Energy-Efficient Design in Wireless OFDMA Guowang Miao † , Nageen Himayat ∗ , Ye (Geoffrey) Li † , and David Bormann ∗ † School of ECE, Georgia Institute of Technology Atlanta, Georgia 30332–0250, email: gmiao3@gatech.edu and liye@ece.gatech.edu ∗ Wireless Commun. Lab./Commun. Tech. Lab., Corp. Tech. Group, Intel Corp. Santa Clara, CA, email: nageen.himayat@intel.com and david.bormann@intel.com Abstract—Energy-efficient transmission is an important aspect of wireless system design due to limited battery power in mobile devices. We consider uplink energy-efficient transmission in OFDMA systems since mobile stations are battery powered. We account for both circuit and transmit power when design- ing energy-efficient communication mechanisms and emphasize energy efficiency over peak rates or throughput. Both link adaptation and resource allocation schemes are developed to optimize the overall bits transmitted per Joule of energy, which allows for maximum energy savings in a network. Our simulation results show that the proposed schemes significantly improve energy efficiency. Index Terms– energy efficiency, OFDMA, bits per Joule, link adaptation, resource allocation I. I NTRODUCTION Power efficiency is becoming increasingly important for wireless communication systems due to limited battery re- sources in mobile devices. Unfortunately, battery technol- ogy has not progressed as fast as silicon technology [1]. Hence, recent energy-efficient management schemes [2]–[4] have focused on minimizing energy consumption rather than throughput maximization [5]. Additionally, orthogonal fre- quency division multiple access (OFDMA) has emerged as one of the prime multiple access schemes for next generation multi-user broadband wireless networks. However, limited research has been done for energy-efficient communication in OFDMA systems. In this paper, we consider uplink energy- efficient transmission in OFDMA systems to improve battery consumption at the mobiles. We account for both circuit and transmit power when designing link adaptation and resource allocation schemes, and emphasize energy efficiency over peak rates or throughput. We initially focus on the case of flat-fading OFDMA channels, and defer the frequency selective case to future work. The rest of the paper is organized as following. In Section II, we briefly describe the system model. Then we develop optimal energy-efficient link adaptation and network resource allocation schemes in Sections III and IV respectively. Finally, we conclude the paper in Section V. II. SYSTEM DESCRIPTION Multiple access is achieved in OFDMA by assigning sub- channels to individual users based on quality of service (QoS) requirement and channel condition. This allows simultaneous data transmission for several users. For simplicity, we investi- gate energy-efficient OFDMA communication with flat fading channels in this paper. Consider uplink transmission in an OFDMA network with one base station (BS) and multiple users, i.e. mobile stations. Denote N and K as the numbers of users and subchannels, respectively. Denote c i as the number of subchannels assigned to User i. Each subchannel will be assigned to one user exclusively at each frame slot. Hence, N  i=1 c i ≤ K. (1) Denote r i as the achievable data rate on each subchannel by User i, then the data rate of User i is R i = r i c i . (2) The BS allocates subchannels to improve the overall network energy efficiency, which is measured by the number of bits transmitted per Joule. Additionally, each user also adjusts transmit power and modulation order for further optimization. III. OPTIMAL ENERGY-EFFICIENT LINK ADAPTATION This section considers per link adaptation schemes that will result in minimum energy consumption, or equivalently, maximum energy-efficiency. Throughout this section, assume c subchannels are assigned. Since we focus on per link energy- efficient optimization, subscripts indicating different users will be dropped subsequently. A. Energy-Efficient Transmission Rate Power consumption of a mobile station in transmission mode consists of two parts. The first is circuit power, denoted as P C , which is independent of data rate and exists whenever the user is in transmission mode. The second is transmit power, P T (R), which depends on data transmission rate, R. For example, we consider an additive white Gaussian noise (AWGN) channel with signal bandwidth W , the achievable data rate is given by the Shannon capacity as R = W log(1 + P T g N o W ), (3) This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 proceedings. 978-1-4244-2075-9/08/$25.00 ©2008 IEEE 3307