2162-2337 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LWC.2019.2943471, IEEE Wireless Communications Letters 1 Three-Stage DRX Scheduling for Joint Downlink Transmission in C-RAN Jia-Ming Liang, Ching-Kuo Hsu, Jen-Jee Chen, Kun-Ru Wu, and Yu-Chee Tseng Abstract---Cloud-RAN (C-RAN) is a promising network architec- ture to provide 5G broadband services. In C-RAN, the base stations (BSs) are separated as the computation entities (i.e., Baseband Units (BBUs)) and the radio entities (i.e., Remote Radio Heads (RRHs)), where the BBUs are put in a centralized cloud to realize centrally control and the RRH are left at cell sites for signal transmission. With C-RAN, multiple collaborative cells can transmit data to user equipments (UEs) by Joint Transmission (JT) technology to improve network efficiency. On the other hand, 3GPP standard has defined the Discontinuous Reception (DRX) mechanism, which allows UEs to turn off their radio interfaces periodically to save their energy. However, how to cooperate DRX with JT under the C-RAN architecture is still an open issue. Therefore, this paper studies how to optimize UEs’ DRX parameters while considering their quality-of- service (QoS) in C-RAN with JT. To solve this problem, we propose an energy-efficient JT scheduling scheme. The main idea of our scheme is to well pair UEs and BSs with the consideration of joint transmission quality and then optimizes DRX parameters to further save energy. Extensive simulation results show that our scheme can improve system throughput, well utilize resource and save UEs’ energy consumption. Index Terms---5G mobile system, cloud radio access networks (C- RAN), joint transmission (JT), discontinuous reception (DRX), sleep scheduling. I. I NTRODUCTION Cloud Radio Access Networks (C-RAN) is an emerging net- work architecture for 5G communications. With a centralized network architecture, the computation entities, i.e., Baseband Units (BBUs), are separated from base stations (BSs), and put in a centralized cloud, and leaves Remote Radio Heads (RRHs) at cell sites. By this way, the radio resources of different RRHs can be centrally managed and multiple cells can be collaborated to improve network efficiency by the Joint Transmission (JT) technology [1]. On the other hand, 3GPP (the 3rd Generation Partnership Project) has defined the Discontinuous Reception (DRX) mechanism, which allows UEs to turn off their radio interfaces when there is no data transmitted from evolved Node B (eNB). However, how to optimize DRX parameters to minimize UEs’ energy under the C-RAN architecture with JT is still an open issue. Thus, we address the DRX optimization problem under UEs’ QoS constraints in C-RAN with JT. The objective is to schedule JT transmissions to improve network efficiency while minimizing UEs’ wake-up periods to save energy under consideration of UEs’ QoS requirements in terms of packet delay Jia-Ming Liang is with the Department of Computer Science and Information Engineering, Chang Gung University, Taoyuan 33302, Taiwan, and also with the Department of General Medicine, Chang Gung Memorial Hospital, Taoyuan 33378, Taiwan. E-mail: jmliang@mail.cgu.edu.tw Ching-Kuo Hsu, Kun-Ru Wu, and Yu-Chee Tseng are with the Department of Computer Science, National Chiao Tung University, Hsinchu 30010, Taiwan. E-mail: {HSUck, kunruwu, yctseng}@cs.nctu.edu.tw Jen-Jee Chen is with the College of Artificial Intelligence and Green En- ergy, National Chiao Tung University, Hsinchu 30010, Taiwan. E-mail: jen- jee@nctu.edu.tw Jen-Jee Chen is the corresponding author. and data rate. To solve this problem, we propose an energy- efficient JT scheduling scheme. The key idea of our scheme is to pair the UEs with BSs by two special cost metrics and then optimize the corresponding DRX parameters to save energy. In the literature, references [2], [3] propose dynamic point selection schemes to balance traffic load of RRHs; however, these works do not exploit JT to improve transmission efficiency. References [4], [5] consider C-RAN using JT. A cooperative caching placement for JT with single cell transmission to min- imize UEs’ data transmission time is proposed in [4]. The study [5] presents a beamforming design for Gaussian broadcast channels to maximize spectral efficiency. However, these works [4], [5] do not consider the optimization of throughput and energy simultaneously. Reference [6] presents a JT scheme with a call admission control (CAC) to maximize network throughput. The work [7] proposes a load-aware JT scheme to maximize coverage and ergodic link throughput. The study [8] develops a semi-dynamic cluster scheme to maximize energy efficiency. The work [9] proposes a power-factorizing mechanism to achieve the maximization of edge UEs’ throughput. However, these works [6]--[9] do not consider the DRX mechanism, which can further save UEs’ energy. II. SYSTEM MODEL A. C-RAN and JT Towards 5G, C-RAN is a promising architecture for mobile networks to support collaborative transmissions. C-RAN sepa- rates a traditional base station into two parts: digital function unit known as BBU and radio function unit known as RRH. BBUs are placed in the BBU pool in a cloud and connect with RRHs via optical fibers. It thus centralizes control. In C-RAN, JT is a way to realize collaborative transmissions and enhance data rate of UEs [1]. When a UE locates in the borders of multiple cells, JT can allocate multiple RRHs to serve it simultaneously. In Fig. 1(a), where Cell 1 and Cell 2 together serve UE 1 , UE 2 , and UE 3 ; without cooperative transmission by JT, UE 1 needs to receive data from Cell 1 at time slots 1 ∼ 2 and switches to Cell 2 at time slots 3 ∼ 4. Note that UE 2 and UE 3 are allocated at the same time slots because the transmissions are interference free and the cells are centrally controlled by C-RAN. Thus, it requires 6 subframes to serve all UEs. In Fig. 1(b), with cooperative transmission by JT, UE 1 can receive data from Cell 1 and Cell 2 at time slots simultaneously. It only requires 4 subframes to serve all UEs, thus improving transmission efficiency. B. DRX mechanism The DRX mechanism is designed to enhance downlink en- ergy consumption of UEs. The DRX configurations, which are determined by eNB, are UE-specific. The UE performs wake- up/sleep operations in a periodic cycle when DRX is activated.