A green radio resource allocation scheme for LTE-A downlink
systems with CoMP transmission
Wen-Ching Chung• Chung-Ju Chang•
Hsin-Ying Teng
Published online: 5 January 2014
Ó Springer Science+Business Media New York 2014
Abstract In this paper, we propose a green radio resource allocation (GRRA) scheme for LTE-advanced downlink systems with coordinated multi-point (CoMP) transmission to support multimedia traffic. The GRRA scheme defines a green radio utility function, which is composed of the required transmission power, assigned modulation order, and the number of coordinated transmission nodes. By maximizing this utility function, the GRRA scheme can effectively save transmission power, enhance spectrum efficiency, and guarantee quality-of-service requirements. The simulation results show that when the traffic load intensity is greater than 0.7, the GRRA scheme can save transmission power by more than 33.9 and 40.1 %, as compared with the conventional adaptive radio resource allocation (ARRA) scheme (Tsai et al. in IEEE Trans Wireless Commun 7(5):1734–1743,2008) with CoMP and the utility-based radio resource allocation (URRA) scheme (Katoozian et al. in IEEE Trans Wireless Commun 8(1):66–71, 2009) with CoMP, respectively. Besides, it enhances the system throughput by approximately 5.5 % and improves Jain’s fairness index for best effort users by more than 155 % over these two ARRA and URRA schemes.
Keywords Green radio Resource allocation CoMP LTE-A QoS Fairness
1 Introduction
Recently, the study of green radio communications has attracted considerable attention because of the high energy-efficiency demand in next-generation wireless systems [1–6]. When a base station (BS) is in an active mode, power supply, processing circuits, and air conditioning account for up to 60 % of the total energy consumption [1]. Therefore, reducing the energy consumption of the BS becomes an important issue. If the BS reduces its transmission power, the quality-of-service (QoS) experienced by users will be compromised. In green radio communications, one of the main design objectives is to reduce the amount of energy consumption while satisfying the QoS requirements [6].
Long term evolution-advanced (LTE-A) was introduced by the 3rd Generation Partnership Project (3GPP) to fulfill the requirements of IMT-Advanced for next-generation cellular systems [7–9]. In the LTE-A system, coordinated multi-point (CoMP) transmission is adopted to save the transmission power, increase the coverage of high data rates, enhance the cell-edge throughput, and/or increase the overall system throughput [10]. CoMP transmission indi-cates that the transmission is carried out among multiple geographically separated transmission nodes, which com-prise a CoMP cooperating set. Therefore, the radio resource allocation (RRA) problem in the LTE-A system with CoMP transmission would involve multiple degrees of freedom in space, time, and frequency; thus, the RRA problem would become very challenging, particularly when the QoS requirement guarantee for multimedia traffic is considered.
The LTE-A downlink system adopts orthogonal fre-quency division multiple access (OFDMA) in the physical layer [8]. OFDMA can eliminate the intra-cell interference due to orthogonality between subcarriers. However, in a W.-C. Chung (&) C.-J. Chang H.-Y. Teng
Department of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu 300, Taiwan, ROC
e-mail: [email protected] DOI 10.1007/s11276-013-0684-8
multi-cell OFDMA system with the frequency reuse factor being equal to one, a user would experience severe inter-cell interference with a resulting low signal-to-noise-and-interference ratio (SINR). CoMP transmission is an effi-cient technique to mitigate inter-cell interference for users. By allowing transmission nodes in neighbor cells to transmit the same signal on the same subcarrier, CoMP transmission transforms inter-cell interference into a valu-able signal.
In the literature, two types of schemes have been pre-sented to implement downlink CoMP transmission: coor-dinated scheduling/beamforming (CS/CB) and joint processing (JP) [11–13]. In the CS/CB scheme, the user scheduling/beamforming decisions are coordinated within the CoMP cooperating set [11]. By using beamforming weights, the interference to other users in different cells can be efficiently mitigated [12]. The JP scheme can also be categorized as joint transmission (JT) and dynamic cell selection (DCS) [13]. In the JT scheme, the data for the selected user is delivered simultaneously from multiple transmission nodes to improve the received signal quality and cancel the active interference for other users. However, in the DCS scheme, the data for the selected user is transmitted from only one transmission node at a time. The other nodes in the CoMP cooperating set are then muted to avoid generating the interference to the served user. In order to effectively utilize the radio resource and enhance the received signal quality, we consider that downlink CoMP transmission is implemented by the JT scheme in this paper.
The study of CoMP transmission has recently attracted considerable attention [14–17]. On the basis of a zero forcing beamforming technique, To¨lli et al. [14] proposed an RRA method for cooperative MIMO–OFDM systems so as to implement the bit and power loading algorithms in practical systems. Gao et al. proposed a cooperating set construction algorithm [15] to determine the best groups of transmission nodes and coordinated users by evaluating channel orthogonality among users. Fodor et al. [16] studied the problems of setting SINR targets and allocating transmission powers for CoMP systems to obtain a better trade-off between the fairness and the multi-cell throughput performance. Moreover, Liu et al. [17] proposed a trans-mission scheme with the joint proportional fairness algo-rithm for CoMP-based systems. This scheme does not allocate the dedicated frequency band to cell-edge users and can obtain the best frequency diversity gain. However, these proposed schemes only considered a single service class. In modern mobile Internet systems, multimedia traffic support is an essential requirement.
On the other hand, the QoS requirement guarantee is an important performance consideration in the design of RRA schemes for wireless communication systems [18–21]. The
QoS requirements contain the required bit error rate (BER), the minimum required transmission rate, the maximum packet delay tolerance, and the maximum packet dropping ratio. On the basis of the fixed priority for each traffic, Yu et al. [18] proposed a cross-layer design for MIMO–OF-DMA systems to guarantee QoS requirements. Tsai et al. [19] proposed an adaptive radio resource allocation (ARRA) scheme for downlink OFDMA/SDMA systems. The ARRA scheme dynamically assigns a high priority value to the urgent user and allows a non-real time (NRT) user to be assigned a higher priority value than a real time (RT) user when an urgent need arises. In [20] and [21], utility-based radio resource allocation (URRA) schemes were proposed to maximize the utility function, which is designed on the basis of the QoS, rate, and fairness factors. However, these proposed schemes were designed for a single transmission node only. For multi-transmission nodes, the design of RRA schemes becomes more com-plicated given that they must determine the transmission nodes for users and consider the interference to/from other transmission nodes.
In this paper, we propose a green radio resource allo-cation (GRRA) scheme for LTE-A downlink systems with CoMP transmission to support multimedia traffic. The GRRA scheme first defines a service priority and a min-imum number of transmission bits for each user on the basis of the degree of the user’s urgency. According to a green radio utility function, it then assigns an appropriate CoMP transmission type, modulation order, and sub-channel to users in a priority value sequence. By coordi-nating the transmission power of each transmission node, the GRRA scheme allows the transmission nodes to simultaneously serve multiple users at the same subchan-nel. Therefore, the proposed scheme can mitigate the interference, thus saving transmission power and increas-ing the system throughput, and can achieve a high Jain’s fairness index [22] under the QoS requirement guarantee. The simulation results show that when the traffic load intensity is greater than 0.7, the GRRA scheme can save transmission power by more than 33.9 % as compared with the adaptive radio resource allocation (ARRA) scheme [19] and by more than 40.1 % as compared with the utility-based radio resource allocation (URRA) scheme [20]. Moreover, the GRRA scheme can achieve a Jain’s fairness index for a best-effort (BE) service that is approximately 155 % greater and a system throughput that is approximately 5.5 % higher than both the ARRA and the URRA schemes.
The remainder of the paper is organized as follows. The system model is introduced in Sect.2. Section3 describes the design of the GRRA scheme. Section 4 presents the simulation results for the performance analysis of the GRRA scheme. Finally, Sect.5 concludes the paper.
2 System model
In the LTE-A downlink system with CoMP transmission, each cell is partitioned into three sectors and one evolved Node B (eNB) is located at the center of the cell. As shown in Fig.1, the eNB is equipped with a 120° directional antenna for each sector. Assume that the frequency reuse factor within the cell is 3, and the interference among sectors in the same cell is then ignored [23]. There are K pieces of single-antenna user equipment (UE) uniformly distributed in the sector. There are also two pieces of remote radio equipment (RRE) in each sector, denoted by RRE1 and RRE2, located at the cell edge of the sector with distance d to the eNB and distanced2 to the sector boundary. Each RRE has a single omnidirectional antenna and is connected with eNB via an optical fiber. In the sector, the eNB and two RREs are operated at the same fre-quency band for the downlink CoMP transmission.
We adopt the JT scheme for CoMP transmission and define a CoMP cooperating set in the sector, denoted by X, which contains RRE1, RRE2, and eNB, numbered as nodes 1, 2, and 3, respectively. Thus, there are totally eight downlink CoMP transmission types in the LTE-A system. Table1lists the eight CoMP transmission types, where Xj is the set of transmission nodes for CoMP transmission type j, 0 B j B 7. Transmission type 0 denotes that no transmission node is involved and thus implies no data transmission; transmission type 7 indicates that the data transmission has RRE1, RRE2, and eNB engaged.
Assume that each sector has N subchannels. In the LTE-A standard [24], each subchannel is composed of 12 sub-carriers. Moreover, the frame time is defined as 10 ms and is divided into ten subframes. Each subframe consists of two slots, and each slot has seven OFDM symbols. The basic resource unit for allocation is one subchannel over one subframe. The channel state information (CSI) corre-sponding to the downlink channels from the eNB and RRE
to the UE are reported by the UE to the eNB. The proposed green radio resource allocation (GRRA) scheme is carried out subframe by subframe at the eNB for each sector. In order to improve the spectrum efficiency for CoMP trans-mission, we consider that the CoMP transmission nodes can simultaneously serve at most three UEs at the same resource unit. The allocation decision and the user data are delivered from the eNB to the RREs by the optical fiber.
Let hk,ni be the channel gain from transmission node i to
UE k on subchannel n. Assume that the coherent time of the wireless channel is longer than the subframe duration. Therefore, the channel gain can be regarded as constant within a subframe. Denote the CoMP transmission type for UE k on subchannel n by sk,n, thus sk;n2 f0; 1; . . .; 7g, and assume that the cooperative signals can be coherently combined at the UE [25]. Let Un be the set of UE that multiplex on subchannel n. Thus, the received signal on subchannel n at UE k, denoted by yk,n, is given by
yk;n¼ X i2Xsk;n hik;n ffiffiffiffiffiffiffipi k;n q xk;n þ X i02XX sk;n X k02U n;k06¼k hik;n0 ffiffiffiffiffiffiffiffi pi0 k0;n q xk0;nþ zk;n; ð1Þ
where pk,ni is the allocated transmission power to UE k on
subchannel n by node i, xk,nis the data symbol, Xsk;nis the
set of nodes of transmission type sk;n;X Xsk;n is the set
containing elements that belong to X but not Xsk;n, and zk,n
is the additive white Gaussian noise (AWGN) with zero mean and variance r2. The first term on the right-hand side of (1) is the desired signal, whereas the second term is the interference signal from other transmission nodes to UE k on subchannel n in the same sector. Suppose that pk,ni is
designed in accordance with the pre-maximum-ratio-combining (pre-MRC) scheme [26,27], which is given by
pik;n¼ jh i k;nj 2 P i2Xsk;njhik;nj 2 2 SINR k Ik;nþ r2 ; ð2Þ
Fig. 1 The LTE-A downlink system with CoMP transmission
Table 1 CoMP transmission types Transmission
type j
Transmission nodes Xj¼ fijnode i in type jg
0 None X0¼ f;g
1 RRE1 X1¼ f1g
2 RRE2 X2¼ f2g
3 eNodeB X3¼ f3g
4 RRE1 & RRE2 X4¼ f1; 2g
5 RRE1 & eNodeB X5¼ f1; 3g
6 RRE2 & eNodeB X6¼ f2; 3g
and Ik;n¼ X i02XX sk;n X k02U n;k06¼k pik00;njhi 0 k;nj 2 ; ð3Þ
where SINRk*is the minimum required SINR for UE k and
Ik,nis the interference power from other nodes on UE k at
subchannel n. The SINRk*with M-QAM modulation can be
obtained by
SINRk¼ lnð5BER
kÞ
1:5 ðM 1Þ; ð4Þ
where BERk*is the BER requirement for UE k. Therefore,
the total allocated power for UE k on subchannel n in the sector, denoted by Pk,n, is given by
Pk;n¼ X
i2Xsk;n
pik;n: ð5Þ
Further, the LTE-A system can support three service classes: real-time (RT), non-real-time (NRT), and best-effort (BE) service classes. Each service class has different QoS requirements. For the RT service, the QoS require-ments are the required BER, the maximum packet delay tolerance, and the maximum packet dropping ratio. For the NRT service, the QoS requirements include the required BER and the minimum required transmission rate. For the BE service, the QoS requirement is the required BER only. Let us denote Rk*, Dk*, and PDk*as the minimum required
transmission rate, the maximum packet delay tolerance, and the maximum packet dropping ratio for UE k, respectively. There are four kinds of traffic types consid-ered in the LTE-A system. The RT service has voice and video traffic; the NRT service has HTTP traffic; and the BE service has FTP traffic. Each traffic has one individual queue at the eNB. Suppose that the queue is large enough to store all the arriving packets. The traffic packet is stored in its own queue in a first-in first-out manner. Packet dropping occurs only when the packet delay time exceeds its maximum packet delay tolerance, but will not occur because of the overflow of buffer occupancy. Retransmis-sion due to erroneous transmisRetransmis-sion is not considered in this study.
3 Green radio resource allocation (GRRA) scheme
The green radio resource allocation (GRRA) scheme first formulates the RRA problem of the LTE-A downlink system with CoMP transmission into utility-based optimi-zation equations. Then, it employs a priority and bit assignment (PBA) algorithm and a priority-based resource allocation (PRA) algorithm to heuristically solve the opti-mization equations.
3.1 Problem formulation
Denote mk,n as the number of bits for UE k with M-QAM
modulation on subchannel n, where mk,n= log2 M and
mk,n= {0, 2, 4, 6}, 1 B k B K, 1 B n B N. If mk,n= 0,
the implication is that subchannel n is not allocated to UE k in this subframe. If mk,n = 2, 4, or 6, the implication is
that subchannel n is assigned to UE k and the data are modulated in the modulation order of QPSK, 16-QAM, or 64-QAM, respectively. Thus the allocated transmission bits of UE k in this subframe, denoted by Rk, are given by
Rk¼ XN n¼1
q mk;n; ð6Þ
where q is the number of OFDM symbols in one basic allocation unit (12 subcarries over 14 symbols), and q = 168.
Define the green radio utility function for UE k on sub-channel n, denoted by Gk,n; a function of the modulation order
index mk,nand the CoMP transmission type sk,n. Denote M as
the modulation order assignment vector, M¼ ½m1;1; . . .; m1;N; . . .; mk;1; . . .; mk;N; . . .; mK;1; . . .; mK;N, and S as the CoMP transmission type assignment vector, S¼ ½s1;1; . . .; s1;N; . . .; sk;1; . . .; sk;N; . . .; sK;1; . . .; sK;N. M and S will be the solution of the utility-based optimization equations for the LTE-A system with CoMP transmission. The GRRA scheme aims to enhance the system throughput and save the trans-mission power. Therefore, Gk,nis designed as
Gk;n¼ mk;nþ ffiffiffiffiffiffiffiffiffi1 jXsk;nj p Pk;n; if Pk;n[ 0;jXsk;nj [ 0; 0; otherwise; ( ð7Þ
wherejXsk;nj is the number of elements in Xsk;nand Pk,nis in
mW. When a transmission occurs, Pk,n[ 0 andjXsk;nj [ 0,
and we have Gk,n[ 0. It can be seen in (7) that mk,nis the
dominant term in Gk,n. A piece of UE with a higher
modulation order will have a larger Gk,nand thus a higher
priority to be served. This implies that the objective of the system throughput enhancement can be achieved. Further, if there are some pieces of UE with the same modulation order, the piece of UE that needs fewer transmission nodes and less transmission power will be served earlier. This implies that the objective of transmission power saving can be attained. It should be noted that Gk,ndoes not consider
the QoS requirements of UE because we do not oversatisfy the QoS requirements of the UE.
Consequently, the utility-based optimization equations of the GRRA scheme for the LTE-A downlink system with CoMP transmission are formulated as follows:
ðM; SÞ ¼ arg max M;S XK k¼1 XN n¼1 Gk;n; ð8Þ
subject to the QoS requirement constraints: ðiÞ Dk Dkif UE k is with RT service, and
ðiiÞ Rk Rkif UE k is with NRT or BE service; 1 k K;
and the system constraints:
ðiÞ X K k¼1 sgnðmk;nÞ 3; 8n; ðiiÞ X K k¼1 XN n¼1 pik;n Pi T;8i; ðiiiÞ Ik;n Ith;8k 2 Un; n; ðivÞ Rk Qk q q; 8k;
where Dk is the packet delay of UE k; sgnðÞ is the sign
function,dxe is the smallest integer greater than x, Qk is
the buffer occupancy of UE k, PTi is the maximum
trans-mission power budget at node i, Ik,nis the interference on
UE k at the subchannel n given in (3), and Ithis the
max-imum allowed interference on the UE. The QoS require-ment constraints are simply set to make the GRRA scheme fulfill the QoS requirement guarantee and achieve a high Jain’s fairness index. The system constraint (i) is the channel allocation constraint; it is set because each sub-channel in the same sector is assumed to be possibly allocated to at most three pieces of UEs. The system constraint (ii) is the total power constraint; it is set because the total power allocation for each OFDMA symbol has a limitation for downlink data transmission at each trans-mission node. The system constraint (iii) is the interference limitation constraint; it is set to reflect the co-channel interference limitation for resource allocation. In this study, we set Ith= 0.1 [28]. The system constraint (iv) is the
buffer occupancy constraint; it is set because the allocated transmission bits to UE k should not be greater than its buffer occupancy.
However, it is complicated and intractable to obtain the optimal solution of the optimization equations given in (8) by an exhaustive search. The GRRA scheme is therefore designed to contain a priority and bit assignment (PBA) algorithm and a priority-based resource allocation (PRA) algorithm to heuristically determine the suboptimal set of assignment vectors (M*, S*) in (8).
3.2 Priority and bit assignment (PBA) algorithm
The PBA algorithm determines the service priority value and the minimum required number of transmission bits assigned to UE k, 1 B k B K, according to its residual time before violating the QoS requirements.
Denote Vk as the residual lifetime of the head-of-line
(HOL) packet of UE k, which indicates the number of subframes remaining for the HOL packet not to violate the QoS requirements. Vkis then designed as
Vk¼
Dk Dk; if UE k is with the RT service,
BkþB0k
R
k Tk
j k
; if UE k is with the NRT or BE service; (
ð9Þ whereb x c is the largest integer smaller than x, Bkis the
number of residual bits in the HOL packet of UE k, Tkis
the time duration that the packet has been buffered in the queue of UE k, and B0kis the number of transmitted bits of
UE k in Tk. The smaller the value of Vk, the more urgent
UE k is. For the RT service, Vkis intuitively defined from
its delay requirement. For the NRT service, the average transmission rate, denoted by Rk, should be greater than or equal to the QoS requirement of the minimum transmission rate. Thus, Vk is derived from the inequality (Bk? B0k)/
(Vk? Tk) C Rk *
. On the other hand, for the BE service, we consider fairness to avoid the deprivation of BE UEs due to bad channel conditions. The Rk*for the BE service is then
set to the maximum average transmission rate among the BE UEs; that is, Rk¼ maxk02U
BERk0, where UBEis the set of
UE with the BE service.
The priority value for UE k, denoted by uk, is designed
as uk¼ 1þ Dk VkþDk ak; if Vk[ 0; 2 ak; otherwise; ( ð10Þ
where akis the default priority constant for UE k. We set
ak= 3, 2, or 1 for UE k with the RT, NRT, or BE service,
respectively, given that the RT service, which has a strict delay requirement, should have the highest priority value and the BE service, which is background traffic, should have the lowest priority value. It can be seen from (10) that the more urgent (less Vk) UE k is, the higher is the priority
value of UE k.
Subsequently, the minimum required number of trans-mission bits allocated to UE k at the current subframe to avoid violating QoS requirements, denoted by ^Rk, is given as
^ Rk¼ Bk Vkq l m q; if Vk Vth; Bk q l m q; otherwise; 8 < : ð11Þ
where Vth is the threshold for Vk. If Vk is below Vth, it
means that UE k is very urgent and its HOL packet should be completely transmitted at the current subframe. There-fore, we set ^Rk¼ dBqke q. Otherwise, the HOL packet should be equally delivered over the next Vk subframes,
and we set ^Rk¼ dVBk
kqe q. If the value of Vthis set to 1,
current subframe. In order to release the system load, in this study, we set Vth= 3.
3.3 Priority-based resource allocation (PRA) algorithm
The PRA algorithm allocates an appropriate CoMP mission type, modulation order, subchannel, and trans-mission power to the UE with the highest priority value to maximize the green radio utility function.
In order to guarantee the QoS requirements, the PRA algorithm serves the hightest-priority UE first. Let Uc be the set of backlogged UE with the highest priority and having bits to be transmitted at the current subframe and Nf be the set of free subchannels. The highest-priority UE
with its best-condition channel, denoted by (k*, n*), is obtained by
ðk; nÞ ¼ arg max
k2Uc;n2Nf
Gk;n: ð12Þ
It should be noted that when the maximal Gk;n is found,
the CoMP transmission type sk;n, the modulation order
mk;n, and the subchannel n* are determined for UE k*.
Here, the assigned modulation order mk;n should satisfy
the system constraint (iv). Before the radio resource is allocated to UE k*, the system constraints (ii) and (iii) are checked. If the power budget at the transmission node
i; i2 Xsk;n, is sufficient for delivering the packet to UE
k* on subchannel n* and the interference to the other selected UE on subchannel n* is less than Ith, the radio
resource is allocated to UE k*at this subframe. The set of the used transmission nodes on the subchannel n*, denoted by Xu;n, is then updated by Xu;n ¼ Xu;nþ fiji 2 Xs
k;ng.
IfjXu;nj ¼ 3, subchannel n*is removed from Nf because
of the system constraint (i). Moreover, the queue length of UE k* and the consumed power at transmission node i, denoted by Pk*, are updated after this allocation. The PRA
algorithm is repeated until all radio resources are allocated to all the pieces of UE or no candidate UE exists. The pseudocode of this PRA algorithm is shown in the ‘‘Appendix’’.
4 Simulation results
4.1 Simulation environment
We use Matlab as the simulation tool. In this simulation, the parameters of the considered LTE-A downlink system are set to be compatible with the 3GPP Evolved Universal Terrestrial Radio Access (E-UTRA) standards [24]. The system bandwidth in each sector is 5 MHz. The maximum system transmission rate for eNB or RRE is equal to 25.2 Mbps, which is obtained when each subchannel delivers
data with the highest modulation order of 64-QAM. The path loss from RRE is modeled as 36.7log(dR) ?
132.8 ? 26log(fc) dB, where dR is the distance between
RRE and UE in kilometers and fcis the carrier frequency in
gigahertz [10]. The shadowing from RRE is assumed to be lognormal with zero mean and a standard deviation of 10 dB. The path loss from the eNB is modeled as 39.09log(dB) ? 130.8 ? 20log(fc) dB, where dB is the
distance between the eNB and the UE in kilometers. The shadowing from the eNB is lognormal with zero mean and a standard deviation of 6 dB. Moreover, the multipath channel is assumed to be a six-tap Rayleigh-faded path with an exponential power delay profile [29]. The other parameters of the LTE-A system are listed in Table2
[25].
The LTE-A system can accommodate four traffic types: voice traffic of the RT service, video traffic of the RT service, HTTP traffic of the NRT service, and FTP traffic of the BE service. The voice traffic is modeled as an ON– OFF model [30]. The video traffic is composed of a sequence of streaming video frames [31]. Each video frame is composed of eight slices (packets). The HTTP traffic is modeled as a sequence of page downloads [31]. Both the main and the embedded object sizes follow a truncated lognormal distribution. The FTP traffic is modeled as a sequence of file downloads [31], where the file size follows a truncated lognormal distribution. The distribution parameters for voice, video, HTTP, and FTP traffic can be found in [30] and [31], and thus details are omitted here. The QoS requirements for each traffic type are listed in Table3 [30].
4.2 Performance evaluation
The proposed GRRA scheme will be compared with the conventional adaptive radio resource allocation (ARRA) scheme [19] and the utility-based radio resource alloca-tion (URRA) scheme [20]. Here, the ARRA and URRA schemes with CoMP transmission are also considered,
Table 2 Parameters of the LTE-A system
Parameters Assumption
Cell radius 1,000 m
Carrier frequency, fc 2 GHz Number of subchannels, N 25 Total eNB Tx power, PT3 43 dBm Total RRE Tx power, PT1(PT2) 30 dBm
Antenna pattern AH(/) = - min (12 (///3dB)2, Am) /3dB= 70°, Am= 25 dB
Antenna gain 14 dBi
where only the transmission node with the largest channel gain is chosen for CoMP transmission. In the ARRA scheme, user priority is defined as the minimum number of bits required to transmit at the current sub-frame so as to guarantee QoS requirements. Because the NRT packet size is greater than the RT packet size, the NRT UE might have a higher priority than the RT UE when it is urgent. In the URRA scheme, the utility value is designed according to the QoS, rate, and fairness factors. For the URRA scheme to have the best perfor-mance on QoS measurements and system throughput, the fairness factor is set to one. The QoS factor for the NRT UE k is modified to be 1þ Rk
R
kþ Rk
such that the NRT service has a higher QoS factor than the BE service. In the following figures, the traffic load intensity of the system is defined as
Traffic Load Intensity
¼total average arrival rates of all traffics maximum system transmission rate ;
ð13Þ
and the traffic load intensity is varied from 0.1 to 0.9. Further, assume that each traffic type has the same number of pieces of UE.
Figure2 shows the average transmission power con-sumed versus the traffic load intensity. It is found that when the traffic load intensity is greater than 0.7, the GRRA scheme can save transmission power by more than 33.9 and 40.1 % as compared with the ARRA and URRA schemes with CoMP, respectively. Moreover, the ARRA (URRA) scheme with CoMP saves transmission power by more than 57.8 % (56.6 %) as compared with the original ARRA (URRA) scheme without CoMP. The reasons for this are as follows. The ARRA (URRA) scheme with CoMP transforms inter-cell interference into more valu-able signals and thus consumes less transmission power to achieve the required SINR than the original ARRA (URRA) scheme without CoMP. By using the green radio utility function of (7), the GRRA scheme selects the UE with a higher modulation order and less transmission
power. Moreover, it allows three transmission nodes to serve one piece of UE by the system constraint (i) and considers the interference among transmission nodes by the system constraint (iii). Thus, it can most efficiently mitigate inter-cell interference and consume the lowest average transmission power. On the other hand, the URRA and ARRA schemes with CoMP are assumed to choose only the transmission node with the largest channel gain to deliver packets. This might cause some interference to the other pieces of UE on the same sub-channel and then more transmission power might be needed to achieve the required SINR.
Figure3 depicts the system throughput versus the traffic load intensity. It can be seen that the RRA schemes with CoMP can enhance the system throughput by more than 80 % as compared with those without CoMP. Although the GRRA scheme and the URRA and ARRA schemes with CoMP have almost similar system throughput, the GRRA scheme has a 5.5 % higher system throughput than the other two schemes. The reasons for this are as follows. As pieces of RRE are established, the maximum system transmission rate for the LTE-A system is increased. Thus, the GRRA scheme and the URRA and ARRA schemes with CoMP have a higher system throughput than the URRA and ARRA schemes without CoMP. Although the GRRA scheme considers fairness for the BE service in the priority value assignment by (9) and (10), it mitigates some co-channel interference among the pieces of UE by the system constraint (iii) so as to increase the received SINR. Further, it allocates the subchannel with the highest modulation order to the UE by (12). Hence, the GRRA scheme can enhance the sys-tem throughput. On the other hand, the URRA and ARRA schemes with CoMP only choose the transmission mode with the largest channel gain. This might increase inter-ference among the pieces of UE and then reduce the
Table 3 The QoS requirements for each traffic type Voice (RT) Video (RT) HTTP (NRT) FTP (BE) Required BER 10-3 10-4 10-6 10-6 Maximum packet delay tolerance 40 ms 10 ms N/A N/A Maximum packet dropping ratio 1 % 1 % N/A N/A Minimum required transmission rate
N/A N/A 100 kbps N/A
N/A not applicable
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5
Traffic Load Intensity
Average Transmission Power (mW/bit)
GRRA URRA w CoMP ARRA w CoMP URRA w/o CoMP ARRA w/o CoMP
system throughput. From the simulation results of Figs.2
and3, it can be seen that the proposed GRRA scheme can efficiently save transmission power and enhance the sys-tem throughput for the LTE-A syssys-tem with multimedia traffic.
Figure4depicts the packet dropping ratio of voice and video traffic. The GRRA and URRA with CoMP schemes can guarantee the dropping ratio requirement for voice traffic for a traffic load intensity up to 0.9 and for video traffic for a traffic load intensity up to 0.8. The ARRA with CoMP scheme has the largest voice and video packet dropping ratios. The phenomena are explained below. Voice traffic has a larger delay tolerance than video traffic, and the number of packets for video traffic is greater than that for voice traffic. Therefore, the dropping ratio requirement for voice traffic can be fulfilled more easily than that for video traffic. In the GRRA and the URRA with CoMP schemes, the RT traffic always has the highest priority value as given in (10) and can be served first. However, the ARRA with CoMP scheme allows the NRT traffic to have a higher priority than the RT traffic when the former becomes urgent, and then the NRT traffic can be first served. Thus, as the number of NRT packets increases, the RT packets cannot be delivered within the delay requirement, and the ARRA with CoMP scheme has the largest RT-service packet drop ratio among the three compared schemes.
Figure5 presents the mean packet delay of voice and video traffic. All three schemes satisfy the packet delay requirements of voice and video traffic. It can be found that the ARRA (URRA) with CoMP scheme has the largest (smallest) mean packet delay, and the GRRA scheme lies in the middle. The reasons for this are the same as those given in Fig.4. In the GRRA scheme, the HOL packet for the UE is equally delivered over the next residual subframes by (11) as it is not urgent.
However, in the URRA scheme, the entire HOL packet is delivered to the UE as it is served. Therefore, the GRRA scheme has a larger RT-service packet delay than the URRA scheme.
Figure6 shows the average transmission rate of HTTP traffic. The ARRA scheme with CoMP has the highest transmission rate of HTTP traffic. The GRRA and the URRA with CoMP schemes can guarantee the minimum transmission rate requirement for HTTP traffic up to a traffic load intensity of 0.81. The reason for this is as follows. In the GRRA and the URRA with CoMP schemes, the NRT service always has a smaller priority value than the RT service. Therefore, as the traffic load is heavy, there is not enough radio resource for HTTP traffic to attain its minimum transmission rate requirement as given in (10). On the other hand, in the ARRA with CoMP scheme, the user priority is defined as the mini-mum number of bits required to transmit at the current subframe. Given that the NRT packet size is greater than the RT packet size, the NRT service has a higher priority than the RT service when it becomes urgent. Therefore, the ARRA with CoMP scheme has the largest NRT-ser-vice transmission rate.
Figure7 presents Jain’s fairness index for FTP traffic. Jain’s fairness index is defined as [22]
0 0.2 0.4 0.6 0.8 1 5 10 15 20 25 30 35 40
Traffic Load Intensity
System Throughput (Mbps)
GRRA URRA w CoMP ARRA w CoMP URRA w/o CoMP ARRA w/o CoMP
Fig. 3 System throughput
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Traffic Load Intensity
Voice Packet Dropping Ratio (%)
(a) GRRA URRA w CoMP ARRA w CoMP 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10
Traffic Load Intensity
Video Packet Dropping Ratio (%)
(b)
GRRA URRA w CoMP ARRA w CoMP
Packet Dropping Ratio Requirement Packet Dropping Ratio Requirement
Jain Fairness Index¼ PR k ð Þ2 KPR2k : ð14Þ
The GRRA scheme improves Jain’s fairness index for FTP traffic by more than 155 % as compared with the URRA and ARRA schemes with CoMP. This is because the GRRA scheme assigns higher priority values to BE UE with lower transmission rates by (9) and (10) so as to attain a higher Jain’s fairness index. On the other hand, the URRA and ARRA schemes with CoMP allocate more
radio resources to BE UE with better channel conditions such that BE UE with worse channel conditions cannot obtain enough radio resource.
5 Conclusions
In this paper, we propose a green radio resource allocation (GRRA) scheme for LTE-A downlink systems with CoMP transmission to support multimedia traffic. The objective of the GRAA scheme is to maximize the green radio utility value, which comprises system transmission power, mod-ulation order, and the number of coordinated transmission nodes. By coordinating the transmission power of each transmission node, it allows transmission nodes to simul-taneously serve multiple users at the same subchannel. The GRAA scheme employs a priority and bit assignment (PBA) algorithm and a priority-based resource allocation (PRA) algorithm. The PBA algorithm defines the service priority value and the minimum number of transmission bits for each piece of UE according to the degree of the UE’s urgency. According to the green radio utility func-tion, the PRA algorithm allocates the radio resource to UE in the priority value order. Therefore, the GRRA scheme can mitigate the interference, thus saving transmission power and increasing the system throughput, and can achieve a high Jain’s fairness index under the QoS requirement guarantee. The simulation results show that when the traffic load intensity is greater than 0.7, the GRRA scheme can save transmission power by more than 33.9 % as compared with the conventional adaptive radio resource allocation (ARRA) scheme [19] with CoMP and by more than 40.1 % as compared with the conventional utility-based radio resource allocation (URRA) scheme [20] with CoMP. Moreover, the GRRA scheme can improve Jain’s fairness index for FTP traffic by more than
0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 30 35
Traffic Load Intensity
Mean Voice Packet Delay (ms)
(a) GRRA URRA w CoMP ARRA w CoMP 0 0.2 0.4 0.6 0.8 1 0 2 4 6 8 10
Traffic Load Intensity
Mean Video Packet Delay (ms)
(b) GRRA URRA w CoMP ARRA w CoMP ⋅ maximum delay tolerance of voice packet = 40 ms ⋅ maximum delay tolerance of video packet = 10 ms
Fig. 5 Mean packet delay of a voice traffic and b video traffic
0 0.2 0.4 0.6 0.8 1 0 100 200 300 400 500 600 700 800 900
Traffic Load Intensity
Average Transmission Rate (kbps)
GRRA URRA w CoMP ARRA w CoMP
Minimum Rate Requirement
Fig. 6 Average transmission rate of HTTP traffic
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Traffic Load Intensity
Jain Fairness Index
GRRA URRA w CoMP ARRA w CoMP
155 % as compared with the URRA and ARRA schemes with CoMP.
Acknowledgments The authors would like to give thanks the anonymous reviewers for their suggestions to improve the presentation of the paper. The work was supported by National Science Council
(NSC), Taiwan, under contract number NSC 100-2221-E-009-102-MY3, and the Ministry of Education, Taiwan, under the ATU plan. Appendix
References
1. Niu, Z., Wu, Y., Gong, J., & Yang, Z. (2010). Cell zooming for cost-efficient green cellular networks. IEEE Communications Magazine, 48(11), 74–79.
2. Chen, Y., Zhang, S., Xu, S., & Li, G. Y. (2011). Fundamental trade-offs on green wireless networks. IEEE Communications Magazine, 49(6), 30–37.
3. Han, C., Harrold, T., Armour, S., Krikidis, I., Videv, S., Grant, P. M., et al. (2011). Green radio: Radio techniques to enable energy-efficient wireless networks. IEEE Communications Magazine, 49(6), 46–54.
4. Bogucka, H., & Conti, A. (2011). Degrees of freedom for energy savings in practical adaptive wireless systems. IEEE Communi-cations Magazines, 49(6), 38–45.
5. Wang, B., Wu, Y., Han, F., Yang, Y. H., & Liu, K. J. R. (2011). Green wireless communications: A time-reversal paradigm. IEEE Journal on Selected Areas in Communications, 29(8), 1698–1710. 6. Ismail, M., & Zhuang, W. (2011). Network cooperation for energy saving in green radio communications. IEEE Wireless Communications, 18(5), 76–81.
7. Parkvall, S., & Astely, D. (2009). The evolution of LTE towards IMT-advanced. Journal of Communications, 4(3), 146–154. 8. 3GPP TR36.913. (2008). Requirements for further advancements
for EUTRA (LTE-advanced). 3rd Generation Partnership Project, Tech. Rep., June 2008.
9. Ghosh, A., Ratasuk, R., Mondal, B., Mangalvedhe, N., & Thomas, T. (2010). LTE-advanced: Next-generation wireless broadband technology. IEEE Transactions on Wireless Communications, 17(3), 10–22.
10. 3GPP TR36.814 (2010). Evolved universal terrestrial radio access (E-UTRA); further advancements for E-UTRA physical layer aspects. 3rd Generation Partnership Project, Tech. Rep., Mar. 2010.
11. Andrews, J. G., Choi, W., & Heath, R. W. Jr. (2007). Over-coming interference in spatial multiplexing MIMO cellular networks. IEEE Transactions on Wireless Communications, 14(6), 95–104.
12. Karakayali, M. K., Foschini, G. J., & Valenzuela, R. A. (2006). Network coordination for spectrally efficient communications in cellular systems. IEEE Transactions on Wireless Communica-tions, 13(4), 56–61.
13. Sawahashi, M., Kishiyama, Y., Morimoto, A., Nishikawa, D., & Tanno, M. (2010). Coordinated multipoint transmission/reception techniques for LTE-advanced. IEEE Transactions on Wireless Communications, 17(3), 26–34.
14. To¨lli, A., Codreanu, M., & Juntti, M. (2008). Cooperative MIMO–OFDM cellular system with soft handover between dis-tributed base station antennas. IEEE Transactions on Wireless Communications, 7(4), 1428–1440.
15. Gao, X., Li, A., & Kayama, H. (2009). Low-complexity downlink coordination scheme for multi-user CoMP in LTE-advanced system. IEEE PIMRC.
16. Fodor, G., Johansson, M., & Soldati, P. (2009). Near optimum power control under fairness constraints in CoMP systems. IEEE GLOBECOM.
17. Liu, J., Chang, Y., Pan, Q., Zhang, X., & Yang, D. (2010). A novel transmission scheme and scheduling algorithm for CoMP-SU–MIMO in LTE-A system. IEEE VTC-Spring.
18. Yu, J., Cai, Y., Ma, Y., Zhang, D., & Xu, Y. (2007). A cross-layer design of packet scheduling and resource allocation for multiuser MIMO–OFDM system. ICICS.
19. Tsai, C. F., Chang, C. J., Ren, F. C., & Yen, C. M. (2008). Adaptive radio resource allocation for downlink OFDMA/SDMA systems with multimedia traffic. IEEE Trans-actions on Wireless Communications, 7(5), 1734–1743. 20. Katoozian, M., Navaie, K., & Yanikomeroglu, H. (2009).
Utility-based adaptive radio resource allocation in OFDM wireless net-works with traffic prioritization. IEEE Transactions on Wireless Communications, 8(1), 66–71.
21. Yen, C. M., Chang, C. J., & Wang, L. C. (2010). A utility-based TMCR scheduling scheme for downlink MIMO/OFDMA sys-tems. IEEE Transactions on Vehicular Technology, 59(8), 4105–4115.
22. Jain, R. K., Chiu, D. M. W., & Hawe, W. R. (1984). A quantitative measure of fairness and discrimination for resource allocation and shared computer system. Technical Report DEC-TR-301, Digital Equipment Corporation. 23. Rahman, M., Yanikomeroglu, H., & Wong, W. (2009).
Inter-ference avoidance with dynamic inter-cell coordination for downlink LTE system. IEEE WCNC, April 2009.
24. 3GPP TR25.814. (2005). Physical layer aspects for evolved UTRA. 3rd Generation Partnership Project, Tech. Rep.
25. Rumney, M. (2009). LTE and the evolution to 4G wireless: Design and measurement challenges. New York: Wiley. 26. Hottinen, A., Tirkkonen, O., & Wichman, R. (2003).
Multi-antenna transceiver techniques for 3G and beyond. New York: Wiley.
27. Madan, R., Mehta, N. B., Molisch, A. F., & Zhang, J. (2008). Energy-efficient cooperative relaying over fading channels with simple relay selection. IEEE Transactions on Wireless Commu-nications, 7(8), 3013–3025.
28. Ngo, D. T., Tellambura, C., & Nguyen, H. H. (2010). Resource allocation for OFDMA-based cognitive radio multicast networks with primary user activity consider-ation. IEEE Transactions on Vehicular Technology, 59(4), 1668–1679.
29. 3GPP TR 25.996. (2010). Universal mobile telecommunications system (UMTS); spacial channel model for multiple input multi-ple output (MIMO) simulations. 3rd Generation Partnership Project, Tech. Rep., Jan. 2010.
30. WiMAX forum. (2007). Wimax system evaluation methodology. V.2.0. Tech. Rep., Dec. 2007.
31. 3GPP TR 25.892. (2004). Feasibility study for OFDM for UTRAN enhancement. 3rd Generation Partnership Project, Tech. Rep., 2004.
Wen-Ching Chungwas born in Taiwan, ROC, in June 1977. He received B.E. and Ph.D. degrees in electrical engineering from National Chiao Tung Univer-sity, Hsinchu, Taiwan, in 1999 and 2006, respectively. Since January 2008, he has joined the Department of Electrical and Computer Engineering of National Chiao Tung University in Taiwan as a Assistant Research Fellow. His research interests are in the areas of radio resources management and link adaption for wireless communication networks and nonlinear control.
Chung-Ju Changwas born in Taiwan, ROC, in August 1950. He received the B.E. and M.E. degrees in electronics engineer-ing from National Chiao Tung University, Hsinchu, Taiwan, in 1972 and 1976, respectively, and the Ph.D. degree in electri-cal engineering from National Taiwan University, Taiwan, in 1985. From 1976 to 1988, he was with Telecommunication Laboratories, Directorate Gen-eral of Telecommunications, Ministry of Communications, Taiwan, as a Design Engineer, Supervisor, Project Manager, and then Division Director. He also acted as a Science and Technical Advisor for the Minister of the Ministry of Communications from 1987 to 1989. In 1988, he joined the Faculty of the Department of Commu-nication Engineering, College of Electrical Engineering and Com-puter Science, National Chiao Tung University, as an Associate Professor. He has been a Professor since 1993 and a Chair Professor since 2009. He was Director of the Institute of Communication Engineering from August 1993 to July 1995, Chairman of Department of Communication Engineering from August 1999 to July 2001, and Dean of the Research and Development Office from August 2002 to July 2004. Also, he was an Advisor for the Ministry of Education to promote the education of communication science and technologies for
colleges and universities in Taiwan during 1995–1999. He is acting as a Committee Member of the Telecommunication Deliberate Body, Taiwan. Moreover, he once served as Editor for IEEE Communica-tions Magazine and Associate Editor for IEEE TransacCommunica-tions Vehicular Technology. His research interests include performance evaluation, radio resources management for wireless communication networks, and traffic control for broadband networks. Dr. Chang is members of the Chinese Institute of Engineers (CIE) and the Chinese Institute of Electrical Engineers (CIEE).
Hsin-Ying Teng was born in Taiwan, ROC, in May 1986. She received the B.E. degree in electrical engineering from National Taiwan University of Science and Technology, in 2008, and M.E. degree in elec-trical engineering from National Chiao Tung University, Hsin-chu, Taiwan, in 2010. Her research interests include radio resource management for wire-less communication networks.
