing, vol. 71, no. 4, pp. 1505-1533, 2015/04/01 2015, doi: 10.1007/s11227-014-1376-6.
[17] S. K. Panda and P. K. Jana, "A multi-objective task scheduling algorithm for heterogeneous multi-cloud environment," in 2015 International Conference on Electronic Design, Computer Networks & Automated Verification (EDCAV), 29-30 Jan. 2015 2015, pp. 82-87, doi: 10.1109/EDCAV.2015.7060544.