近幾年對於第五代行動通訊(5G)的標準規格，國際組織3GPP持續的在進行訂制，5G可以分成三大應用場景，分別是增強型行動寬頻通訊(Enhanced Mobile Broadband, eMBB)，超可靠度和低延遲通訊(Ultra-reliable and Low Latency Communications, URLLC)，以及大規模機器型通訊(Massive Machine Type Communications, mMTC)，其中URLLC為了將來能應用在工業自動化生產、無人駕駛等等，極高的可靠度(high reliability)與極低的時間延遲(low latency)便是他所追求的目標，然而如此嚴格的需求將對現有網路通訊系統帶來巨大的挑戰，因此本篇論文將焦點放在URLLC的資源分配上。
對於要上行的突發性URLLC流量採用無允諾上行(Uplink Grant-free)方式可以有效減少用戶設備(User Equipment, UE)與基地台(Base Station, BS)之間的授權延遲，而一般基地台會分配給URLLC的UE專用資源(dedicated resource)或共享資源池(shared resource pool)進行上行，無論使用哪種資源皆各有優缺點存在，因此本篇論文將依據URLLC流量特性對UE分組，在頻寬有限的情況下做最有效益的資源配置，並搭配強化學習的方式決定共享資源池的大小。從模擬結果可以看出，在各種情況下皆能有效節省資源，同時達到高可靠度與低延遲的目標，尤其是高負載的情況更能體現優勢。

In recent years, the international organization 3GPP has continued to customize the standard specifications of the fifth generation mobile communication (5G). 5G can be divided into three major application scenarios, namely Enhanced Mobile Broadband (eMBB), Ultra-reliable and Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC). URLLC can be used in industrial automation production, unmanned driving, etc. in the future. The purposes of URLLC is high reliability and low latency. However, such strict requirements will bring great challenges to existing network communication systems, so this paper will focus on the resource allocation of URLLC.
For the burst URLLC traffic to be uplinked, the Uplink Grant-free method can effectively reduce the request delay between the UE and the BS. BS may allocate the dedicated resource or shared resource pool to URLLC UE for uplink. No matter which resource is used, there are strength and weaknesses. Therefore, this paper will group UEs according to the traffic characteristics. Under limited bandwidth, it will make the most effective resource allocation, and use reinforcement learning to determine the size of the shared resource pool. From the simulation results, it shows that the resources can be effectively saved in various situations, and achieve the goals of high reliability and low latency at the same time, especially in high load case.

[1] ITU, "IMT Vision – Framework and overall objectives of the future development of IMT for 2020 and beyond," 09 2015. [Online]. Available: https://www.itu.int/dms_pubrec/itu-r/rec/m/R-REC-M.2083-0-201509-I!!PDF-E.pdf. [Accessed 10 06 2019].
[2] 3GPP TR 38.913, “Study on scenarios and requirements for next generation access technologies,” 3GPP Tech. Rep., V15.0.0, Jul. 2018.
[3] 3GPP TR 38.211, “NR; Physical channels and modulation,” 3GPP Tech. Rep., V16.0.0, Dec. 2019.
[4] [Online]. Available: https://www.sharetechnote.com/html/5G/5G_FrameStructure.html.
[5] 3GPP TR 38.912, “Study on New Radio (NR) access technology,” 3GPP Tech. Rep., V15.0.0, Jul. 2018.
[6] [Online]. Available: http://howltestuffworks.blogspot.com/2019/11/5g-nr-time-domain-slots-and-slot-formats.html.
[7] 3GPP TS 38.213, “NR; Physical layer procedures for control,” 3GPP Tech. Rep., V16.1.0, Apr. 2020.
[8] [Online]. Available: https://blogs.keysight.com/blogs/inds.entry.html/2018/09/07/5g_flexible_numerolo-XHto.html.
[9] 3GPP TS 38.331, “NR; Radio Resource Control (RRC); Protocol specification,” 3GPP Tech. Rep., V16.0.0, Apr. 2020.
[10] [Online]. Available: http://www.techplayon.com/5g-nr-grant-free-dynamic-scheduling-transmission-without-grant-twg/.
[11] Azad Ravanshid et al., “Multi-connectivity functional architectures in 5g,” in 2016 IEEE International Conference on Communications Workshops (ICC), May, 2016.
[12] [Online]. Available: https://www.accton.com/Technology-Brief/the-emergence-of-5g-mmwave/.
[13] [Online]. Available: https://blogs.nvidia.com.tw/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/.
[14] [Online]. Available: https://blogs.oracle.com/datascience/reinforcement-learning-deep-q-networks.
[15] [Online]. Available: https://medium.com/%E9%9B%9E%E9%9B%9E%E8%88%87%E5%85%94%E5%85%94%E7%9A%84%E5%B7%A5%E7%A8%8B%E4%B8%96%E7%95%8C/%E6%A9%9F%E5%99%A8%E5%AD%B8%E7%BF%92-ml-note-reinforcement-learning-%E5%BC%B7%E5%8C%96%E5%AD%B8%E7%BF%92-dqn-%E5%AF%A6%E4%BD%9Catari-game-7f9185f.
[16] 3GPP TS 22.104, “Service requirements for cyber-physical control applications in vertical domains,” 3GPP Tech. Rep., V17.1.0, Sep. 2019.
[17] R1-1705463, “UL grant-free transmission for URLLC,” 3GPP TSGRAN WG1 #88, Apr. 2017.
[18] Zhiyi Zhou, Rapeepat Ratasuk, Nitin Mangalvedhe, and Amitava Ghosh, "Resource Allocation for Uplink Grant-Free Ultra-Reliable and Low Latency Communications," IEEE 87th VTC Spring, 2018.
[19] Salah Eddine Elayoubi, Patrick Brown, Matha Deghel, and Ana Galindo-Serrano, "Radio Resource Allocation and Retransmission Schemes for URLLC Over 5G Networks," IEEE Journal on Selected Areas in Communications, April, 2019.
[20] Nurul Huda Mahmood et al., "Uplink Grant-Free Access Solutions for URLLC services in 5G New Radio," International Symposium on Wireless Communication Systems, 2019.
[21] Luca Buccheri et al., "Hybrid Retransmission Scheme for QoS-defined 5G Ultra-Reliable Low-Latency Communications," IEEE Wireless Communications and Networking Conference, 2018.
[22] R1-1705654, “UL grant-free transmission for URLLC,” 3GPP TSG RAN WG1 #88, Apr. 2017.
[23] S. Xing, et.al, “Advanced Grant-free Transmission for Small Packets URLLC Services”, International Conference on Communications Workshops (ICC Workshops), 2019.
[24] 3GPP TS 38.212, “NR; Multiplexing and channel coding,” 3GPP Tech. Rep., V16.1.0, Mar. 2020.