Throughput Improvement for OFDMA Femtocell ... - IEEE Xplore

2 downloads 0 Views 413KB Size Report
Abstract—In this paper, we propose a novel scheme to improve the throughput of femtocell networks. The proposed scheme consists of two steps: spectrum ...
Throughput Improvement for OFDMA Femtocell Networks Through Spectrum Allocation and Access Control Strategy Mingjie Feng, Da Chen, Zhiqiang Wang, Tao Jiang and Daiming Qu Wuhan National Laboratory for Optoelectronics Department of Electronics and Information Engineering Huazhong University of Science and Technology Wuhan, 430074, China Email: [email protected] Abstract—In this paper, we propose a novel scheme to improve the throughput of femtocell networks. The proposed scheme consists of two steps: spectrum allocation for femtocells and dealing with nearby macro-users with access control strategy. Since the positions of femtocells are relatively stable, we firstly deal with inter-femtocell interference and formulate it as an integer programming problem. Then we analyze the effects of different access strategies on system throughput for each subchannel and propose a novel access control strategy. Finally, the superiority of the proposed scheme is verified by simulation results.

Index Terms — femtocell, spectrum allocation, throughput, access control. I. I NTRODUCTION Femtocell is regarded as an effective solution for the indoor coverage problem and provide high data rates in future wireless networks [1]. However, the spectrum utilized by femtocell is provided by wireless operator, resulting in interference with existing cellular networks and interference among different femtocells. To deal with the interference problem, previous research focused on power control [2], [3] and spectrum partitioning [4], [5]. In [6], the author proposed a fractional frequency reuse (FFR) scheme that mitigate the interference, but the interference reduction is at the expense of decreased spectrum utilization. An adaptive fractional frequency reuse scheme had been proposed in [7], and the frequency reuse factor is determined according to specific interference level. In [8], [9], cognitive radio was applied to address crosstier interference, where the femtocell opportunistically utilize the available channel by recognizing interference signature. However we still face the problem of miss detection and false alarm, and the optimal channel selection strategy is not well studied yet. An alternative way to handle cross-tier interference is considering access control [10]. For the closed access scheme, only registered users are served by femtocell, while all the users are permitted to access to femtocell in open access scheme. Study to analyze capacity under different access schemes in the uplink has been conducted in [11]. In [12], the author defined the femtocell coverage area, and calculated the average throughput in each area with different access strategies in the downlink.

978-1-4577-1719-2/12/$26.00 ©2012 IEEE

In this paper, we firstly deal with inter-femtocell interference by formulating the femtocell spectrum allocation as an integer programming problem. In our assumption, closely deployed femtocells can not use the same channel, and we aim to maximize the resource block of all femtocells. After that, for both uplink and downlink spectrum of macrocell, we analyze the throughput of each interfering nearby macro-user and femto-user, then schedule the access scheme in order to maximize the throughput of the whole cellular system. By allocating proper spectrum resource for each femtocell and providing the optimal access strategy for each interfering macro-user, our scheme offer better throughput performance compared with several conventional schemes. The remainder of this paper is organized as follows. The system model considered in this paper is described in Section II. The proposed femtocell spectrum allocation is discussed in Section III. Then, the access strategy for each user is analyzed in Section IV. The performance of the proposed scheme is verified in Section V. Finally, conclusions are given in Section VI.

387

Outer region Inner region

Dth

Macrocell Base station

Rc

Femtocell Area Femtocell Base station

Fig. 1.

System model.

II. S YSTEM MODEL The system model of this paper is shown in Fig. 1. We consider an OFDMA cellular system, and the radius of macrocell and femtocell is Rc and Rf , respectively. The channel of a macrocell with bandwidth of Bc is divided into N subchannels, and the bandwidth of each sub-channel Bu is equal to the bandwidth required by a macro-user. For simplicity, we assume that the N sub-channels are sufficient for macrousers, thus each macro-user correspond to one sub-channel in our analysis. We assume both femtocells and macro-users are randomly distributed in the macrocell area, each macrouser is allocated with random and different sub-channel. The transmission power of macro-user Pu has 5 levels according to the distance with macrocell base station, and the macrocell base station use fixed transmission power of Pc . We consider a Rayleigh fading channel with unit average power, and experience path loss and wall penetration loss. The wall penetration loss and the path loss exponent is denoted as L and α, respectively. The noise power is σ 2 . The femtocell spectrum allocation method has several categories in previous study. One category is that femtocell located in a macrocell area utilize the spectrum of this macrocell, thus, interference management is required to coordinate between two tiers. For the other category, femtocell utilize the spectrum of other macrocell [13], [14], but interference problem still exist in cell-edge area. In [12], the macrocell area is divided into inner region and outer region with boundary Dth , which indicate that femtocells located in the two regions have different features. In this paper, we allocate the inner region with the spectrum of another macrocell, while allocate the outer region with the spectrum of the corresponding macrocell. For the inner region, due to longer distance with other macrocell and the low transmission power femtocell base station, interference between femtocell and other macrocell is neglectable in our study. For the outer region, the distance between femtocell base station and macrocell base station is relatively further, therefore we can mainly focus on dealing with the interference between cell-edge macro-user and interference between femtocell base station and macrocell base station is mitigated. III. F EMTOCELL SPECTRUM ALLOCATION In this section, we consider maximize the total spectrum bandwidth of all femtocells while mitigate inter-femtocell interference and guarantee a basic resource allocation for each femtocell. Define the interference list as a M × M binary matrix with element Fij , where Fij = 1 indicates that femtocell i and j will have inter-femtocell interference when they utilize the same sub-channel, and Fij = 0 indicates that femtocell i and j will not bring interference to each other even they utilize the same sub-channel. Therefore, as shown in Fig. 2, the interference list can be obtained by  1, Dij < Rf + Ri j Fi = . (1) 0, else where Dij is the distance between two femtocell base stations, and Ri is the interfering radius of femtocell. We only consider

Femtocell area

Rf

Dij

Ri

Rf Femtocell area

Ri Fig. 2.

Inter-femtocell interference.

path loss in this section to obtain Ri , and Ri can be obtained using the path loss model between two femtocells. Denote W as the spectrum resource of all femtocells, δnm is an indicator and is defined as 

sub − channel n is assigned to femtocell m else (2) In order to ensure the fairness among femtocells, and avoid unnecessary resource allocation, the spectrum resource allocated to each femtocell should have an upper bound and denoted as Am . Therefore, the spectrum allocation of femtocell is formulated as n δm =

1, 0,

W =

M N P P

n=1 m=1

n δm

max W δ  j  δin δjn Fi = 0 N P n s.t.  δm ≤ Am

(3)

n=1

However, obtaining the optimal solution of this problem by exhaustive searching brings huge complexity since the number of all possible combinations is exponentially increasing with the number of femtocells multiplexing the number of subchannels (M × N ). Therefore, we define the independent femtocells set, and consider a resource block of 10 subchannels for analysis. In an independent femtocell set, a femtocell only brings interference to femtocells in the same set, while does not bring interference to femtocells in other independent sets. With this approach, we can obtain the optimal solution of spectrum allocation in each independent set with lower complexity, hence the optimal solution is the combination of the solutions of these independent sets. By

388

considering a resource block of 10 sub-channels, we assign the N/10 resource blocks for femtocells and the computational complexity is obviously reduced. In our model, femtocells in different regions utilize different sets of sub-channels, thus we should conduct spectrum allocation for femtocells located in inner region and out region separately. Note that the value of Am also affects the final result, and it is related with the density of femtocells. Obviously, femtocell that is interfered by more other femtocells will be allocated with less spectrum resource to maximize the total spectrum resource while avoiding inter-femtocell interference. Therefore, when Am is large, the fairness among femtocells will not be satisfactory, while when Am is small, the number of total spectrum resource will significantly reduced. In this paper, we set Am to be a constant that consider both fairness and the total benefits, future work may search the optimal value of Am with respect to femtocell density. IV. T HROUGHPUT ANALYSIS BASED ACCESS CONTROL STRATEGY

Since femtocells located in the inner region utilize the spectrum of another macrocell and thus the cross-tier interference is neglectable, this section focus on addressing the cross-tier interference in outer region. We stand on the point of wireless operator, and aim to maximize the throughput of the cellular system. Previous research [10], [12] had shown the fact that there is a tradeoff in the throughput performance between femtocell and macrocell, and femto-users tend to adopt closed access to maintain high data rates while macro-users prefer open access to improve their QoS especially in cell-edge area. In this paper, we analyze the throughput relationship between interfering macro-users and femto-users in detail, then propose a novel access control strategy and the procedure to realize it.

  Pf R−α g0 Tu =Bu log2 1 + P LR−α g0 + σ 2  u  −α P D g u 0  (4) P + Bu log2 1 + Pf LD−α g0 + σ 2 cochannelFBS

Open access:  Pu LR−α g0  P Pf LD−α g0 + σ 2



Tu = Bu log2 1 +

Where R and D denote the distance between user and femtocell base station, macrocell base station, respectively. g0 is the exponentially distributed channel power with unit mean. For sub-channels used by the downlink of the macrocell, macro-user is interfered by co-channel femtocells, and interfering femtocells also have the choice of open or closed access. In closed access, femtocells should reduce transmission power on the sub-channel to guarantee the QoS of macro-user, while in open access, the sub-channel is used to serve macro-user and not available for femtocell. Therefore, the sum throughput on this sub-channel for both access strategies can be represented as Closed access:   Pf 0 R−α g0 Tu =Bu log2 1 + P LD−α g0 + σ 2  c  −α P D g c 0  (6) P + Bu log2 1 + Pf 0 LR−α g0 + σ 2 cochannelFBS

Open access: 

A. Throughput analysis on each sub-channel In our scheme, the interference management is mainly accomplished by femtocell base station. Since frequent frequency switching and handover will deplete the battery power of a macro-user equipment (MUE), we assume that macro-user will not conduct channel switching for both access strategies. For analytical tractability, we neglect the difference between uplink and downlink of femtocell users and use the downlink for our study. For sub-channels used by the uplink of the macrocell, if a sub-channel is occupied by a macro-user and nearby femtocells at the same time, femtocells are interfered by this macro-user, and have the choice of open or closed access. Closed access means that femtocell continue occupying the sub-channel while tolerate the interference, while open access means the sub-channel is used to serve macro-user and not available for femtocell. Therefore, the sum throughput on this sub-channel for both access strategies can be represented as Closed access:

(5)

cochannelFBS

Tu = Bu log2 1 +

 Pf LR−α g0  P Pf 0 LR−α g0 + σ 2

(7)

cochannelFBS

Where Pf 0 is the adjusted transmission power of femtocell base station. B. Access control strategy In our model, we assume the initial system state is closed access, and the access results will be updated after a period of time. Both femtocell base station (FBS) and MUE can identify the received signal, and macrocell base station (MBS) schedule the whole system through control channels. 1) uplink: For the uplink sub-channels of macrocell, when MUE and femtocells are utilizing the same sub-channel, the access control strategy has the following steps: • If half of the spectrum of a FBS is used for MUE transmission, this FBS is unavailable for open access. • On each sub-channel, FBS detects the interference signal of nearby MUE, then obtain the throughput loss if adopting open access. The throughput of macro-user

389

if adopting open access can be obtained based on the received signal strength of MUE. The calculation results and the availability information are sent to MBS for scheduling. • The throughput of MUE when adopting closed access is easily obtained at the MBS, thus the throughput gain from closed access to open access for MUE is also obtained. Then, MBS compare the maximal throughput gain of MUE if access to an available FBS with the sum throughput loss of all interfering FBSs. If the throughput gain of the MUE is greater than the throughput loss of all FBSs, the MUE is scheduled to access to the FBS. If the throughput gain of MUE is less than the throughput loss of all FBSs, the MUE continue served by MBS in the uplink. • If the MUE is determined to access to a FBS, reducing the transmission power of MUE may further improve the system throughput since lower power will generate less interference to other FBSs while the reduced power is sufficient for open access transmission. Thus, the MBS replace the transmission power of MUE to lower levels and repeat the calculation. Hence, the optimal transmission power of MUE when access to a FBS can be found. • MBS sends to access result and transmission power information to MUE. 2) downlink: For the downlink sub-channels of macrocell, when MUE and femtocells are utilizing the same sub-channel, the access control strategy has the following steps: • If half of the spectrum of a FBS is used for MUE transmission, this FBS is unavailable for open access. • MUE detects and identifies the interference signals of nearby FBSs. Then MUE reports the received signal strength of MBS and each interfering FBSs to MBS. • MBS informs each of the interfering FBS, and each FBS reduce the transmission power on this sub-channel to Pf 0 . • Each interfering FBS calculate the throughput loss on this sub-channel if adopting open access and send the results to MBS. The calculation results and the availability information are sent to MBS for scheduling. • MBS calculates the throughput of MUE with both access schemes based on power information reported from MUE. Then, MBS compare the maximal throughput gain of MUE when served by an available FBS with the sum throughput loss of all interfering FBSs. If the throughput gain of the MUE is greater than the throughput loss of all FBSs, the MUE is scheduled to access to the FBS. Otherwise, the MUE will continue served by MBS in the downlink. • If the access scheme is open access, MBS will inform the corresponding FBS to serve the macro-user. V. S IMULATION RESULTS The performance of the proposed scheme is verified by computer simulation with system parameters shown in Table I. Interference power threshold Pi is set to determine whether

TABLE I S IMULATION PARAMETERS Pc

43dBm

Pf

20dBm

Pf 0

15dBm

Pu

10dBm,15dBm,20dBm,25dBm,30dBm

Rc

500m

Rf

30m

Dth

250m

Bc

20MHz

Bu

200KHz

α

3.5

L

0.8

σ2

-105dBm

Am

70

Pi

-60dBm

two femtocells can use the same sub-channel in Section III, and whether to handle access control in Section IV. To evaluate the spectrum allocation of the proposed scheme, the proposed scheme is compared with conventional schemes. For the validation of the access control strategy, we compare the proposed scheme with closed access scheme and open access scheme. We assume that in open access scheme, any interfering macro-users will be served by femtocells. Fig. 3 depicts the total throughput of femto-users and macro-users with different spectrum allocation schemes. In the universal frequency reuse scheme, all femtocells located in the macrocell utilize the spectrum of this macrocell, and each femtocell utilize the whole spectrum of this macrocell. For the improved universal frequency reuse scheme, femtocells located in the inner region utilize the spectrum of another macrocell , while femtocells located in the outer region utilize the spectrum of this macrocell, and each femtocell utilize the whole spectrum of the corresponding macrocell. We can observe that when the number of femtocell is small, universal frequency reuse scheme and improved universal frequency scheme can obtain higher throughput than the proposed scheme, but when the number of femtocell gets larger, the proposed scheme offer better performance. This is because there is enough space for spatial reuse of spectrum when there are few femtocells, and each femtocell can use the whole spectrum with slight interference to each other. As the more femtocells are deployed, the spectrum resource upper bound Am is necessary to reduce interference. We can also infer that the optimal value of Am is related to the number of femtocells. As the number of femtocells increases, Am should decrease, which means in dense deployed situation, the constraint of resource allocation for each femtocell should be tight to reduce interference and improve the whole system performance. The proposed scheme could offer better performance since the maximal number of sub-channels are allocated to each femtocell while avoiding inter-femtocell interference. Fig. 4 shows the total throughput performance compared

390

provide throughput improvement compared with conventional schemes.

9 Improved Universal Reuse Universal Reuse Proposed

Total System Throughput (Gbps)

8

ACKNOWLEDGMENT

7

This work was supported by the Project-sponsored by SRF for ROCS, SEM, the National & Major Project with Grant 2012ZX03003004, the National Science Foundation of China with Grant 61172052 and Grand 60872008, the Program for New Century Excellent Talents in University of China under Grant NCET-08-0217, the Research Fund for the Doctoral Program of Higher Education of the Ministry of Education of China under Grant 200804871142, the Science Found for Distinguished Young Scholars of Hubei in China with Grant 2010CDA083, and Major Program of Ministry of Science and Technology of China (No. 2010ZX03003-002-03).

6 5 4 3 2 1 0

Fig. 3.

0

20

40 60 Number of Femtocell

80

100

Total system throughput for different spectrum allocation schemes.

R EFERENCES

with fully open access and closed access scheme. In order to better demonstrate the effectiveness of the proposed scheme, we make the comparison for the sum throughput of interfering users, and use the spectrum allocation strategy of universal frequency reuse. As expected, the proposed access control strategy could achieve higher throughput than fully open and closed scheme while guarantee the benefits of both macrousers and femto-users. With user selection and power adjustment in the proposed strategy, each macro-user is given with the best access strategy to maximize the total throughput. 35 Sum Throughput of Interfering Users (Mbps)

Proposed Closed Access

30

Open Access 25

20

15

10

5

0

0

20

40 60 Number of Femtocell

80

100

Fig. 4. Comparison of the throughput of interfering users with different access strategies.

VI. C ONCLUSION In this paper, we considered the design of a spectrum allocation and access control strategy for femtocell networks. The spectrum allocation is based on femtocell location, and aim to maximize the resource for femtocells while mitigate inter-femtocell interference. We also considered finding the optimal access strategy for macro-users to improve the total system throughput while mitigate the cross-layer interference. The simulation results demonstrate that the proposed scheme

[1] V. Chandrasekhar, J. G. Andrews, and A. Gatherer, “Femtocell networks: a survey,” IEEE Commun. Mag., vol. 46, no. 9, pp. 59-67, Sep. 2008. [2] H.-S. Jo, C. Mun, J. Moon, and J.-G. Yook, “Interference mitigation using uplink power control for two-tier femtocell networks,” IEEE Trans. Wireless Commun., vol. 8, no. 10, pp. 4906-4910, Oct. 2009. [3] H.-S. Jo, C. Mun, J. Moon, and J.-G. Yook, “Self-optimized coverage coordination in femtocell networks,” IEEE Trans. Wireless Commun., vol. 9, no. 10, pp. 2977-2982, Oct. 2010. [4] V. Chandrasekhar, J. G. Andrews, “Spectrum allocation in tiered cellular networks,” IEEE Trans. Commun., vol. 57, no. 10, pp. 3059- 3068, Oct. 2009. [5] I. Guvenc, M.-R. Jeong, F. Watanabe, and H. Inamura, “A hybrid frequency assignment for femtocells and coverage area analysis for cochannel operation,” IEEE Comm. Lett., vol. 12, no. 12, pp. 880-882, Dec. 2008. [6] R. Giuliano, C. Monti, and P. Loreti, “WiMAX fractional freuqency reuse for rural environments,” IEEE Wireless Commun., vol. 15, no. 3, pp. 6065, June. 2008. [7] Heui-Chang Lee, Dong-Chan Oh, and Yong-Hwan Lee, “Mitigation of Inter-Femtocell Interference with Adaptive Fractional Frequency Reuse,” in Proc. IEEE International Conference on Communications, pp. 1-5, May. 2010. [8] Chiao Lee, Jane-Hwa Huang, Li-Chun Wang. “Distributed Channel Selection Principles for Femtocells with Two-Tier Interference,” in Proc. IEEE Vehicular Technology Conference, pp. 1-5, May. 2010. [9] Yang-Yang Li, Macuha. M., Sousa. E.S., Sato. T.. Nanri. M., “Cognitive Interference Management in 3G femtocells,” in Proc. IEEE Personal, Indoor and Mobile Radio Communications, pp. 1119-1122, Sep. 2009. [10] Guillaume de la Roche, Alvaro Valcarce, David Lpez-Prez, and Jie Zhang, “Access Control Mechanisms for Femtocells,” IEEE Commun. Mag., vol. 48, no. 1, pp. 33-39, Jan. 2010. [11] P. Xia, V. Chandrasekhar, and J. G. Andrews, “Open vs. closed access femtocells in the uplink,” IEEE Trans. Wireless Commun., vol. 9, no. 12, pp. 3798-3809, Dec. 2010. [12] Han-Shin Jo, Ping Xia, and Jeffrey G. Andrews, “Downlink Femtocell Networks: Open or Closed?,”in Proc. IEEE International Conference on Communications, pp. 1-5, June. 2011. [13] Ju Yong Lee, Sueng Jae Bae, Young Min Kwon, Min Young Chung, “Interference Analysis for Femtocell Deployment in OFDMA Systems Based on Fractional Frequency Reuse,” IEEE Comm. Lett., vol. 15, no. 4, pp. 425 - 427, Mar. 2011. [14] Haddad. Y, Porrat. D, “A Two-Tier Frequency Reuse Scheme,” in Proc. IEEE Personal, Indoor and Mobile Radio Communications, pp. 203 207, Sep. 2010.

391