0018-9545 (c) 2013 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/TVT.2014.2376038, IEEE Transactions on Vehicular Technology 1 Low-Complexity Power-Efficient Schedulers for LTE Uplink with Delay-Sensitive Traffic M. Kalil * , A. Shami * , A. Al-Dweik *† , and S. Muhaidat ‡ * Western University, London, Ontario, Canada, emails: {mkalil3, ashami2, aaldweik}@uwo.ca † Department of Electrical and Computer Engineering, Khalifa University, Abu Dhabi, UAE. emails: dweik@kustar.ac.ae, muhaidat@ieee.org Abstract—This work investigates power-efficient scheduling for the uplink of Long Term Evolution (LTE) systems. The aim is to minimize the total transmit power while satisfying particular delay requirements. The scheduling process is formulated as a dynamic programming (DP) problem. We show that the global optimal solution requires the knowledge of future arrival rates and future channel gains for all users. Alternatively, we propose two low complexity heuristic schedulers. The first heuristic sched- uler controls the maximum allowable transmit power (MATP) for each user based on the queue length. In particular, if the queue length of a user is relatively small, the scheduler reduces the user’s transmission rate as well as the MATP to minimize the total transmit power. On the other hand, when the queue length is large, the scheduler increases both the transmission rate and the MATP to satisfy the delay requirements. The second heuristic scheduler controls the minimum acceptable bit per Watt ratio (BPWR) for each user. Users can only transmit if their BPWRs are greater than an acceptable level, which allows only high power efficient transmission. The minimum acceptable BPWR is changed adaptively based on the queue length of each user. The performance of the heuristic schedulers is evaluated and compared to the optimal solution and other existing schedulers. Keywords—LTE, scheduling, dynamic programming, SC-FDMA. I. I NTRODUCTION The recent years have witnessed unremitting advances in the wireless technology domain, which served to grow the mobile data market. New wireless applications and services have emerged, and accessing data services via mobile devices has increased considerably. The size of the global mobile data traffic has increased 18 times over the period 2000-2013 [? ]. It is expected to increase 11-fold by 2018 [? ]. Furthermore, 11-fold growth in the aggregate smartphone traffic is expected between 2013 and 2018 [? ]. To keep up with the increase in mobile data traffic, Long Term Evolution (LTE) technology has been developed to support high performance radio-access technology. LTE supports high data rate links and enables users to run multiple concurrent applications with heterogeneous quality of service (QoS) requirements, such as live streaming of audio, video, and social media applications. However, to maintain a Copyright (c) 2013 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending a request to pubs-permissions@ieee.org. fixed error performance, increasing the transmitted data rate is accompanied with a power increase to keep the energy per bit unchanged. Furthermore, as the total number of bits transmitted per unit time grows, the total transmission power per unit time becomes substantially higher as well. Unfortu- nately, the increasing demand for transmission power is quite higher than the improvement in batteries’ capacity. As most end-user devices are powered from small size batteries, high data rate transmission would reduce the average operation time per charge of battery-powered devices. Consequently, the development of power-efficient transmission techniques has become an important design consideration to improve the battery life of mobile devices. In the literature, there has been increasing interest to better understand and model the power consumption of smartphones. For example, Zhangt et al. [? ] designed an online power model that estimates the power consumption of different com- ponents in Android smartphones including central processing unit (CPU), liquid-crystal display (LCD), global positioning system (GPS), audio, Wi-Fi and cellular interfaces. The work reported in [? ] models the impact of wireless signal strength on smartphone energy by analysing traces collected from 3785 smartphones. A power model of a commercial LTE network is presented in [? ], where an application is designed and installed on Android smartphones to collect traces of the power consumption. The study suggests that the power consumption of LTE is 23 times higher than the power consumption of WiFi interfaces. In LTE systems, the LTE uplink is based on single carrier frequency division multiple access (SC-FDMA). Compared to orthogonal frequency division multiple access (OFDMA), SC- FDMA has lower peak-to-average power ratio (PAPR). The Low PAPR advantage of SC-FDMA is achieved by localized- mapping of the resource blocks (RBs), where each user should be mapped to a subset of contiguous RBs. Resource scheduling in OFDMA-based systems has been widely investigated in the literature [? ]. Many schedulers have been developed to optimize different allocation metrics such as the sum rate maximization [? ], total transmit power min- imization [? ], and fairness [? ]. Several solutions have been presented based on game theory [? ], convex optimization and dual decomposition [??? ], dynamic backpressure policies [? ], and interior point methods [? ]. However, the contiguity constraint of the SC-FDMA changes the scheduling problem into a non-convex optimization problem [?? ], and prevents the