Energy-Efficient User Scheduling and Power Allocation for NOMA Wireless Networks

Transcription

1 THE UIVERSITY OF EW SOUTH WALES SCHOOL OF ELECTRICAL EGIEERIG AD TELECOMMUICATIOS Eergy-Efficiet User Schedulig ad Power Allocatio for OMA Wireless etworks By Yag Zhag Thesis submitted as a requiremet for the degree of Master of Egieerig i Telecommuicatios Submissio Date: October 27, 2017

2 Abstract Previous research focuses o icreasig the etwork sum rate, as it is well ackowledged that there will be a exlosio o the trasformatio of data i the ext geeratio, so the sectrum efficiecy ad eergy efficiecy is gettig more ad more imortat ad recious, it is ecessary to take eergy efficiecy ito accout whe costructig our etwork. This aer is goig to comare the merits ad drawbacks of OMA ad OMA system, which oe is more eergy efficiet. More imortatly, it is aimig to exlore the fudametal riciles of OMA system such as suerositio codig (SC) ad successive iterferece cacellatio (SIC) techiques. Later art of this aer will discuss roosed algorithms to otimize the eergy efficiecy erformace by imrovig the scheme of user schedulig ad ower allocatio i OMA system. Keywords: OMA, OMA, SC, SIC. II

3 Ackowledgemets First of all, I would take this oortuity to exress my very great areciatio to Dr. Derrick Wig Kwa g or his valuable ad costructive suggestios durig the laig ad develomet of this research work. His willigess to give his time so geerously has bee very much areciated. I would also like to exted my thaks to the techicias of the laboratory of the electrical egieerig deartmet for their hel i offerig me the resources i ruig the rogram. Fially, I wish to thak my arets for their suort ad ecouragemet throughout my study. III

4 Cotets Abstract ii Ackowledgemets iii 1 Itroductio OMA key techique A SIC OMA key techique B ower reuse OMA key techique C Ideedet of user feedback CSI System model ad roblem formulatio OMA system with two users i oe subchael OMA system with multile users i multile subchael Problem formulatio Proosed User Pairig ad Power Allocatio Algorithm User airig algorithm Radom airig Chael state sortig-airig algorithm Power allocatio algorithm Power allocatio for each user i the same sub chael Power allocatio amog subchael Simulatio Result Results

5 4.2 Discussio Coclusio Future Work Bibliograhy 23 Aedix 1 29 A.1 MATLAB CODE

6

7 Chater 1 Itroductio With the dramatical growth i the umber of mobile subscribers ad smarthoes, the demad for higher data rate ad a more reliable commuicatio chael is becomig icreasigly fierce. Accordig to [1], it is goig to be 1000 times icrease i caacity comared to the curret etworks, it is redicted that there will be over 100 billio devices which meas there will be multile coectios. More tha 7.6 billio of mobile subscribers could use their cell hoes, smart hoes, ipad or latos dowload ad uload ucoutable iformatio ad data. etwork oerators are ruig at LTE latforms, oeratig the etworks, but it is ot able to bear the weight of the exlosio of data [2]. The oerator is facig the challege today about how do they deliver the customer exeriece with all this drivers ad techology ad how do they drive their ROI rofitability. Therefore, there is a eed for a ew iformatio techology which is more eergy efficiet to cotribute to gree commuicatio. I the fourth geeratio, OFDM has bee widely adoted because it ca effectively elimiate the iterferece betwee sigal waveforms ad highly icreased the sum rate of the wireless system. OFDM is a advaced techology which has high trasmittig rate, the cocet of this advaced techology is to trasform the serial data ito arallel data ad multilexed ito relatively low seed ad modulated by differet carriers. The symbol badwidth is dramatically elarged by 1

8 the arallel chael. The erformace of ati-fadig ad ati- iterferece will be imroved as well. The sectrums of each subcarrier are ot mutually overlaig by usig multilexig method i traditioal frequecy divisio. A eormous umber of filters are required at trasmitter ad receiver side, this dramatically icreases the costructio cost of the system ad its comlexity [4] [5]. Simultaeously, Sufficiet frequecy itervals must be maitaied betwee the subcarriers to deal with the iter carrier mutual crosstalk, which will reduce the sectrum efficiecy of the system. Although The data trasmissio rate of OFDM could reach to hudreds of megabytes er secod or eve Gigabit, ad it may be able to meet the eeds of mobile commuicatios alicatios to some extet for a eriod of time. With the oularizatio of smart termials ad the growig demad for ew mobile services, the demad for wireless trasmissio rate will icrease exoetially, ad the trasmissio rate of wireless commuicatios will still be difficult to meet the eeds of future mobile commuicatios alicatios. To deal with more strict ad harsh situatio i 5G, maage more ad more data, to ursuit higher sum rate ad caacity, more sectrum efficiecy ad eergy efficiecy. OFDM is ot eough to meet those deliverables, Therefore, OMA is well cosidered to achieve higher system throughut ad kee the low cost at the same time. I OMA, A sigle radio resource ca oly be assiged to oe user, such as frequecy divisio or time divisio, while i OMA system, a resource ca be allocated for multi users. I some scearios, such as the ear-far effect ad wide coverage of multi-ode access scearios, esecially the ulik itesive scee, o-orthogoal multile access multilexig by usig ower multilexig has obvious erformace advatages comared to traditioal OMA access, it is more suitable for the future deloymet of the system [2]. The reset research [1] has verified the OMA effect i cellular dowlik sceario which ca icrease the overall throughut of macro cell to a large extet. The o-orthogoal multile access through a combiatio of serial iterferece cacellatio or maximum likelihood demodulatio to achieve the ultimate caacity limit, so the difficulty of desig ad imlemetatio is desiged a low comlexity ad effective receiver algorithm. This 2

9 is oe of the hottest research toics worldwide. From the above iformatio, we ca see that OMA is ew techology which icororates some 3G ad 4G techiques ad ideas. For examle, subchael ad carriers were metioed i 4G ad SIC was iitially utilized i 3G. While for 3G, it alies CDMA techology to trasmit o-orthogoal sigal, all users are sharig oe chael, ad RAKE receiver is used to receive sigal. But there is a serious earfar roblem, so ower cotrol is adoted i 3G. The techology limits the ower to the user ear the cell ceter, guarateed the arrived sigal at the receivig termial, the ower of each user is equal. I the 4G system, it is based o OFDM techology, differet users take orthogoal trasmissio, thus the ear-far roblem is ot ifluecig so much ad ower cotrol is ot ecessary as well. As OFDM alied Adative modulatio codig (AMC) techology, modulatio codig ca be automatically adjusted accordig to lik state iformatio, so as to rovide the best ass ad trasmit rate to the user, however, to some extet it has to rely o the lik status iformatio of the user feedback [7-12]. Comared to CDMA ad OFDM, OMA alies orthogoal trasmissio betwee subchaels, so the ear-far roblem is ot as much as it i 3G, ad multile access iterferece roblems is ot that serious as well. Because it is ot deedig o the users feedback CSI, whe alyig AMC ad ower multilexig techiques, it could easily deal with variable liks, eve uder a fast-movig eviromet, it ca still kee good seed erformace [22]. More imortatly, OMA allow oe subchael ca be shared by multile users, uder the same trasmit rate, it ca defiitely icrease the sectrum efficiecy comare with 4G. I the light of the research ublished by some of the world's leadig research orgaizatios, the sectrum efficiecy will be oe of the key oits of 5G. From this oit of view, OMA, which ca ot oly satisfy the mobile service seed demad but also ehace the sectrum efficiecy, is likely to be adoted as the ew multi-access techology i 5G [9]. Figure 1 clearly shows the differece betwee 3G, 4G ad 5G. 3

10 Figure 1. The overview of 3G, 4G ad 5G. Cosiderig the imortace of OMA techology, it is ecessary to uderstad the riciles ad key techiques of OMA system. Here are several key techologies alied i OMA system: A. Successive Iterferece Cacellatio (SIC) [23]. At the trasmitter side, similar to the CDMA system, by itroducig the iterferece iformatio to the system, it ca acquire higher sectrum efficiecy, but also meet the multile access iterferece (MAI) roblem. To elimiate the multile access iterferece roblems, may achievemets have bee made i the rocess of the third-geeratio mobile commuicatio system, successive iterferece cacellatio (SIC) is oe of them. The detectio method used for multile users is alied at the receiver side i OMA. SIC techique is used for cacellig the iterferece by gradual elimiatio strategy, the user was distiguished oe after aother. The by doig amlitude comesatio ad multi-ath iterferece cacellatio o each user, the system total multi-access iterferece was wied off from the received sigal. B. Power reuse. To elimiate the multi-access iterferece at the receiver side, it is ecessary to judge the iterferece cacellatio riory of the differet user, ad the 4

11 judgmet is based o the sigal ower of the user [25] [26]. To achieve otimum erformace ad distiguish the differet users at the same time, the base statio allocated differet ower level for trasmittig multile users sigal. Power reuse is ot widely used i covetioal multi-access schemes. The base statio follows the relevat algorithm to do the ower allocatio rather tha just simly cotrol the sigal ower. C. Ideedet of user feedback CSI. I ractice cellar etworks, due to the delay of user feedback CSI ad dyamic etwork system. It is hard to rovide real effective chael state iformatio i real time. Although may advaced techologies are ot relyig o this ad ca acquire stable erformace gai, OMA which is alied SIC techology is more suitable for this kid of situatio, which ca achieve better erformace i high-seed movig scees ad costruct better mobile ode backhaul [28]. I this aer, we focus o icreasig eergy efficiecy of OMA system from two differet asects. The first is to fid out the otimal user schedulig scheme to maximize the system s eergy efficiecy [29]. The secod aroach, based o the roosed scheme i the first art, is to cosider ad determie the ercetage of the ower used for each user i the same subchael, roose a ower allocatio algorithm for OMA system ad measure the eergy efficiecy erformace of the etworks by usig the uit of bits er Joule [30]. The later art of the aer is arraged as follows, chater 2 will rovide the system model followed with roblem formulatio. I chater 3, I will discuss the roosed user airig as well as ower allocatio algorithm for OMA system. The, the simulatio result will be show i chater 4 with discussio. Fially, I will coclude this aer with summary ad future work. 5

12 Chater 2 System Model ad Problem Formulatio We are tryig to establish a dowlik of OMA system etwork. Ad users are ruig SIC techique at their termials, assume that there are M users ad subchaels i the system ad we use m to deote the mobile user ad deote the th subchael i this system. There is distace R which deote the radius of the cellular etwork ad M users are radomly distributed i this area. The total badwidth is B ad it is slit ito subcarriers. The total ower of the system is P s ( s ) ad the subchael s total ower is 1 P M ( i, i1 ). I OMA system, we assume that a great umber of users could be arraged i the same subcarrier ad they are radomly aired together, further assume that the users will exeriece the same Rayleigh fadig i the same subchael. 2.1 OMA System with Two Users i Oe Subchael. 6

13 Figure 2. The system model of two users aired i oe subchael. I this model, the received sigal i this system is exressed as: yi m i m i h s h s, i=m; h Where i is the chael gai of user m or. is AWG. I figure 2, it is clearly show that users are multilexed i BS by ower domai, ad more ower is allocated to the user with oorer chael coditio user m, which ca be uderstood as the cell-edge user. For examle, there are two users are allocated i oe subchael. Whe receivig sigals, both of user m ad user should receive the same sigal which combied both of their sigals together ad have their messages. User i ear the base statio ad has a good chael coditio, so the user ca aly SIC at receiver side to elimiate the iterferece from user m, which meas that user decode the user m sigal ad the remove it from the combied sigal, therefore, there is oly user s sigal ad oise left. So, comared to the revious sigal user ca achieve better erformace with the 7

14 higher rate. While for user m, it is farther from the base statio tha user, ad its chael coditio must worse tha user, so it ca ot aly SIC to cacel the iterferece from user, so it decodes its ow messages by treatig the user s sigal as oise. Because the base statio allocated more ower to user m, so it ca still able to decode its sigal ad acquire some rate. [31][32] Therefore, the ricile of OMA trasmittig rocess ca be regarded as the aired users are allocated with differet levels of ower accordig to their chael coditio, ad all users are alyig SIC algorithm i their ow receivers. 2.2 OMA System with Multile Users i Multile Subchaels. Cosider M users are arraged i subchaels, the received sigal of user m o subchael ca be show as: y h s h s Z m, m, m, m i, m, i m, i1, im M Where s m ad s i are modulated sigals, is the subchael chael gai which i this case, as I metioed before, is Rayleigh fadig chael gai. Z m, is AWG oise whose mea is zero ad variace is 2. I OMA system, it is allowed multile users to share oe subchael at the same time ad frequecy, so it will create a large amout of iterferece, if we ot usig SIC techology at the receiver side, the received SIR is writte as: SIR m, = h m, m, M 2 2 i, hm, i1, im m, Hm, = M 1 H i1, im 2 i, m, Where H h m, m, o subchael. 2 2 rereset the chael gai ormalized by oise of user m 8

15 Before we move o to the SIC decodig, oe of the imortat rocedure is to determie the SIC decodig order. I order to icrease the sum rate of the system, we always set higher ower level to those users with bad chael coditios, for examle, for ay two users m i coditio h i, better tha h j, ad m j, we always set i the same subchael with chael i, j,. Therefore, we ca rak the users riority based o the chael coditio. Ad the raked order ca be exressed as: h h h h h 1, 2, m, m1, M, Whe we aly SIC at the receiver side, accordig to the chael coditio order, we ca cacel those iterfereces from the user with the chael coditio worse tha you, ad the we treat those users sigal which chael coditio better tha you as oise. Therefore, we ca achieve a better SIR with SIC. The received sigal to iterferece lus oise ratio with SIC at the receiver side is writte as: SIR = 1 m, m, m, m1 i1 Therefore, the sum rate of oe subchael is: H H i, m, R =Blog 1 1 m, m, m, 2 m1 i1 H H i, m, As there are M users i subchael, ad there are subchaels i the system, so the total system sum rate ca be exress as: R= R = Blog 1 M M m, m, R m, 2 m1 1 1 m1 1 m1 1 i, Hm, i1 H 2.3 Problem Formulatio. 9

16 We are exlorig the eergy efficiecy of the system ad use bits er Joule as its uit. We eed to fid out both the system rate ad the eergy cosumtio. The system sum rate has described i art B, so we eed to defie eergy cosumtio ext. There are two arts eergy cosumtio i this system, oe is sigal trasmit ower cosumtio ad the other oe is circuit ower cosumtio such as decodig. Therefore, the total eergy efficiecy of the system ca be give by: E= c R 1 1 Subject to 1 s, 0 Where R is oe subchael sum rate ad is its subchael ower, c circuit ower cosumtio. Our task is to icrease the eergy efficiecy of the OMA system as much as ossible which is to maximize E. is max E. max 0,1, 0 c R 1 1 Subject to 1 s Where is the otimal ower roortioal factor of each user ad it eeds to be determied i the algorithm of ower allocatio. However, this otimizatio roblem is a o-covex roblem, there are o systematic aroaches for solvig the roblem efficietly. The followig research will aalysis it o user schedulig ad ower allocatio searately. 10

17 Chater 3 Proosed User Pairig ad Power Allocatio Algorithm 3.1 User airig Algorithm. It is imortat to air the users at the trasmitter because it ca make sure the badwidth ca be fully utilized, the users are allocated ito each sucarrier symmetrically as well as ehacig user fairess. The followig algorithm are assumig two users. users 2 subchael which mea that each suchael are assiged with 1) Radom airig. [35] Radom airig is the most easier way for user airig, which meas that the base statio chooses users radomly ad allocated ito a radom emty suchael. This method resets oor erformace because the user chael state iformatio does ot used. Some users with similar chael coditio may have large iterferece due to the ower allocated to each of them may be same level, 11

18 both of them caot aly SIC ad both of them may have to treat the other oe as oise. Therefore, the radom airig is defiitely usuitable. 2) Chael state sortig-airig algorithm. [35-41] To icrease the system sum rate, decrease the iterferece betwee two users i the same subchael ad ehace the user fairess, the sortig- airig algorithm are imlemeted i this aer, which is air the best chael-coditio user with the worst chael-coditio user for each subchael. However, the last air may have large iterferece betwee these two users because their chael coditios may have o much differece. But the other chaels are erformig very good, therefore, this airig method ca miimize the total system iterferece betwee two users i the same subchael. 12

19 Figure 3. Process of Chael state sortig-airig algorithm. The airig rocess is show i figure 3. Firstly, it will register users ito the system ad the radomly select a emty subchael, the lace the cadidate user i ascedig order accordig to their chael coditio. Pair the users with the best ad worst chael coditio ad assig them ito the chael. The rest of users will the select aother emty subchael ad sortig agai accordig to the ew chael coditio i the ew subchael, aother aired user will come out from the ew ascedig order. This loo will sto whe all the users are all aired ad assiged ito subchaels. If there is subchael left ad it is emty, the system will register users ad go through all the rocedure agai to make sure all subchaels are fully used ad the system maximized rate is achieved. 3.2 Power allocatio algorithm. After user airig, Base statio eed to allocate differet levels of ower to differet sub chaels, ad i each sub chael, it is also required to determie the ower roortioal factors for those two users assiged i the same sub chael. Therefore, there are two arts of ower allocatio roblems i this algorithm. 1) Power allocatio for each user i the same sub chael. [39-46] Accordig to the Shao s chael caacity theory, the trasmit ower allocates to oe user affect the achievable sum rate of the sub chael. Assume there are U two users 1 U ad 2 that are to be multilexed over a subchael with the H chael coditio 1 H2. The ricile of SIC decodig sequeces makes that U1 U ca cacel the iterferig ower term of 2 U, whereas 2 treats the sigal U ower of 1 as oise. I this sectio, the roblem of fidig the aroriate ower U for 1 U ad 2 to maximize the subchael trasmit rate ca be formulated as: 13

20 h max R u log 1 u log 1 h 1, h 1 2 Subject to C : P, max C : 0, To solve this roblem, for fixed should meet 1 Pmax 2 to maximize 2, R is mootoically icreasig with R 2. The we have: 1. The 1 h1 Pmax 2 R 2 u1 log2 1 u2 log2 1 h2 2 1 h 1 2 = u log 1 h u log 1 h u log 1 h max The dr 2 u1h1 u2h2 d 1 h l 2 1 h l 2 u2h2 -u1h 1-u1 u2 h1h 2 2 1h 1h l = Sice 1 0, 2 0 u, ad 1 u2 to erform OMA oeratio, the zero oit of dr d 2 2 is 2 u h u h u u h h Thus, the maximum value of R is obtaied whe P As max 2 u h u h u u h h P , 1 max 2 is the total ower for a subchael. Whe cosiderig a sigle subchael, it is a fixed value which is assiged by the BS, I will discuss the allocatio scheme P amog subchaels i the ext art. c is the circuit ower which is also a costat value, therefore, as we get the maximum R of a subchael, the maximum eergy efficiecy for this subchael which is E 14 _ max = R c _ max max ca be obtaied.

21 2) Power allocatio amog subchael based o Dikelbach method. [43-67] I our revious work, user airig algorithm ad ower allocatio solutio for two users i the same sub chael have bee discussed. The airig structure ad ower allocatio scheme allows the use of the solutio i allocatig ower for users withi a air (i oe sub chael). However, the ower allocatio across the airs (across differet sub chaels) have to be determied. To maximize the system erformace, the BS delivers differet levels of ower amog those subchaels ad the roblem of maximizig eergy efficiecy for the whole system is formulated as max 0, s 1 max 0, s 1 Subject to c 1 1 R 1 1 u log 1 h u log 1 h u log 1 h 1, 2 1, 1, 2 1, 2, 2, 2 2, 2,, 0 s c 1 Due to the comlexity of this covex roblem, it is very hard to fid the otimal oit of it, therefore, i order to solve the roblem efficietly, we covert (13) to Dikelbach reresetatio that ca be simly writte by Where, 0, s 1 max F q qd 1, log 2 1 1, 1, log 2 1 1, 2, 2, log2 1 2, 2, R u h u h u h 1 1 c 1 D q D ad which ca rereset our system eergy efficiecy. If there is a solutio F makes q 0, which meas 15

22 qd D 0 q [69-71] That meas that if there is a makes F q 0 solutio ca obtai a better q. therefore, we relace q by, this testified that this set of q ew = D, ad redo the calculatio. There must be a solutio otimal F which makes =0 ad F other solutio will all make <0, the we get the maximum q whe =0 with the solutio otimal F maximum eergy efficiet oit., which also meas that we get the The algorithm of this Dikelbach Method is show below. [51] [52] 1: Iitialize q k, set the iteratio umber k 0. F q q D do 2: while the objective fuctio 0 k 3: =0 4: For subcarrier idex =1 to do 5: = 1, log 2 1 1, 1, log 2 1 1, 2, 2, log2 1 2, 2, u h u h u h 6: ed for 7: If 8: Break 9: else 10: k 11: ed if 12: ed while Fq qk +1 c q = 1. This algorithm describes the roosed scheme of Dikelbach method to maximize 16

23 the system eergy efficiecy. It shows a simle way for solvig the otimizatio roblem with two users as a air multilex i oe subcarrier. To clarify the imrovemet of the system eergy efficiecy, the aroach of equal ower allocated o each subchael will be used i the comariso i the simulatio art. 17

24 Chater 4 Simulatio Result I this sectio, simulatio results are reseted to evaluate the erformace of the roosed user airig ad ower allocatio algorithms for OMA system. I the simulatio, we cosider oe base statio, ad there are 6 users are uiformly distributed i the cell rage of 500m with the radom locatios. The miimum distace betwee a user ad BS is 30 meters, the badwidth ad carrier frequecy for the system are 0.2Mhz ad 2.0Ghz resectively. To reduce the demodulatig comlexity of the SIC receiver, we set 64 subchaels i this simulatio ad each subchael ca be allocated by oe matched air with two users i this OMA system. Ad the OFDM scheme as the comaratio ca oly be allocated by oe user i a subchael. For the subchael ower allocatio, we comare the roosed otimal algorithm with equal ower allocatio scheme. For further aalysis the relatioshi betwee eergy efficiecy, we set the circuit ower cosumtio we vary the from 0.5W to 1W ad 1.5W to see the differet erformace of our roosed algorithm. 18

25 Figure 4. Eergy efficiecy of the system vs. BS ower. Figure 4 resets the erformace of eergy efficiecy versus the maximum total trasmit ower of BS. The maximum trasmit ower of BS is from 10dBm to 30 dbm ad there are 6 users assiged i the system. Fig. 4 shows that the eergy efficiecy grows with the icreasig of maximum BS ower, ad it stabilized whe it gets the otimal oit, which meas that whatever how much more ower you add o the system, the redudat ower caot be utilized to icrease the system eergy efficiecy. Whe cosiderig the system erformace with differet levels of circuit ower cosumtio, it is oticeable that whe circuit cosumtio is small, the erformace of eergy efficiecy will be better. Moreover, as the circuit ower icrease, the otimal oit will move a little bit to right had side which meas that the BS eeds more ower to get to the otimal eergy efficiecy oit whe icrease the circuit ower cosumtio. 19

26 Figure 5. Eergy efficiecy of OFDM ad OMA with same circuit ower. I order to further comare the eergy efficiecy erformace of OMA ad OMA. We lot the eergy efficiecy for OFDM ad OMA uder the same coditios, with 6 users ad 64 subcarriers i the system, the circuit ower cosumtio are all 1w. From figure 5, we ca clearly see that for 2 users as a air i a subcarrier, the OMA system is always eergy efficiet tha OFDM. 20

27 Chater 5 Coclusio I this aer, I have rovided the reasos that our wireless etwork eed to focus o eergy efficiecy i the ext geeratio, ad also examied the merit ad demerit of OMA ad OMA system, basically ad techically. There are three key techologies alied i OMA system, SIC, ower reuse ad imlemetig full CSI. They are the mai cotributors to the better erformace of OMA system i eergy efficiecy cometitio. I sectio two, we have set a system model ad try to roose a eergy-efficiet user schedulig scheme ad ower allocatio algorithm. While these are ot easy because it is a o-covex otimizatio roblem ad there are o systematic aroaches for solvig the roblem efficietly. Therefore, i the rest of this aer, we roosed eergy-efficiet resource allocatio algorithms for 2-user OMA system, icludig chael state sortig-airig algorithm which is roosed to achieve maximum eergy efficiecy uder the kow subchael ower. For ower allocatio, there are two arts of ower allocatio task, oe is ower assigmet amog multilexed users i the same subcarrier ad the other oe is ower allocatio amog differet subcarriers. Sice the objective fuctio is o-covex, Dikelbach Method was utilized to simlify the equatio ad fid the otimal solutio. Thus, the otimal eergy efficiecy oit could be obtaied. 21

28 5.1 Future work For multile users airig ad ower allocatio roblems, it is much more comlex. For examle, whe alyig SIC, user with good chael coditio eed to decode several other user sigals ad remove them from the origial sigal to achieve high data rate. While, it will cosume more circuit o decodig, as the simulatio result shows the circuit ower cosumtio is a big ifluece agaist eergy efficiecy. Therefore, there is a tradeoff betwee reducig the circuit ower cosumtio ad maximizig data rate at the same time, which eed to do further aalysis. 22

29 Refereces [1] Z. Wei, J. Yua, D. W. K. g, M. Elkashla, ad Z. Dig, "A Survey of Dowlik o-orthogoal Multile Access for 5G Wireless Commuicatio etworks," acceted by ZTE Commuicatios, Se [2] X. Che, Z. Zhag, C. Zhog, ad D. W. K. g, "OMA i Dowlik SDMA with Limited Feedback: Performace Aalysis ad Otimizatio," IEEE JSAC SI i o-orthogoal Multile Access for 5G Systems, vol. 35, o. 10, , Oct [3] Q. Yag, H.-M. Wag, D. W. K. g, ad M. H. Lee, "Exloitig MultileAtea Techiques for o-orthogoal Multile Access," IEEE JSAC SI i oorthogoal Multile Access for 5G Systems, vol. 35, o. 10, , Oct [4] Q. Wu, G. Y. Li, W. Che, D. W. K. g ad R. Schober, "A Overview of Sustaiable Gree 5G etworks," i IEEE Wireless Commuicatios, vol. 24, o. 4, , [5] Y. Che, S. Zhag, S. Xu ad G. Y. Li, "Fudametal trade-offs o gree wireless etworks," i IEEE Commuicatios Magazie, vol. 49, o. 6, , Jue [6] C. Ha et al., "Gree radio: radio techiques to eable eergy-efficiet wireless etworks," i IEEE Commuicatios Magazie, vol. 49, o. 6, , Jue [7] D. W. K. g, E. S. Lo, ad R. Schober, "Wireless Iformatio ad Power Trasfer: Eergy Efficiecy Otimizatio i OFDMA Systems," IEEE Tras. Wireless Commu., vol. 12, , Dec [8] D. W. K. g, E. S. Lo, ad R. Schober, "Eergy-Efficiet Resource Allocatio i OFDMA Systems with Hybrid Eergy Harvestig Base Statio," IEEE Tras. Wireless Commu., vol. 12, , Jul

30 [9] D. W. K. g, E. S. Lo, ad R. Schober, "Eergy-Efficiet Resource Allocatio i Multi-Cell OFDMA Systems with Limited Backhaul Caacity," IEEE Tras. Wireless Commu., vol , Oct [10] Lei Li ad Zhisheg iu, "A itegrated subchael schedulig algorithm for adative modulatio ad codig (AMC) MIMO-OFDM wireless systems," Commuicatios, 2004 ad the 5th Iteratioal Symosium o Multi- Dimesioal Mobile Commuicatios Proceedigs. The 2004 Joit Coferece of the 10th Asia-Pacific Coferece o, 2004, vol.1. [11] D. W. K. g, E. S. Lo, ad R. Schober, "Eergy-Efficiet Resource Allocatio i OFDMA Systems with Large umbers of Base Statio Ateas," IEEE Tras. Wireless Commu., vol , Se [12] D. W. K. g, E. S. Lo, ad R. Schober, "Eergy-Efficiet Resource Allocatio for Secure OFDMA Systems," IEEE Tras. o Veh. Techol, vol. 61, , Jul [13] Kee-Bog Sog, A. Ekbal, Seog Taek Chug ad J. M. Cioffi, "Adative modulatio ad codig (AMC) for bit-iterleaved coded OFDM (BIC-OFDM)," i IEEE Trasactios o Wireless Commuicatios, vol. 5, o. 7, , July [14] D. W. K. g, E. S. Lo, ad R. Schober, "Dyamic Resource Allocatio i MIMO- OFDMA System with Full-Dulex ad Hybrid Relayig," IEEE Tras. Commu, vol. 60, , May [15] D. W. K. g, E. S. Lo, ad R. Schober, "Secure Resource Allocatio ad Schedulig for OFDMA Decode-ad-Forward Relay etworks," IEEE Tras. Wireless Commu., vol. 10, , Oct [16] D. W. K. g ad R. Schober, "Resource Allocatio ad Schedulig i Multi-Cell OFDMA Systems with Decode-ad-Forward Relayig etworks," IEEE Tras. Wireless Commu., vol. 10, , Jul [17] D. W. K. g ad R. Schober, "Cross-Layer Schedulig for OFDMA Amlify-ad- Forward Relay etworks," IEEE Tras. o Veh. Techol., vol. 59, , Mar [18] V. Lau, W. K. g, ad D. S. W. Hui, "Asymtotic Tradeoff betwee Cross-Layer Goodut Gai ad Outage Diversity i OFDMA Systems with Slow Fadig ad Delayed CSIT," IEEE Tras. Wireless Commu., vol. 7, , Jul

31 [19] L. Dai, B. Wag, Y. Yua, S. Ha, C. l. I ad Z. Wag, "o-orthogoal multile access for 5G: solutios, challeges, oortuities, ad future research treds," i IEEE Commuicatios Magazie, vol. 53, o. 9, , Setember [20] Z. Dig, Z. Yag, P. Fa ad H. V. Poor, "O the Performace of o-orthogoal Multile Access i 5G Systems with Radomly Deloyed Users," i IEEE Sigal Processig Letters, vol. 21, o. 12, , Dec [21] R. Q. Hu ad Y. Qia, "A eergy efficiet ad sectrum efficiet wireless heterogeeous etwork framework for 5G systems," i IEEE Commuicatios Magazie, vol. 52, o. 5, , May [22] Ha, S., Chih-Li, I., Xu, Z., & Su, Q. (2014). Eergy efficiecy ad sectrum efficiecy co-desig: From OMA to etwork OMA. E-LETTER. [23] Higuchi, K., & Bejebbour, A. (2015). o-orthogoal multile access (OMA) with successive iterferece cacellatio for future radio access. IEICE Trasactios o Commuicatios, 98(3), [24] L. Fag ad L. B. Milstei, "Aalysis of coded successive iterferece cacellatio i multicarrier DS/CDMA systems," 2000 IEEE Sixth Iteratioal Symosium o Sread Sectrum Techiques ad Alicatios. ISSTA Proceedigs (Cat. o.00th8536), Parsiay, J, 2000, vol.1. [25] Y. Saito, Y. Kishiyama, A. Bejebbour, T. akamura, A. Li ad K. Higuchi, "o- Orthogoal Multile Access (OMA) for Cellular Future Radio Access," 2013 IEEE 77th Vehicular Techology Coferece (VTC Srig), Dresde, 2013, [26] O. Aydi, E. A. Jorswieck, D. Aziz ad A. Zaoe, "Eergy-Sectral Efficiecy Tradeoffs i 5G Multi-Oerator etworks With Heterogeeous Costraits," i IEEE Trasactios o Wireless Commuicatios, vol. 16, o. 9, , Set [27] Y. Saito, Y. Kishiyama, A. Bejebbour, T. akamura, A. Li ad K. Higuchi, "o- Orthogoal Multile Access (OMA) for Cellular Future Radio Access," 2013 IEEE 77th Vehicular Techology Coferece (VTC Srig), Dresde, 2013,

32 [28] M. Irfa, B. S. Kim ad S. Y. Shi, "A sectral efficiet satially modulated oorthogoal multile access for 5G," 2015 Iteratioal Symosium o Itelliget Sigal Processig ad Commuicatio Systems (ISPACS), usa Dua, 2015, [29] V. W.S. Wog, R. Schober, D. W. K. g, ad L.-C. Wag, "Key Techologies for 5G Wireless Systems," The Istitutio of Egieerig ad Techology, Cambridge Uiversity Press, Ar. 2017, ISB-10: [30] Z. Wei, D. W. K. g, J. Yua, ad H-M. Wag "Otimal Resource Allocatio for Power-Efficiet MC-OMA with Imerfect Chael State Iformatio," IEEE Tras. Commu., vol. 65, o. 9, , Set [31] F. Fag, H. Zhag, J. Cheg ad V. C. M. Leug, "Eergy efficiecy of resource schedulig for o-orthogoal multile access (OMA) wireless etwork," 2016 IEEE Iteratioal Coferece o Commuicatios (ICC), Kuala Lumur, 2016, [32] Y. Su, D. W. K. g, Z. Dig, ad R. Schober, "Otimal Joit Power ad Subcarrier Allocatio for Full-Dulex Multicarrier o-orthogoal Multile Access Systems," IEEE Tras. Commu., vol. 65, o. 3, , Mar [33] F. W. Murti ad S. Y. Shi, "User airig schemes based o chael quality idicator for ulik o-orthogoal multile access," 2017 ith Iteratioal Coferece o Ubiquitous ad Future etworks (ICUF), Mila, 2017, [34] W. Liag, Z. Dig, Y. Li ad L. Sog, "User Pairig for Dowlik o-orthogoal Multile Access etworks usig Matchig Algorithm," i IEEE Trasactios o Commuicatios, vol. PP, o. 99, [35] Z. Dig, P. Fa ad H. V. Poor, "User Pairig i o-orthogoal Multile Access Dowlik Trasmissios," 2015 IEEE Global Commuicatios Coferece (GLOBECOM), Sa Diego, CA, 2015, [36] X. Zhag, T. H. Chag, Y. F. Liu, C. She ad G. Zhu, "Ulik ad dowlik user airig i full-dulex multi-user systems: Comlexity ad algorithms," 2017 IEEE Iteratioal Coferece o Acoustics, Seech ad Sigal Processig (ICASSP), ew Orleas, LA, 2017,

33 [37] J. He ad Z. Tag, "Low-comlexity user airig ad ower allocatio algorithm for 5G cellular etwork o-orthogoal multile access," i Electroics Letters, vol. 53, o. 9, , [38] J. Guo, X. Wag, J. Yag, J. Zheg ad B. Zhao, "User Pairig ad Power Allocatio for Dowlik o-orthogoal Multile Access," 2016 IEEE Globecom Workshos (GC Wkshs), Washigto, DC, 2016, [39] J. He, Z. Tag ad Z. Che, "Fast ad efficiet user airig ad ower allocatio algorithm for o-orthogoal multile access i cellular etworks," i Electroics Letters, vol. 52, o. 25, , [40] Z. Q. Al-Abbasi ad D. K. C. So, "User-Pairig Based o-orthogoal Multile Access (OMA) System," 2016 IEEE 83rd Vehicular Techology Coferece (VTC Srig), ajig, 2016, [41] M. B. Shahab, M. F. Kader ad S. Y. Shi, "A Virtual User Pairig Scheme to Otimally Utilize the Sectrum of Uaired Users i o-orthogoal Multile Access," i IEEE Sigal Processig Letters, vol. 23, o. 12, , Dec [42] Y. Ma, X. Wag, Y. T. Guo ad M. Zhao, "Otimal ower allocatio for eergy efficiecy maximizatio i 5G dowlik multiuser systems," 2016 Progress i Electromagetic Research Symosium (PIERS), Shaghai, 2016, [43] J. Hajiour, J. M. iya ad D. W. K. g, "Eergy-Efficiet Resource Allocatio i Buffer-Aided Wireless Relay etworks," i IEEE Trasactios o Wireless Commuicatios, vol. 16, o. 10, , Oct [44] J. Zhag, L. Xiag, D. W. K. g, M. Jo ad M. Che, "Eergy Efficiecy Evaluatio of Multi-tier Cellular Ulik Trasmissio uder Maximum Power Costrait," i IEEE Trasactios o Wireless Commuicatios, vol. PP, o. 99, [45] X. Che, D. W. K. g ad H. H. Che, "Secrecy wireless iformatio ad ower trasfer: challeges ad oortuities," i IEEE Wireless Commuicatios, vol. 23, o. 2, , Aril [46] Q. Wu, M. Tao, D. W. Kwa g, W. Che ad R. Schober, "Eergy-Efficiet Resource Allocatio for Wireless Powered Commuicatio etworks," i IEEE Trasactios o Wireless Commuicatios, vol. 15, o. 3, , March

34 [47] J. A. Oviedo ad H. R. Sadjadour, "A ew OMA aroach for fair ower allocatio," 2016 IEEE Coferece o Comuter Commuicatios Workshos (IFOCOM WKSHPS), Sa Fracisco, CA, 2016, [48] G. ai, S. S. Das ad A. Chatterjee, "Low Comlexity User Selectio with Otimal Power Allocatio i Dowlik OMA," i IEEE Wireless Commuicatios Letters, vol. PP, o. 99, [49] T. Maglayev, R. C. Kizilirmak ad Y. H. Kho, "Otimum ower allocatio for o-orthogoal multile access (OMA)," 2016 IEEE 10th Iteratioal Coferece o Alicatio of Iformatio ad Commuicatio Techologies (AICT), Baku, 2016, [50] Z. Yag, W. Xu, C. Pa, Y. Pa ad M. Che, "O the Otimality of Power Allocatio for OMA Dowliks With Idividual QoS Costraits," i IEEE Commuicatios Letters, vol. 21, o. 7, , July [51] F. Kara, Ö. F. Gemici, İ. Hökelek ad H. A. Çira, "Otimal ower allocatio for DL OMA systems," th Sigal Processig ad Commuicatios Alicatios Coferece (SIU), Atalya, 2017, [52] J. Choi, "Joit Rate ad Power Allocatio for OMA with Statistical CSI," i IEEE Trasactios o Commuicatios, vol. PP, o. 99, [53] F. A. Khales ad G. A. Hodtai, "A evaluatio of the coverage regio for dowlik o-orthogoal Multile Access (OMA) based o Power Allocatio Factor," 2017 Ira Worksho o Commuicatio ad Iformatio Theory (IWCIT), Tehra, 2017, [54] M. S. Ali, H. Tabassum ad E. Hossai, "Dyamic User Clusterig ad Power Allocatio for Ulik ad Dowlik o-orthogoal Multile Access (OMA) Systems," i IEEE Access, vol. 4, , [55] X. Zhag, Q. Gao, C. Gog ad Z. Xu, "User Grouig ad Power Allocatio for OMA Visible Light Commuicatio Multi-Cell etworks," i IEEE Commuicatios Letters, vol. 21, o. 4, , Aril [56] W. Cai, C. Che, L. Bai, Y. Ji ad J. Choi, "Subcarrier ad ower allocatio scheme for dowlik OFDM-OMA systems," i IET Sigal Processig, vol. 11, o. 1, , [57] Z. Yag, Z. Dig, P. Fa ad. Al-Dhahir, "A Geeral Power Allocatio Scheme to Guaratee Quality of Service i Dowlik ad Ulik OMA Systems," i IEEE 28

35 Trasactios o Wireless Commuicatios, vol. 15, o. 11, , ov [58] J. Choi, "Power Allocatio for Max-Sum Rate ad Max-Mi Rate Proortioal Fairess i OMA," i IEEE Commuicatios Letters, vol. 20, o. 10, , Oct [59] P. Xu ad K. Cumaa, "Otimal Power Allocatio Scheme for o-orthogoal Multile Access With $\alha $ -Fairess," i IEEE Joural o Selected Areas i Commuicatios, vol. 35, o. 10, , Oct [60]. Otao, Y. Kishiyama ad K. Higuchi, "Performace of o-orthogoal access with SIC i cellular dowlik usig roortioal fair-based resource allocatio," 2012 Iteratioal Symosium o Wireless Commuicatio Systems (ISWCS), Paris, 2012, [61] L. Shi, B. Li ad H. Che, "Pairig ad Power Allocatio for Dowlik o- Orthogoal Multile Access Systems," i IEEE Trasactios o Vehicular Techology, vol. PP, o. 99, [62] W. Cai, C. Che, L. Bai, Y. Ji ad J. Choi, "Subcarrier ad ower allocatio scheme for dowlik OFDM-OMA systems," i IET Sigal Processig, vol. 11, o. 1, , [63] Y. Su, D. W. K. g, ad R. Schober, "Resource Allocatio for MCOMA Systems with Cogitive Relayig," acceted for resetatio, IEEE Globecom 2017 worksho, Sigaore [64] Y. Su, D. W. K. g, ad R. Schober, "Otimal Resource Allocatio for Multicarrier MISO-OMA Systems," i Proceedigs of the IEEE ICC [65] Z. Wei, D. W. K. g, ad J. Yua, "Joit Power ad Subcarrier Allocatio for Cooerative OMA Systems," ivited aer, i Proceedigs of the IEEE VTC [66] Z. Wei, J. Guo, D. W. K. g, ad J. Yua, "Fairess Comariso of Ulik OMA ad OMA," i Proceedigs of the IEEE VTC [67] Z. Wei, D. W. K. g, ad J. Yua, "Power-Efficiet Resource Allocatio for MC- OMA with Statistical Chael State Iformatio," i Proceedigs of the IEEE Globecom. 2016, Washigto, DC, USA, Dec [68] X. Li, W. Xu, Z. Feg, X. Li ad J. Li, "Matchig-Theory-Based Sectrum Utilizatio i Cogitive OMA-OFDM Systems," 2017 IEEE Wireless 29

36 Commuicatios ad etworkig Coferece (WCC), Sa Fracisco, CA, 2017, [69] Q. Zhag, J. Zhag, J. Feg ad Y. Yag, "eurodyamic Aalysis for a Class of oliear Fractioal Programmig," 2008 Iteratioal Coferece o Comutatioal Itelligece ad Security, Suzhou, 2008, [70] J. Gog ad F. You, "A efficiet global otimizatio algorithm for mixed-iteger oliear fractioal rograms with searable cocave terms," 2015 America Cotrol Coferece (ACC), Chicago, IL, 2015, [71] W. Dikelbach,"O oliear fractioal rogrammig." Maagemet sciece ,