Radio-Efficient Adaptive Modulation and Coding: Green Communication Perpective Liqiang Zhao, Jian Cai, and Hailin Zhang State Key Laboratory of Integrated Service Network Xidian Univerity Xi an, Shaanxi, China E-mail: lqzhao@mail.xidian.edu.cn Abtract- To leen the environmental impact of the communication indutry, Green Communication ha achieved increaing attention from government, academia, and indutry. However, how to define a green communication ytem ha been an open topic. In thi paper, we carry out a comprehenive analyi of the b//hz bandwidth efficiency, the b//hz/m area pectral efficiency, the b/tenu power efficiency, the b//hz/w power efficiency, and the (b m)//hz/w green efficiency of a wirele link. After comparing the performance of the green efficiency with the other efficiencie, we preent an adaptive modulation and coding cheme to approach the maximum green efficiency, which provide an effective way to Green Communication. Keyword- Green Communication, Efficiency, Adaptive Modulation and Coding I. INTRODUCTION Given that there i a worldwide growth in the number of mobile ubcriber, with the majority of the tele-traffic evolving from low data rate peech and modet-rate text meaging toward high data rate multimedia ervice (e.g., online muic and video), an increaing contribution of information technology to the overall energy conumption of the world i oberved. Therefore, there i a need to reduce the energy requirement of radio acce network. Hence, energy conumption will become a more important contraint in the deign of future mobile communication ytem [1-]. It ha become clear that 3GPP Long Term Evolution (LTE) i the preferred technology for the next generation mobile communication ytem. Novel tranmiion technologie, uch a multiple input and multiple output (MIMO), and orthogonal frequency diviion multiplexing (OFDM), will be adopted for LTE. However, the development of uch mobile broadband communication ytem come at a ignificant energy cot, which i unutainable and unuitable for futureproof communication. Therewith, over pat two year, Green Communication ha received much attention from government, academia, and indutry [3]. The aim of Green Communication i to invetigate and create innovative method for the reduction of the total power needed to operate the future mobile communication ytem and to identify appropriate network architecture and radio technologie which facilitate the Thi work i upported by the 111 Project (B08038), State Key Laboratory of Integrated Service Network (ISN090105), Program for New Century Excellent Talent in Univerity (NCET-08-0810), National Natural Science Foundation of China (No. 6077137), and UK-China Science Bridge: R&D on (B)4G Wirele Mobile Communication. required power reduction. Up to now, the continuou invetment in Green Communication bring about a wealth of theoretical knowledge and practical engineering olution, which can mainly be departed into two categorie. The firt part conider highly efficient power amplifier deign. For example, [4] preented high efficiency cla F amplifier. The econd part focue on network topology, MAC protocol, and routing deign [5-6], which i alo called Green Network. However, a widely accepted choice of a criterion characterizing the overall efficiency of a wirele network remain an open problem [7]. Our work ha provided a uniform efficiency for green network deign and dicued the b//hz/w power efficiency and the b//hz bandwidth efficiency of wirele meh network [8]. Hence, firtly, in ection II we preent of a novel definition of green efficiency. Secondly, a much invetment in wirele ytem aim at improving the bandwidth efficiency [9] and the power efficiency [10-11], in ection III, we carry out a comprehenive analyi of the bandwidth, power and green efficiency of a wirele link. Thirdly, in ection IV, baed on the bandwidth, power, and green efficiency criteria, we preent a heuritic approach for a wirele ytem to achieve the maximum band, power, and green efficiency. Fourthly, in ection V imulation tudie are carried out to evaluate the performance of the propoed approach. Finally, the concluding remark and the ubject for further tudy are given in Section VI. II. GREEN EFFICIENCY CRITERION FOR WIRELESS SYSTEMS By definition, efficiency i the ratio of the attained utility to the conumed reource. Clearly, therefore, the notion of efficiency i cloely related to the definition of the utility and the reource. In wirele communication, a uer aim at tranmitting it packet uccefully under it quality-of-ervice (QoS) requirement and available reource contraint over a certain ditance to it receiver. Hence, the radio utility metric hould include the uccefully packet in bit, QoS metric (uch a bandwidth in b/, delay and jitter in econd, and packet lo rate) and the tranmiion ditance in meter. Hence, the definition of green efficiency can be expreed a 978-1-444-8331-0/11/$6.00 011 IEEE
arg max Radio Efficiency (1) S.T.1 (atifying the minimal QoS requirement) Bandwidth > Bandwidthmin Delay < Delaymax Jitter < Jittermax PacketLoRate < PacketLoRate where Bandwidth min i the minimum required bandwidth, Delay max, Jitter max, and PacketLoRate max are the maximum tolerant delay, jitter, and packet lo rate repectively. S.T. (atifying the contraint of available reource) max Allocated timelot Available timelot Allocated ubcarrier Available ubcarrier Allocated antenna Available antenna Tranmitted power Max tranmiion power There ha been a lot of work on the ue of utility function in the networking area. And ome of thee work [1] could be applied here too. However, unfortunately, the above problem ha been proven to be NP-hard [13], o we cannot hope an algorithm that can find the theoretical optimum and run in polynomial time. Hence, if not conidering the QoS requirement and reource contraint, the definition of the utility could be implified a bit-time-meter (b m). On the other hand, available radio reource up to now can be claified into five categorie, time, frequency, pace, code and power in five different domain. Hence, a novel concept of green efficiency i preented in [8], which i defined a the number of bit time tranmiion ditance per reource category in a domain, correponding to (bit-time-meter)-perecond-per-hertz-per-antenna-per-code-per-w ((b m)//hz/antenna/code/w). Thi definition i generic and all the five categorie of radio reource can be deignated. In view of the frequency pectrum and the power, two major reource categorie, many reearcher conider the b//hz bandwidth efficiency and the b//hz/w power efficiency repectively a the principal efficiency criterion. Obviouly, the bandwidth and power efficiency are two pecial cae of green efficiency. According to the Shannon-Hartley theorem, the bandwidth efficiency of a communication ytem i defined a the number of bit per unit bandwidth, which in the cae of ingle-input ingle-output (SISO) ytem correpond to bit-per-econdper-hertz (b//hz), while in multiple-input multiple-output (MIMO) ytem i equivalent to bit-per-econd-per-hertz-perantenna (b//hz/antenna). Furthermore, [14] define the area pectral efficiency a bit-per-econd-per-hertz-per-m (b//hz/m ). On the other hand, a hot of pread-pectrum method intentionally acrifice the b//hz performance for the ake of achieving a better bit-per-econd-per-watt (b//w) power efficiency. Moreover, [11] define the power efficiency a the number of bit per thermal noie ignal energy unit (TNEU). By no mean hould thee method be claified a le efficient, ince in the appropriate circumtance they are capable of coniderable improving the overall performance of the entire network. For implicity, in the following, we only conider the b//hz bandwidth efficiency, the b//hz/m area pectral efficiency, the b/tenu power efficiency, the b//w power efficiency, and the (b m)//hz/w green efficiency, and give a comprehenive analyi of the four efficiencie. III. EFFICIENCIES OF A WIRELESS SYSTEM A. Bandwidth, area pectral, and power efficiencie The ytem capacity i defined a the maximum poible tranmiion rate uch that the probability of error i arbitrarily mall [15], which i quantified by the Shannon-Hartley theorem a: ( ) ( b ) C = Blog 1+γ (), where B i the channel bandwidth in Hz. γ = S/N 0 i the average ignal-to-noie ratio (SNR) in db recorded at the receiver, where S denote the ignal power, and N 0 denote the noie power. The above equation decribe the capacity of a Gauian channel, auming an infinite duration of the tranmitted ignal, a well a an infinite detection/decoding complexity. Subequently, the b//hz bandwidth efficiency of a communication ytem i defined a C η b = = log ( 1+ γ ) ( b / / Hz) (3). B The b//hz/m area pectral efficiency i defined a η ASE ( 1+ γ ) ( b / / Hz m ) ηb log = = / (4), A πd where we conider a circle area A = πd, with the tranmitter at the center and a radiu of d. The b/tenu power efficiency i defined a η ( + γ ) η log = (5), γ1 b TENU = / γ ( b TENU ) where TNEU refer to the amount of ignal energy identical to the variance of the complex-valued AWGN ample recorded at the receiver. The b//hz/w power efficiency i defined a ( 1+ γ ) ηb log ηw = = ( b / / Hz / W ) (6), where P t i the tranmitted power in W. B. The (b m)//hz/w green efficiency The (b m)//hz/w green efficiency i defined a
( 1+ γ ) d log η m = ηw d = (( b m) / / Hz / W ) (7). P t where d donete the tranmiion ditance between the tranmitter and receiver. For implicity, we conider the free-pace propagation model in thi paper. And the free-pace power received by an antenna at a ditance d from the tranmitter i given by GtGrλ S = P PL = P (8), ( 4π ) d L t t where PL i the propagation lo, G t and G r are the tranmitter and receiver antenna gain repectively, λ i the wavelength, and L i the ytem lo not related to propagation (L 1). Subtituting (8) into (7), we can derive an explicit formula of the green efficiency veru the tranmiion power and tranmiion ditance, P G G t t r d log λ 1 + ( 4 ) N d L π 0 η m = (( b m) / / Hz / W ) (9). Obviouly, the green efficiency decreae with the increaing of the tranmiion power, and increae with the increaing of the tranmiion ditance, a hown in Fig. 1. However, when the tranmiion ditance i larger than about 0km, if the tranmiion ditance continue increaing, the green efficiency will decreae a the bit error rate i very large. Figure 3. The (b m)//hz/w green efficiency and the b//hz/m area pectral efficiency a a function of the tranmiion ditance Figure 1. The (b m)//hz/w green efficiency Figure. The (b m)//hz/w green efficiency and the b//hz bandwidth efficiency a a function of SNR Figure 4. The (b m)//hz/w green efficiency and the b/tenu power efficiency a a function of SNR Moreover, the green efficiency i inverely proportional to the bandwidth and area pectral efficiency repectively, a hown in Fig. -5, a we conider the power a a category of
reource, and the tranmiion ditance a a category of utilitie. By adapting the modulation and coding cheme to the received SNR, we can approach the optimal b//hz bandwidth efficiency [16] and b/tenu power efficiency, a hown in Fig. 6. However, we cannot approach the optimal b//hz/m area pectral efficiency and b//hz/w power efficiency by mean of AMC, a hown in Fig. 7-8. The highet order modulation cheme, 56QAM, work bet, but it area pectral efficiency and power efficiency are far below the upper bound. Moreover, for implicity, temporal coding i not conidered in thi paper. Figure 5. The (b m)//hz/w green efficiency and the b//hz/w power efficiency a a function of the tranmitting power IV. GREEN EFFICIENT MODULATION AND CODING SCHEME In mot practical cenario characterized by variou performance-limiting factor including channel fading, interference a well a latency and complexity contraint, the actual attainable bandwidth, power, and green efficiency are coniderably lower than thoe predicted by Equation (3-7). However, adaptive modulation and coding (AMC) i a poible olution to approach the above efficiencie. Figure 7. AMC cheme for the b//m area pectral efficiency A. Adaptive modulatin and coding for bandwidth, area pectral, and power efficiencie Figure 8. AMC cheme for the b//hz/w power efficiency Figure 6. AMC hceme for the b/tenu power efficiency B. Adaptive modulatin and coding for the (b m)//hz/w green efficiency It i much complex to approach the upper bound of the green efficiency by mean of AMC a hown in Fig. 9-10. At a low tranmiion ditance, we adopt a higher order modulation
cheme, e.g., 56QAM, a it work better to provide a higher green efficiency. With the increaing of tranmiion ditance, we adopt a lower order modulation cheme, e.g., 16QAM, a it work better to provide a higher green efficiency. However, if the tranmiion ditance continue increaing, no modulation cheme can approach the upper bound of the green efficiency. That i to ay, we have to limit the coverage ize of a cell. Figure 9. Green efficient AMC cheme v. tranmiion ditance Figure 10. Green efficient AMC cheme v. SNR Similarly, at a high SNR, we adopt a higher order modulation cheme, e.g., 56QAM, a it work better to provide a higher green efficiency. With the decreaing of SNR, we adopt a lower order modulation cheme, e.g., 16QAM, a it work better to provide a higher green efficiency. However, if SNR continue decreaing, no modulation cheme can approach the upper bound of the green efficiency. I. CONCLUSION For a greener network, thi paper ha propoed a new efficiency criterion, the (b m)//hz/w green efficiency. Firtly, we carried out a comprehenive comparion of the green efficiency with the b//hz bandwidth efficiency, the b//hz/m area pectral efficiency, the b/tenu power efficiency, the b//hz/w power efficiency. Secondly, we preent an adaptive modulation and coding cheme to approach the upper bound of the green efficiency. REFERENCES [1] Smart 00: Enabling the low carbon economy in the information age, The Climate Group, Globe e-sutainability Initiative (GeSI), 008. [] G. P. Fettwei and E. Zimmermann, ICT energy conumption trend and challenge, the 11th International Sympoium on Wirele Peronal Multimedia Communication, Finland, September 008. [3] S. Fletcher, Green Radio@ Sutainable Wirele Network, VCE Core5 Programme, Available: http://kn.theiet.org/magazine/rateit/communication/green-radioarticle.cfm. [4] F. M. Ghannouchi and M. M. Ebrahimi, Invere cla F power amplifier for WiMAX application with 74% efficiency at.45 GHz, IEEE ICC, Dreden, Germany, June, 009. [5] L. Chiaraviglio, M. Mellia, and F.Neri, Energy-aware backbone network: a cae tudy, IEEE ICC, Germany, June, 009. [6] J. C. C. Retrepo, C. G. Gruber, and C. M. Machuca, Energy Profile aware routing, IEEE ICC, Germany, June, 009. [7] G. Gow and R. Smith, Mobile and Wirele Communication: An Introduction, Open Univerity Pre/McGraw-Hill, 005 [8] L. Zhao, W. Cheng, W. Zhang, X. Zhang, and H. Zhang, Power and Bandwidth Efficiency of Wirele Meh Network: Green Network Perpective, ubmitted to IEEE Journal of Selected Area on Communication. [9] A. Goldmith and S. G.Chua, Variable-rate variable-power MQAM for fading channel, IEEE Tranaction on Communication, Vol.45, No. 10, pp.118-130, 1997. [10] S. Verdu, Spectral efficiency in the wideband regime, IEEE Tranaction on Information Theory, vol. 48, no. 6, pp. 1319-1343, 00. [11] J. Akhtman and L. Hanzo, Power veru bandwidth efficiency in wirele communication: the economic perpective, IEEE VTC Fall, Alaka, USA, 009. [1] Frank Kelly, The mathematic of traffic in network, The Princeton Companion to Mathematic, Princeton Univerity Pre, 008, pp. 86-870. [13] M. S. Garey, D.S. Johnon, Computer and Intractability: Guide to the Theory of NP-Completene, W. H. Freeman, New York, 1979. [14] M. S. Alouini and A. J. Goldmith, Area pectral efficiency of cellular mobile radio ytem, IEEE Tran. Veh. Technol., vol. 48, no. 4, pp. 1047-1066, July 1999. [15] C. E. Shannon, A mathematical theory of communication, Bell Sytem Technical Journal, vol. 7, pp. 379.43 and 63.656, June and October 1948. [Online]. Available: http://plan9.belllab.com/cm/m/what/hannonday/hannon1948.pdf. [16] 3GPP TR 36.94, Radio Frequency Syetem Scenario, the 3rd Generation Partnerhip Project (3GPPTM), 009.1. [Online]. Available: http://www.3gpp.org/ftp/spec/html-info/3694.htm