2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

Downlink Resource Allocation with Cooperative Transmission in EPON-WiMAX Integrated Network Shang-Hsiu Tien

Hung-Yi Teng

Ren-Hung Hwang

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Abstract—The rapid growth of the demand of multimedia applications in recent years has posted a much higher bandwidth demand than before. As a result, the technology of broadband access is becoming more and more important. Integrating Ethernet Passive Optical Network (EPON) and Worldwide Interoperability for Microwave Access (WiMAX) has been considered as a very promising solution for broadband access. Although EPON-WiMAX has received many attentions in research, yet most of the previous researches focus on the uplink packet scheduling and bandwidth allocation. Thus, in our research, we focus on the downlink scheduling and resource allocation problem. In particular, we focus on utilizing the cooperative transmission technique. Based on the advantage of EPONWiMAX architecture, the resource allocation problem utilizing cooperative transmission can be done at the optical line terminal (OLT) in a centralized manner. We first formulate the problem as a linear programming (LP) problem. Due to the high complexity of LP, we propose a heuristic algorithm to solve the problem. Our simulation results show that the proposed algorithm is able to increase the system throughput while guaranteeing the QoS of different classes of traffic. Keywords-EPON-WiMAX integrated network; resource allocation; cooperative transmission

I.

downlink;

INTRODUCTION

In recent year, a variety of multimedia applications such as high-definition television and video-on-demand have posted a high bandwidth demand. As a result, the broadband access technology has become more and more important. Integrating Ethernet Passive Optical Network (EPON) [1] and Worldwide Interoperability for Microwave Access (WiMAX) [2][3] has been considered as a very promising solution for achieving broadband Fixed/Mobile convergence [4][5]. Fundamentally, EPON and WiMAX are well-matched with each other in the aspect of service objective, multiple access mechanism, and capacity hierarchy. In general, the architecture of the EPONWiMAX integrated network is a tree topology, consisting one EPON Optical Line Terminal (OLT), one 1:N passive optical splitter, and multiple integrated units, referred to as ONU-BS, which include an EPON Optical Network Unit (ONU) and a WiMAX Base Station (BS). Advantages of the integrated EPON-WiMAX network include providing broadband Internet access, supporting mobile users, and reducing the network design and maintenance costs.

For seamless integration of EPON and WiMAX, uplink packet scheduling and resource allocation have received much more attention. Yang et al. [8] presented a converged network architecture based on the concept of virtual ONU-BS (VOB) and proposed a QoS-aware dynamic bandwidth allocation (DBA) scheme. Jung et al. [9] investigated three possible integrated architectures and proposed a centralized scheduling mechanism to enhance end-to-end delay and provide better QoS provisioning for the lower priority traffic. Dias Piquet and Saldanha Fonseca [10] assessed the performance of a standardcompliant WiMAX uplink scheduler and showed that the scheduler can provide QoS support to the SSs. Relatively less attention has been focused on downlink packet scheduling and resource allocation on the EPON-WiMAX integrated network. Our previous work [11] presented a two-stage design of downlink packet scheduling and resources allocation to achieve high system throughput while QoS of each traffic class could be guaranteed. But, the inter-cell collaboration was not considered. Emphasizing on inter-cell cooperative transmission, Gong et al. [12] proposed three schemes to optimize ONU-BS user association and resources allocation (BUA-RA) in terms of minimizing the number of rejected connection requests. However, QoS guarantee for different traffic classes was not considered, which is essential to WiMAX. In this paper, we focus on the downlink scheduling and resource allocation problem. In particular, we focus on how to properly utilize the cooperative transmission technique. Based on the advantage of EPON-WiMAX architecture, we formulate the resource allocation problem with inter-cell collaboration as a linear programming (LP) problem. Due to the high computation complexity of LP, we further propose a heuristic algorithm to solve the problem. Our simulation results show that

To achieve throughput enhancement, advanced inter-cell collaboration is possible by employing cooperation transmission techniques such as Space-Time Coding [6], Spatial Multiplexing

978-1-4799-2358-8/14/$31.00 ©2014 IEEE

[7], and Beamforming among distributed ONU-BSs. In this paper, we assume Space-Time Coding is employed to perform inter-cell collaboration. In the EPON-WiMAX integrated network, the OLT plays an important role to support inter-cell collaboration in a centralized manner. In this manner, several ONU-BSs can encode the data with orthogonal space-time codes and transmit cooperatively to a subscriber station (SS). Then, the SS can aggregate the same copy of the data and decodes them in high chance. Although inter-cell collaboration is promising to enhance throughput of SSs located at cell edge. Unfortunately, it may consume more system resources as resources need to be allocated at cooperating BSs. Therefore, how to decide whether use inter-cell collaboration for a cell edge SS is crucial.

750

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

to Interference plus Noise Ratio (SINR), and the selected MCS choice.

the proposed algorithm is able to increase the system throughput while guaranteeing the QoS requirements of each traffic class. The rest of the paper is organized as follows. Section II gives an overview of our system model. In Section III, we propose our approaches to effectively allocate resources so that the system throughput can be improved. Simulation results are presented in Section IV. Conclusions are finally drawn in Section V. II.

1

(4) (5) log 1 (6) In equations (4), (5) and (6), SLER denotes the slot error rate; is the total number of time slots in a MAC PDU which equals / ; is the mean capacity, the constants A and B are chosen to fit the selected MCS curve for the given channel model. 10

SYSTEM MODEL

A. Network Environment Figure 1 shows the architecture of an EPON-WiMAX integrated network which is composed of an EPON as backhaul and multiple WiMAX networks as front-end wireless access networks. In the EPON-WiMAX integrated network, the major components include an optical line terminal (OLT), a passive splitter, and multiple ONU-BSs. As shown in Figure 1, the OLT is connected to multiple ONU-BSs via a passive splitter. It has powerful computing capacity to manage packet traffic efficiently. As proposed in [5], the ONU-BS could be a hybrid device integrating an EPON ONU and a WiMAX BS. It functions as a bridge device to achieve seamless integration between EPON and WiMAX. The resource management functions are located at the OLT so that the resource can be allocated to all SSs in a centralized manner. We assumed that each ONU-BS can access full context of all SSs within its coverage, including rate requirement, distance and signal strength, etc. And, it periodically sends the information of its SSs to the OLT using MultiPoint Control Protocol (MPCP) [15]. Therefore, the OLT can obtain the required information of all SSs in the EPON-WiMAX integrated network. B. Problem Formulation In our model, without cooperative transmission, the data rate of a subscriber station (SS) n is expressed as follows [11]:

∙

However, by employing cooperative transmission techniques such as Space-Time Coding, a SS is possible to be cooperatively served by multiple ONU-BSs. Therefore, SINR of SS n can be estimated as follows. (7) ∑ ∑

(8) (9)

1

where

1 (10) 0 For simplicity, we use Log-Distance path loss model to estimate the received power strength of SS n from ONU-BS m. ∙

(11)

In this study, we aim to maximize the sum of individual received bandwidth (throughput) of each SS assuming that there are M ONU-BSs and N SSs in the network. Our problem is formulated as how does the OLT properly select the serving ONU-BS(s) and the MCS for each SS such that the aggregated network throughput could be maximized. Our objective function is written as follows: ∑ (12) Maximize which is subject to the following constraints: 1) Total time slot constraint ∑ 1, ∀ 1… (13) 2) Data rate requirement of each user ∙ ∙ ,∀ 1 … (14) ∙ 3) BS-User association constraint ∑ 3, ∀ 1… (15) 4) MCS selection constraint , (16) Equation (13) indicates that for each ONU-BS (m), the number of the time slots ( ) that are allocated to the SS does not exceed the total number of time slots (B). Equation (14) indicates that the data rate requirement of each SS must be satisfied by the selected solution. Equation (15) represents that a SS is only able to associate with three ONU-BSs at most [14]. Equation (16) indicates that through an appropriate selection of MCS, the packet loss rate ( ) of SS n should be less than the threshold specified in the WiMAX standard. Table I shows important notations used in this study.

∙ ∙ (1) where represents the number of bits per slot to the MCS denotes the normalized time slots adopted by SS n, allocated to SS n, B is the total wireless bandwidth (in number of time slots). Given a specific MCS choice m, the number of bits which can be transmitted in a time slot is ∗ ∗ (2) denotes the number of carrier symbols per slot, denotes the number of bits per carrier symbol to MCS denotes the channel coding rate to MCS m in m, and physical layer. A time slot is the minimal data allocation unit which is defined to be a collection of subchannels along contiguous OFDM symbols. Let Dn be the amount of data sent from the OLT to SS n in a predefined period. Based on the MCS choice adopted by SS n, the time slots allocated to SS n is where

∙

1

(3)

Each payload of upper layer is fragmented into fixed-size (LByte) SDUs. Each MAC_PDU contains 10-Byte MAC header, L-Byte SDU and 4-Byte CRC. Let be the MAC_PDU size. Wang et al. [13] presented equations (4)–(6) for calculating the packet loss rate given the signal strength, represented by Signal

TABLE I. LIST OF NOTATIONS

751

Symbol

Meaning

M

The total number of ONU-BS in the system

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

N

The total number of SS in the system

B

The total wireless bandwidth (time slots)

Tsmn Amn

The associating indicator of ONU-BS m and SS n

Rn

The actual bandwidth (bps) allocated to SS n

x

The SINR value of SS n

mn

The modulation and coding scheme adopted by SS n

Pn

The packet loss rate of SS n

TPER

The threshold of packet loss rate in WiMAX

III.

PROPOSED SCHEMES

0, ∀j 0 … J, ∀n 4) Variables Ts ∈ 0,1, … , x ∈ 0, 1

…

1

1

1

Step

(17) 1…M

(21)

1 … N

(22) (23) (24)

th transmission mode is used for SS n otherwise

3:

Calculate

_

,

.

If

_ 1 , the serving set will be selected. Otherwise, the combination of ONU-BSs that produces the largest gain will be selected. Step 4: Select the SS n that has the minimal resource consumption to perform resource allocation. If there are multiple SSs have the same resource consumption, the SSs chose the serving set will be selected preferentially. Step 5: Check if all the selected ONU-BS(s) have enough resource. If yes, the selected ONU-BS(s) will serve the SS n and the OLT update the time slot usage of the selected ONU-BS(s) by equations (27) and (28). Otherwise, the SS n will not be served at this frame. Go back to Step 4 until the system resource run out or no SS needs to be allocated. (27) (28)

Then the linear programming problem can be written as follows. Objective: Maximize ∑ R 1) Total time slot constraint ∑ ∑ Ts B , ∀m b 2) Data rate requirement of each user

1 0

(20)

(26)

0 0 0

…

0

0 0 0 …

0

0 0 0

…

… 0

…

… J

… … … …

(19)

Step 1: For each SS, find three closest ONU-BSs denoted as , , and . , , and are sorted according to SINR. , , , Step 2: Set the serving sets and , , . Calculate the consumed resources (time slots) of the serving sets , , and denoted as , , and using equation (26).

TABLE II. TABLE OF TRANSMISSION MODE 0 0 1

1…N

A. The MaxGain heuristic Algorithm be the request rate of SS n. Let be the total time Let slot of ONU-BS m, be the available time slot of ONU-BS m, be the allocated time slot of ONU-BS m ( and ). The OLT conducts the following steps to allocate resources.

A. Linear Programming Since equations (13) and (15) are non-linear, we apply the following approach, similar to [12], to linearize the equations. The possible transmission modes are defined as, J, which indicate a combination of ONU-BSs is selected to serve a SS. Table II shows the binary representation of the jth transmission is used to indicate whether the mode. The binary variable mth ONU-BS is served in the jth transmission mode. For example, the transmission mode j=0 represents none of ONUBSs is selected to serve a SS and j=J represents the SS is served by ONU-BSs M-2, M-1, and M. Due to equation (14), the maximal number of transmission modes for each SS is 1 M .

0 1 0

0, ∀n

(25) Equation (18) ensures that the total allocated time slots to each SS do not exceed the total time slots of a single WiMAX frame. Equations (19) and (20) indicate that the data rate requirement of each SS must be guaranteed. Equations (21) and (22) restrict that a SS can only use a single transmission mode. Equations (23) and (24) define the valid ranges of the decision variables.

Cooperative transmission is a promising technique to enhance throughput of SSs located at cell edge. Unfortunately, it may consume more resources if not properly designed. Therefore, for a cell edge SS, how to decide whether to use cooperative transmission technique or not is crucial. To solve the resource allocation problem, we present two approaches in this paper. The first approach is to formulate this problem into a linear programming problem. Thus, the optimal solution can be obtained by using LP solvers such as LINDO or CPLEX. Due to the high computation complexity of LP, we further propose a heuristic algorithm to obtain a near optimal solution.

j 0 1 2

Ts

The request rate (bps) of SS n SINRn

R R , ∀n 1 … N 3) BS-User association constraint ∑ x 1 , ∀n 1 … N

The normalized resource (%) allocated to SS n from ONU-BS m The resource (time slots) allocated to SS n from ONU-BS m

Tpmn

∑

R

(18)

752

IV.

PERFORMANCE EVALUATION

In this section, we investigate the performance of our approaches via simulations. The simulation program is written by C language and the LP problem is solved by LINDO v.6.1. The simulations were conducted on a computer with Intel Core i5-3230M CPU and 4GB memory. We adopt system throughput

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

and computation time as the performance metrics to evaluate transmission and computation efficiency of our approaches.

B. Simulation results Figure 2 represents the overall system throughput of three approaches. When the number of SS is more than 45, the LP approach is not able to yield an optimal solution since the system resource is not sufficient to satisfy QoS requirements of all SSs. The overall system throughput of the Max_Gain approach is very close to that of the LP approach. When the number of SS is 45, the throughput of the LP approach is only 5.24% higher than that of the Max_Gain approach. However, the Max_Gain approach significantly outperforms than the NCT approach in all cases. When the number of SS is 25, the throughput of the Max_Gain approach is 22.53% higher than that of the NCT approach. As the number of SSs grows, the difference between the Max_Gain approach and the NCT approach slightly decreased. It is because the system resources become scarce and therefore the throughput of cell edge SSs cannot be enhanced via cooperative transmission.

A. Simulation parameters In the simulation, the network topology consists of one OLT and seven ONU-BSs as shown in Fig. 1. The distance between ONU-BSs is 1400 meters. Simulation parameters and QoS requirements of the four traffic classes are summarized in Table III and IV, respectively. Each SS has one connection in each traffic classes, namely, four connections in total. In addition, all ONU-BSs have adaptive modulation and coding schemes capacity as shown in Table V. We simulate a general scenario where all SSs are randomly placed. The number of SS is varied from 25 to 70. We compare our approach (Max_Gain) with two approaches under different number of SSs; in the first approach, the ONU-BS does not employ cooperative transmission technique (NCT); the other approach is the LP approach. The solution yielded by the LP approach is regarded as a benchmark. We assume that the strict priority scheduling is adopted by ONU-BSs when scheduling four classes of traffic. For each setting, we run simulation 30 times and calculate the average results as well as the 95% confidence interval.

Next, we further examine the throughput for individual traffic class. Due to shortage to space, we only discuss the performance of UGS and BE traffics here. Figure 3 represents the throughput of UGS traffic. Again, the LP approach cannot provide the optimal solution when the number of SS is more than 45. When the number of SS is less than 50, the simulation results show that the Max_Gain approach is able to provide a very near optimal solution in UGS traffic. Also, the Max_Gain approach significantly outperforms than the NCT approach in UGS traffic regardless of the number of SS. We also observe the similar trend in rtPS and nrtPS traffics. This fact indicates that the Max_Gain approach can effectively allocate resource to each SS so that the QoS requirements of UGS traffic is satisfied. Figure 4 shows the throughput of BE traffic. In Fig.4, when the number of SS is less than 55, the Max_Gain approach can provide a better throughput of BE traffic than the NCT approach. But, when the number of SS is more than 55, the throughput of BE traffic of the Max_Gain approach decreases slightly. The reason is that the Max_Gain approach can serve the number of SS more than the NCT approach due to inter-cell collaboration. And, most resources are preferentially allocated to the UGS, rtPS and nrtPS traffics. As a result, BE traffic can only obtain less resources when the system resource become scarce. However, this fact will not be a problem since BE traffic does not have any QoS requirements.

TABLE III. SYSTEM PARAMETERS USED IN SIMULATIONS Parameter

Value

WiMAX system bandwidth

10 MHz

WiMAX frame duration

5ms

FFT Size

1024

DL:UL

2:1

Number of OLT

1

Number of ONU-BS

7

Number of SS

25 - 70

Number of connection

100 - 280 100 1 * 10 4

TABLE IV. SPECIFICATION OF TRAFFIC CLASSES Class

Traffic rate

Traffic model

Inter-arrival time

QoS parameters

UGS

64kbps

CBR

20ms

Delay: 20ms

rtPS

176kbps

VBR (poisson)

20ms

nrtPS

232kbps

20ms

BE

360kbps

VBR (poisson) CBR

Delay: 60ms Min. reserve rate: traffic rate Min. reserve rate: traffic rate

Last, we examine computation efficiency of our approaches. The computation time of the LP and Max_Gain approaches is shown in Table VI. It is obvious to see that the LP approach takes very much time to obtain the optimal solution especially when the number of SS is 45. And, it cannot yield an optimal solution when the number of SS is more than 45. However, no matter how many SSs have, the Max_Gain approach only requires no more than 1 millisecond to obtain a near optimal solution. This result indicates that the Max_Gain approach can easily handle fast and frequent state fluctuation due to user mobility in EPON-WiMAX integrated networks.

20ms

TABLE V. MCSS USED IN SIMULATIONS Modulation BPSK QPSK QPSK 16-QAM 16-QAM 64-QAM 64-QAM

Coding rate 1/2 1/2 3/4 1/2 3/4 2/3 3/4

SINR threshold 3.0 6.0 8.5 11.5 15.0 19.0 21.0

Byte/slot 3 6 9 12 18 24 27

TABLE VI. COMPUTATION TIME (MILLISECOND)

753

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

V.

[5]

CONCLUSION

EPON-WiMAX is a promising solution for multimedia applications with mobility support. In this study, we focused on the downlink resource allocation problem with inter-cell collaboration in the EPON-WiMAX integrated network. We proposed two approaches in this paper. The first approach is to formulate this problem into a linear programming problem. Due to the high computation complexity of LP, we further propose a heuristic algorithm to obtain a near optimal solution very efficiently. Our simulation results show that the proposed algorithm can increase the system throughput by properly employing cooperative transmission while guaranteeing the QoS of each traffic class.

[6]

[7]

[8]

[9]

ACKNOWLEDGMENT The authors would like to thank the National Science Council of the Republic of China, Taiwan for financially supporting this research under Contract No. NSC 100－2221－ E－194－012－MY3.

[10]

[11]

REFERENCES [1]

G. Kramer and G. Pesavento, “Ethernet passive optical network (EPON): building a next-generation optical access network,” IEEE Communications Magazine, vol. 40, no. 2, pp.66–73, Feb 2002. IEEE 802.16, “IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access Systems,” 2004 IEEE 802.16e, “IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems,” Feb. 2005. C. Ranaweera, E. Wong, C. Lim, and A. Nirmalathas, “Next Generation Optical-Wireless Converged Network Architectures,” IEEE Network Magazine, pp. 22-27, Mar./Apr. 2012.

[2]

[3]

[4]

[12]

[13]

[14]

[15]

G. Shen, R. S. Tucker, and C.-J. Chae, “Fixed Mobile Convergence Architectures for Broadband Access: Integration of EPON and WiMAX,” IEEE Communication Magazine, vol.45, no.8, pp.44–50, Aug. 2007. V. Tarokh, N. Seshadri, and A. R. Calderbank, ”Space-Time Codes for High Data Rate Wireless Communication: Performance Criterion and Code Construction,” IEEE Transaction on Information Theory, vol. 44, no. 2, pp. 744–764, 1998. Qinghua Li, Xintian Lin, Jianzhong Zhang and Wonil Roh, “Advancement of MIMO technology in WiMAX: from IEEE 802.16d/e/j to 802.16m,” IEEE Communications Magazine, vol. 47 , no. 6, pp. 100– 107, 2009. K. Yang, S. Ou, G. Ken, and H.-H. Chen, “Convergence of Ethernet PON and IEEE 802.16 Broadband Access Networks and its QoS-Aware Dynamic Bandwidth Allocation Scheme,” IEEE Journal on Selected Areas in Communications, vol. 27, no. 2, pp. 101–116, 2009. B. Jung, J.Y. Choi, Y.-T. Han, M.-G. Kim, M. Kang, “Centralized Scheduling Mechanism for Enhanced End-to-End Delay and QoS Support in Integrated Architecture of EPON and WiMAX,” Journal of Lightwave Technology, vol. 28, no. 16, pp. 2277–2288, 2010. M. Dias Piquet, and N.L. Saldanha Fonseca, “Performance Evaluation of a Scheduler for the ONU-BS of Integrated EPON-WiMAX Networks,” IEEE Latin America Transactions, vol. 10, no. 3, pp. 1838–1843, 2012. Hung-Yi Teng, Chung-Chih Kuo and Ren-Hung Hwang, “Downlink packet scheduling and resource allocation in EPON-WiMAX hybrid access networks,” EURASIP Journal on Wireless Communications and Networking, 2012:333. M. Gong, B. Lin, P.-H. Ho, P. Hung, “Adaptive BU association and resource allocation in integrated PON-WiMAX networks,” Wireless Communications and Mobile Computing, vol. 12, no. 14, pp. 1284–1301, 2012. J. Wang, M. Venkatachalam, and Y. Fang, “System Architecture and Cross-Layer Optimization of Video Broadcast over WiMAX,” IEEE Journal on Selected Areas in Communications, vol. 25, no. 4, pp. 712– 721, 2007. B. Lin, P.-H. Ho, X. Shen, and F. C.-W. Su, “Network Planning for NextGeneration Metropolitan-Area Broadband Access under EPON-WiMAX Integration,” in Proc. IEEE Globecom 2008, New Orleans, LA, USA, pp. 1–5, 2008. IEEE Std. 802.3ah, Ethernet in the First Mile, June 2004.

50,000,000 45,000,000

Throughput (bps)

40,000,000 35,000,000 30,000,000 25,000,000 20,000,000 LP

15,000,000

MAX_GAIN

10,000,000

NCT

5,000,000 0 25

Figure 1. EPON-WiMAX integrated network

35

40

45 50 Number of SS

55

60

65

70

Figure 2. The overall system throughput 18,000,000

5,000,000 4,500,000

16,000,000

4,000,000

14,000,000

3,500,000

Throughput (bps)

Throughput (bps)

30

12,000,000

3,000,000

10,000,000

2,500,000 2,000,000

LP

1,500,000 1,000,000 500,000

8,000,000

LP

6,000,000 MAX_GAIN

MAX_GAIN

4,000,000

NCT

2,000,000

NCT

0

0 25

30

35

40

45 50 Number of SS

55

60

65

25

70

30

35

40

45 50 Number of SS

55

Figure 4. Throughput of BE traffic

Figure 3. Throughput of UGS traffic

754

60

65

70

Downlink Resource Allocation with Cooperative Transmission in EPON-WiMAX Integrated Network Shang-Hsiu Tien

Hung-Yi Teng

Ren-Hung Hwang

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Dept. of Computer Science and Information Engineering National Chung-Cheng University Chiayi, Taiwan, R.O.C [email protected]

Abstract—The rapid growth of the demand of multimedia applications in recent years has posted a much higher bandwidth demand than before. As a result, the technology of broadband access is becoming more and more important. Integrating Ethernet Passive Optical Network (EPON) and Worldwide Interoperability for Microwave Access (WiMAX) has been considered as a very promising solution for broadband access. Although EPON-WiMAX has received many attentions in research, yet most of the previous researches focus on the uplink packet scheduling and bandwidth allocation. Thus, in our research, we focus on the downlink scheduling and resource allocation problem. In particular, we focus on utilizing the cooperative transmission technique. Based on the advantage of EPONWiMAX architecture, the resource allocation problem utilizing cooperative transmission can be done at the optical line terminal (OLT) in a centralized manner. We first formulate the problem as a linear programming (LP) problem. Due to the high complexity of LP, we propose a heuristic algorithm to solve the problem. Our simulation results show that the proposed algorithm is able to increase the system throughput while guaranteeing the QoS of different classes of traffic. Keywords-EPON-WiMAX integrated network; resource allocation; cooperative transmission

I.

downlink;

INTRODUCTION

In recent year, a variety of multimedia applications such as high-definition television and video-on-demand have posted a high bandwidth demand. As a result, the broadband access technology has become more and more important. Integrating Ethernet Passive Optical Network (EPON) [1] and Worldwide Interoperability for Microwave Access (WiMAX) [2][3] has been considered as a very promising solution for achieving broadband Fixed/Mobile convergence [4][5]. Fundamentally, EPON and WiMAX are well-matched with each other in the aspect of service objective, multiple access mechanism, and capacity hierarchy. In general, the architecture of the EPONWiMAX integrated network is a tree topology, consisting one EPON Optical Line Terminal (OLT), one 1:N passive optical splitter, and multiple integrated units, referred to as ONU-BS, which include an EPON Optical Network Unit (ONU) and a WiMAX Base Station (BS). Advantages of the integrated EPON-WiMAX network include providing broadband Internet access, supporting mobile users, and reducing the network design and maintenance costs.

For seamless integration of EPON and WiMAX, uplink packet scheduling and resource allocation have received much more attention. Yang et al. [8] presented a converged network architecture based on the concept of virtual ONU-BS (VOB) and proposed a QoS-aware dynamic bandwidth allocation (DBA) scheme. Jung et al. [9] investigated three possible integrated architectures and proposed a centralized scheduling mechanism to enhance end-to-end delay and provide better QoS provisioning for the lower priority traffic. Dias Piquet and Saldanha Fonseca [10] assessed the performance of a standardcompliant WiMAX uplink scheduler and showed that the scheduler can provide QoS support to the SSs. Relatively less attention has been focused on downlink packet scheduling and resource allocation on the EPON-WiMAX integrated network. Our previous work [11] presented a two-stage design of downlink packet scheduling and resources allocation to achieve high system throughput while QoS of each traffic class could be guaranteed. But, the inter-cell collaboration was not considered. Emphasizing on inter-cell cooperative transmission, Gong et al. [12] proposed three schemes to optimize ONU-BS user association and resources allocation (BUA-RA) in terms of minimizing the number of rejected connection requests. However, QoS guarantee for different traffic classes was not considered, which is essential to WiMAX. In this paper, we focus on the downlink scheduling and resource allocation problem. In particular, we focus on how to properly utilize the cooperative transmission technique. Based on the advantage of EPON-WiMAX architecture, we formulate the resource allocation problem with inter-cell collaboration as a linear programming (LP) problem. Due to the high computation complexity of LP, we further propose a heuristic algorithm to solve the problem. Our simulation results show that

To achieve throughput enhancement, advanced inter-cell collaboration is possible by employing cooperation transmission techniques such as Space-Time Coding [6], Spatial Multiplexing

978-1-4799-2358-8/14/$31.00 ©2014 IEEE

[7], and Beamforming among distributed ONU-BSs. In this paper, we assume Space-Time Coding is employed to perform inter-cell collaboration. In the EPON-WiMAX integrated network, the OLT plays an important role to support inter-cell collaboration in a centralized manner. In this manner, several ONU-BSs can encode the data with orthogonal space-time codes and transmit cooperatively to a subscriber station (SS). Then, the SS can aggregate the same copy of the data and decodes them in high chance. Although inter-cell collaboration is promising to enhance throughput of SSs located at cell edge. Unfortunately, it may consume more system resources as resources need to be allocated at cooperating BSs. Therefore, how to decide whether use inter-cell collaboration for a cell edge SS is crucial.

750

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

to Interference plus Noise Ratio (SINR), and the selected MCS choice.

the proposed algorithm is able to increase the system throughput while guaranteeing the QoS requirements of each traffic class. The rest of the paper is organized as follows. Section II gives an overview of our system model. In Section III, we propose our approaches to effectively allocate resources so that the system throughput can be improved. Simulation results are presented in Section IV. Conclusions are finally drawn in Section V. II.

1

(4) (5) log 1 (6) In equations (4), (5) and (6), SLER denotes the slot error rate; is the total number of time slots in a MAC PDU which equals / ; is the mean capacity, the constants A and B are chosen to fit the selected MCS curve for the given channel model. 10

SYSTEM MODEL

A. Network Environment Figure 1 shows the architecture of an EPON-WiMAX integrated network which is composed of an EPON as backhaul and multiple WiMAX networks as front-end wireless access networks. In the EPON-WiMAX integrated network, the major components include an optical line terminal (OLT), a passive splitter, and multiple ONU-BSs. As shown in Figure 1, the OLT is connected to multiple ONU-BSs via a passive splitter. It has powerful computing capacity to manage packet traffic efficiently. As proposed in [5], the ONU-BS could be a hybrid device integrating an EPON ONU and a WiMAX BS. It functions as a bridge device to achieve seamless integration between EPON and WiMAX. The resource management functions are located at the OLT so that the resource can be allocated to all SSs in a centralized manner. We assumed that each ONU-BS can access full context of all SSs within its coverage, including rate requirement, distance and signal strength, etc. And, it periodically sends the information of its SSs to the OLT using MultiPoint Control Protocol (MPCP) [15]. Therefore, the OLT can obtain the required information of all SSs in the EPON-WiMAX integrated network. B. Problem Formulation In our model, without cooperative transmission, the data rate of a subscriber station (SS) n is expressed as follows [11]:

∙

However, by employing cooperative transmission techniques such as Space-Time Coding, a SS is possible to be cooperatively served by multiple ONU-BSs. Therefore, SINR of SS n can be estimated as follows. (7) ∑ ∑

(8) (9)

1

where

1 (10) 0 For simplicity, we use Log-Distance path loss model to estimate the received power strength of SS n from ONU-BS m. ∙

(11)

In this study, we aim to maximize the sum of individual received bandwidth (throughput) of each SS assuming that there are M ONU-BSs and N SSs in the network. Our problem is formulated as how does the OLT properly select the serving ONU-BS(s) and the MCS for each SS such that the aggregated network throughput could be maximized. Our objective function is written as follows: ∑ (12) Maximize which is subject to the following constraints: 1) Total time slot constraint ∑ 1, ∀ 1… (13) 2) Data rate requirement of each user ∙ ∙ ,∀ 1 … (14) ∙ 3) BS-User association constraint ∑ 3, ∀ 1… (15) 4) MCS selection constraint , (16) Equation (13) indicates that for each ONU-BS (m), the number of the time slots ( ) that are allocated to the SS does not exceed the total number of time slots (B). Equation (14) indicates that the data rate requirement of each SS must be satisfied by the selected solution. Equation (15) represents that a SS is only able to associate with three ONU-BSs at most [14]. Equation (16) indicates that through an appropriate selection of MCS, the packet loss rate ( ) of SS n should be less than the threshold specified in the WiMAX standard. Table I shows important notations used in this study.

∙ ∙ (1) where represents the number of bits per slot to the MCS denotes the normalized time slots adopted by SS n, allocated to SS n, B is the total wireless bandwidth (in number of time slots). Given a specific MCS choice m, the number of bits which can be transmitted in a time slot is ∗ ∗ (2) denotes the number of carrier symbols per slot, denotes the number of bits per carrier symbol to MCS denotes the channel coding rate to MCS m in m, and physical layer. A time slot is the minimal data allocation unit which is defined to be a collection of subchannels along contiguous OFDM symbols. Let Dn be the amount of data sent from the OLT to SS n in a predefined period. Based on the MCS choice adopted by SS n, the time slots allocated to SS n is where

∙

1

(3)

Each payload of upper layer is fragmented into fixed-size (LByte) SDUs. Each MAC_PDU contains 10-Byte MAC header, L-Byte SDU and 4-Byte CRC. Let be the MAC_PDU size. Wang et al. [13] presented equations (4)–(6) for calculating the packet loss rate given the signal strength, represented by Signal

TABLE I. LIST OF NOTATIONS

751

Symbol

Meaning

M

The total number of ONU-BS in the system

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

N

The total number of SS in the system

B

The total wireless bandwidth (time slots)

Tsmn Amn

The associating indicator of ONU-BS m and SS n

Rn

The actual bandwidth (bps) allocated to SS n

x

The SINR value of SS n

mn

The modulation and coding scheme adopted by SS n

Pn

The packet loss rate of SS n

TPER

The threshold of packet loss rate in WiMAX

III.

PROPOSED SCHEMES

0, ∀j 0 … J, ∀n 4) Variables Ts ∈ 0,1, … , x ∈ 0, 1

…

1

1

1

Step

(17) 1…M

(21)

1 … N

(22) (23) (24)

th transmission mode is used for SS n otherwise

3:

Calculate

_

,

.

If

_ 1 , the serving set will be selected. Otherwise, the combination of ONU-BSs that produces the largest gain will be selected. Step 4: Select the SS n that has the minimal resource consumption to perform resource allocation. If there are multiple SSs have the same resource consumption, the SSs chose the serving set will be selected preferentially. Step 5: Check if all the selected ONU-BS(s) have enough resource. If yes, the selected ONU-BS(s) will serve the SS n and the OLT update the time slot usage of the selected ONU-BS(s) by equations (27) and (28). Otherwise, the SS n will not be served at this frame. Go back to Step 4 until the system resource run out or no SS needs to be allocated. (27) (28)

Then the linear programming problem can be written as follows. Objective: Maximize ∑ R 1) Total time slot constraint ∑ ∑ Ts B , ∀m b 2) Data rate requirement of each user

1 0

(20)

(26)

0 0 0

…

0

0 0 0 …

0

0 0 0

…

… 0

…

… J

… … … …

(19)

Step 1: For each SS, find three closest ONU-BSs denoted as , , and . , , and are sorted according to SINR. , , , Step 2: Set the serving sets and , , . Calculate the consumed resources (time slots) of the serving sets , , and denoted as , , and using equation (26).

TABLE II. TABLE OF TRANSMISSION MODE 0 0 1

1…N

A. The MaxGain heuristic Algorithm be the request rate of SS n. Let be the total time Let slot of ONU-BS m, be the available time slot of ONU-BS m, be the allocated time slot of ONU-BS m ( and ). The OLT conducts the following steps to allocate resources.

A. Linear Programming Since equations (13) and (15) are non-linear, we apply the following approach, similar to [12], to linearize the equations. The possible transmission modes are defined as, J, which indicate a combination of ONU-BSs is selected to serve a SS. Table II shows the binary representation of the jth transmission is used to indicate whether the mode. The binary variable mth ONU-BS is served in the jth transmission mode. For example, the transmission mode j=0 represents none of ONUBSs is selected to serve a SS and j=J represents the SS is served by ONU-BSs M-2, M-1, and M. Due to equation (14), the maximal number of transmission modes for each SS is 1 M .

0 1 0

0, ∀n

(25) Equation (18) ensures that the total allocated time slots to each SS do not exceed the total time slots of a single WiMAX frame. Equations (19) and (20) indicate that the data rate requirement of each SS must be guaranteed. Equations (21) and (22) restrict that a SS can only use a single transmission mode. Equations (23) and (24) define the valid ranges of the decision variables.

Cooperative transmission is a promising technique to enhance throughput of SSs located at cell edge. Unfortunately, it may consume more resources if not properly designed. Therefore, for a cell edge SS, how to decide whether to use cooperative transmission technique or not is crucial. To solve the resource allocation problem, we present two approaches in this paper. The first approach is to formulate this problem into a linear programming problem. Thus, the optimal solution can be obtained by using LP solvers such as LINDO or CPLEX. Due to the high computation complexity of LP, we further propose a heuristic algorithm to obtain a near optimal solution.

j 0 1 2

Ts

The request rate (bps) of SS n SINRn

R R , ∀n 1 … N 3) BS-User association constraint ∑ x 1 , ∀n 1 … N

The normalized resource (%) allocated to SS n from ONU-BS m The resource (time slots) allocated to SS n from ONU-BS m

Tpmn

∑

R

(18)

752

IV.

PERFORMANCE EVALUATION

In this section, we investigate the performance of our approaches via simulations. The simulation program is written by C language and the LP problem is solved by LINDO v.6.1. The simulations were conducted on a computer with Intel Core i5-3230M CPU and 4GB memory. We adopt system throughput

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

and computation time as the performance metrics to evaluate transmission and computation efficiency of our approaches.

B. Simulation results Figure 2 represents the overall system throughput of three approaches. When the number of SS is more than 45, the LP approach is not able to yield an optimal solution since the system resource is not sufficient to satisfy QoS requirements of all SSs. The overall system throughput of the Max_Gain approach is very close to that of the LP approach. When the number of SS is 45, the throughput of the LP approach is only 5.24% higher than that of the Max_Gain approach. However, the Max_Gain approach significantly outperforms than the NCT approach in all cases. When the number of SS is 25, the throughput of the Max_Gain approach is 22.53% higher than that of the NCT approach. As the number of SSs grows, the difference between the Max_Gain approach and the NCT approach slightly decreased. It is because the system resources become scarce and therefore the throughput of cell edge SSs cannot be enhanced via cooperative transmission.

A. Simulation parameters In the simulation, the network topology consists of one OLT and seven ONU-BSs as shown in Fig. 1. The distance between ONU-BSs is 1400 meters. Simulation parameters and QoS requirements of the four traffic classes are summarized in Table III and IV, respectively. Each SS has one connection in each traffic classes, namely, four connections in total. In addition, all ONU-BSs have adaptive modulation and coding schemes capacity as shown in Table V. We simulate a general scenario where all SSs are randomly placed. The number of SS is varied from 25 to 70. We compare our approach (Max_Gain) with two approaches under different number of SSs; in the first approach, the ONU-BS does not employ cooperative transmission technique (NCT); the other approach is the LP approach. The solution yielded by the LP approach is regarded as a benchmark. We assume that the strict priority scheduling is adopted by ONU-BSs when scheduling four classes of traffic. For each setting, we run simulation 30 times and calculate the average results as well as the 95% confidence interval.

Next, we further examine the throughput for individual traffic class. Due to shortage to space, we only discuss the performance of UGS and BE traffics here. Figure 3 represents the throughput of UGS traffic. Again, the LP approach cannot provide the optimal solution when the number of SS is more than 45. When the number of SS is less than 50, the simulation results show that the Max_Gain approach is able to provide a very near optimal solution in UGS traffic. Also, the Max_Gain approach significantly outperforms than the NCT approach in UGS traffic regardless of the number of SS. We also observe the similar trend in rtPS and nrtPS traffics. This fact indicates that the Max_Gain approach can effectively allocate resource to each SS so that the QoS requirements of UGS traffic is satisfied. Figure 4 shows the throughput of BE traffic. In Fig.4, when the number of SS is less than 55, the Max_Gain approach can provide a better throughput of BE traffic than the NCT approach. But, when the number of SS is more than 55, the throughput of BE traffic of the Max_Gain approach decreases slightly. The reason is that the Max_Gain approach can serve the number of SS more than the NCT approach due to inter-cell collaboration. And, most resources are preferentially allocated to the UGS, rtPS and nrtPS traffics. As a result, BE traffic can only obtain less resources when the system resource become scarce. However, this fact will not be a problem since BE traffic does not have any QoS requirements.

TABLE III. SYSTEM PARAMETERS USED IN SIMULATIONS Parameter

Value

WiMAX system bandwidth

10 MHz

WiMAX frame duration

5ms

FFT Size

1024

DL:UL

2:1

Number of OLT

1

Number of ONU-BS

7

Number of SS

25 - 70

Number of connection

100 - 280 100 1 * 10 4

TABLE IV. SPECIFICATION OF TRAFFIC CLASSES Class

Traffic rate

Traffic model

Inter-arrival time

QoS parameters

UGS

64kbps

CBR

20ms

Delay: 20ms

rtPS

176kbps

VBR (poisson)

20ms

nrtPS

232kbps

20ms

BE

360kbps

VBR (poisson) CBR

Delay: 60ms Min. reserve rate: traffic rate Min. reserve rate: traffic rate

Last, we examine computation efficiency of our approaches. The computation time of the LP and Max_Gain approaches is shown in Table VI. It is obvious to see that the LP approach takes very much time to obtain the optimal solution especially when the number of SS is 45. And, it cannot yield an optimal solution when the number of SS is more than 45. However, no matter how many SSs have, the Max_Gain approach only requires no more than 1 millisecond to obtain a near optimal solution. This result indicates that the Max_Gain approach can easily handle fast and frequent state fluctuation due to user mobility in EPON-WiMAX integrated networks.

20ms

TABLE V. MCSS USED IN SIMULATIONS Modulation BPSK QPSK QPSK 16-QAM 16-QAM 64-QAM 64-QAM

Coding rate 1/2 1/2 3/4 1/2 3/4 2/3 3/4

SINR threshold 3.0 6.0 8.5 11.5 15.0 19.0 21.0

Byte/slot 3 6 9 12 18 24 27

TABLE VI. COMPUTATION TIME (MILLISECOND)

753

2014 International Conference on Computing, Networking and Communications, Wireless Networks Symposium

V.

[5]

CONCLUSION

EPON-WiMAX is a promising solution for multimedia applications with mobility support. In this study, we focused on the downlink resource allocation problem with inter-cell collaboration in the EPON-WiMAX integrated network. We proposed two approaches in this paper. The first approach is to formulate this problem into a linear programming problem. Due to the high computation complexity of LP, we further propose a heuristic algorithm to obtain a near optimal solution very efficiently. Our simulation results show that the proposed algorithm can increase the system throughput by properly employing cooperative transmission while guaranteeing the QoS of each traffic class.

[6]

[7]

[8]

[9]

ACKNOWLEDGMENT The authors would like to thank the National Science Council of the Republic of China, Taiwan for financially supporting this research under Contract No. NSC 100－2221－ E－194－012－MY3.

[10]

[11]

REFERENCES [1]

G. Kramer and G. Pesavento, “Ethernet passive optical network (EPON): building a next-generation optical access network,” IEEE Communications Magazine, vol. 40, no. 2, pp.66–73, Feb 2002. IEEE 802.16, “IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access Systems,” 2004 IEEE 802.16e, “IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems,” Feb. 2005. C. Ranaweera, E. Wong, C. Lim, and A. Nirmalathas, “Next Generation Optical-Wireless Converged Network Architectures,” IEEE Network Magazine, pp. 22-27, Mar./Apr. 2012.

[2]

[3]

[4]

[12]

[13]

[14]

[15]

G. Shen, R. S. Tucker, and C.-J. Chae, “Fixed Mobile Convergence Architectures for Broadband Access: Integration of EPON and WiMAX,” IEEE Communication Magazine, vol.45, no.8, pp.44–50, Aug. 2007. V. Tarokh, N. Seshadri, and A. R. Calderbank, ”Space-Time Codes for High Data Rate Wireless Communication: Performance Criterion and Code Construction,” IEEE Transaction on Information Theory, vol. 44, no. 2, pp. 744–764, 1998. Qinghua Li, Xintian Lin, Jianzhong Zhang and Wonil Roh, “Advancement of MIMO technology in WiMAX: from IEEE 802.16d/e/j to 802.16m,” IEEE Communications Magazine, vol. 47 , no. 6, pp. 100– 107, 2009. K. Yang, S. Ou, G. Ken, and H.-H. Chen, “Convergence of Ethernet PON and IEEE 802.16 Broadband Access Networks and its QoS-Aware Dynamic Bandwidth Allocation Scheme,” IEEE Journal on Selected Areas in Communications, vol. 27, no. 2, pp. 101–116, 2009. B. Jung, J.Y. Choi, Y.-T. Han, M.-G. Kim, M. Kang, “Centralized Scheduling Mechanism for Enhanced End-to-End Delay and QoS Support in Integrated Architecture of EPON and WiMAX,” Journal of Lightwave Technology, vol. 28, no. 16, pp. 2277–2288, 2010. M. Dias Piquet, and N.L. Saldanha Fonseca, “Performance Evaluation of a Scheduler for the ONU-BS of Integrated EPON-WiMAX Networks,” IEEE Latin America Transactions, vol. 10, no. 3, pp. 1838–1843, 2012. Hung-Yi Teng, Chung-Chih Kuo and Ren-Hung Hwang, “Downlink packet scheduling and resource allocation in EPON-WiMAX hybrid access networks,” EURASIP Journal on Wireless Communications and Networking, 2012:333. M. Gong, B. Lin, P.-H. Ho, P. Hung, “Adaptive BU association and resource allocation in integrated PON-WiMAX networks,” Wireless Communications and Mobile Computing, vol. 12, no. 14, pp. 1284–1301, 2012. J. Wang, M. Venkatachalam, and Y. Fang, “System Architecture and Cross-Layer Optimization of Video Broadcast over WiMAX,” IEEE Journal on Selected Areas in Communications, vol. 25, no. 4, pp. 712– 721, 2007. B. Lin, P.-H. Ho, X. Shen, and F. C.-W. Su, “Network Planning for NextGeneration Metropolitan-Area Broadband Access under EPON-WiMAX Integration,” in Proc. IEEE Globecom 2008, New Orleans, LA, USA, pp. 1–5, 2008. IEEE Std. 802.3ah, Ethernet in the First Mile, June 2004.

50,000,000 45,000,000

Throughput (bps)

40,000,000 35,000,000 30,000,000 25,000,000 20,000,000 LP

15,000,000

MAX_GAIN

10,000,000

NCT

5,000,000 0 25

Figure 1. EPON-WiMAX integrated network

35

40

45 50 Number of SS

55

60

65

70

Figure 2. The overall system throughput 18,000,000

5,000,000 4,500,000

16,000,000

4,000,000

14,000,000

3,500,000

Throughput (bps)

Throughput (bps)

30

12,000,000

3,000,000

10,000,000

2,500,000 2,000,000

LP

1,500,000 1,000,000 500,000

8,000,000

LP

6,000,000 MAX_GAIN

MAX_GAIN

4,000,000

NCT

2,000,000

NCT

0

0 25

30

35

40

45 50 Number of SS

55

60

65

25

70

30

35

40

45 50 Number of SS

55

Figure 4. Throughput of BE traffic

Figure 3. Throughput of UGS traffic

754

60

65

70