Cost Efficient Algorithm for ONU Placement in Fiber ...

Available online at www.sciencedirect.com

ScienceDirect Procedia Computer Science 46 (2015) 1303 – 1310

International Conference on Information and Communication Technologies (ICICT 2014)

Cost Efficient Algorithm for ONU Placement in Fiber-Wireless (FiWi) Access Networks Uma Rathore Bhatta,*, Nitin Chouhana, Rakhsa Upadhyaya a

Dept. of Electronics and Telecommunication,Institute of Engineering and technology, Devi Ahilya University, Indore,452001,India

Abstract The FiWi network is a combination of optical backend and wireless front end. One of the key issues in FiWi network is ONU placement since it is related to network deployment cost. The present work proposes a cost efficient algorithm for ONU placement. Initially we place ONUs in center region of each grid. Then we minimize the number of ONUs by ensuring that all the wireless routers are connected to at least one ONU. Simulation results show that proposed algorithm offers better solution for optimal placement of ONUs. Hence, by using proposed algorithm we can make cost-efficient FiWi network. © This is an open access article under the CC BY-NC-ND license © 2015 2014The TheAuthors. Authors.Published PublishedbybyElsevier ElsevierB.V. B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the International Conference on Information and Communication Peer-review under responsibility of organizing committee of the International Conference on Information and Communication Technologies (ICICT 2014). Technologies (ICICT 2014) Keywords: Fiber-Wireless (FiWi) ; Passive Optical Network (PON) ; Wireless Mesh Network (WMN);ONU Placement.

1. Introduction In recent years, the increased number of internet users has created a demand of low cost and high bandwidth network. The user wants to access internet in Anywhere-Anytime fashion. The existing networks such as Passive Optical Network (PON) and Wireless Mesh Network (WMN) provide broadband services to user. PON provides higher bandwidth capacity to users but due to usage of optical devices the overall network cost increases. Also the PON suffers from limited connectivity due to requirement of physical channel to the end-users. According to According to technique used PON may be categorized as TDM PON, WDM PON, Hybrid TDM/WDM PON and

* Corresponding author. Tel.: +91-942-533-3837; fax: +91-731-276-4385. E-mail address:[email protected]

1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the International Conference on Information and Communication Technologies (ICICT 2014) doi:10.1016/j.procs.2015.01.055

1304

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

OFDM PON. Another technology, WMN provides services to user at lower cost with better flexibility, which is ideal for mobile users. The limitation of WMN is lower bandwidth and interference problem. The WMN network includes WiFi, WiMAX and cellular technology. Therefore, various researches are carried to integrate the technical merits of PON and WMN network to form Fiber–Wireless (FiWi) network1-5. Initially FiWi called as Hybrid optical/wireless broadband access network. FiWi provides many advantages to users such as cost-effectiveness, higher bandwidth, better flexibility and reliability. The architecture of FiWi is shown in Fig. 1. FiWi consist PON network at back-end whereas WMN network at frontend. The main component of back-end i.e. PON has Optical Line Terminal (OLT), Remote Node (RN), Optical Network Unit (ONU) and fibers whereas WMN has wireless gateways, wireless routers and end-users. The ONU and wireless gateway combined to work as an interpreter between two ends. The optical signal to wireless signal and vice-versa is converted through ONU. In FiWi, to access internet services packets are sent to primary ONU via wireless router. The ONU forwards this packet to OLT via feeder fiber, RN and distribution fiber. Finally OLT inject the packets into internet backbone. In a reverse fashion the services are provided to the users.

Fig. 1. Architecture of FiWi Network

Survivability, ONU placement and energy saving are the key issues in FiWi network. Survivability is defined as the ability to provide services to users if any failure occurs in the network. Mainly survivability is related to backend since it has tree topology. ONU placement means how optimally network designer can place the ONUs in the network i.e. with fewer number of ONUs all wireless routers can communicate. This issue is directly related to network deployment cost. Due to increasing concern in global warming, energy saving is the current issue in all the networks. If an ONU has low traffic then it transfers its traffic to other ONUs and goes into sleeping mode and if traffic increases then ONU comes into awakening mode. By adding such criterion in ONU we can save energy in FiWi network. As mentioned earlier ONU works as an interpreter between both PON and WMN technology. Therefore, placement of ONU is very important in FiWi. Generally two types of traffic are mainly considered in FiWi viz. Internet traffic and p2p traffic6,7. Internet traffic originates from user and goes to internet backbone whereas p2p traffic originates from one user and goes to another user. In this paper, we propose a cost efficient algorithm for ONU placement, taking into account both type of traffic. ONUs are placed in the center region of each grid first, in the proposed algorithm, and then we minimize the number

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

1305

of ONUs in such a way that with minimum number of ONUs all the wireless routers can communicate to its primary ONU. The rest of the paper is organized as follows. Section 2 describes related work done for ONU placement in FiWi technology. Sections 3 explain system model. The proposed algorithm for ONU placement is described in section 4. Section 5 shows simulation results. Finally, we conclude the paper in section 6. 2. Related Work Various algorithms are proposed for ONU placements with the consideration of internet and p2p traffic. The algorithms which consider internet traffic only are random, deterministic, greedy, SA approach and primal model812 . The algorithms which consider both type of traffic are Tabu search 6 and LBOP algorithm7. All the algorithms8- 12, divide the network into multiple non-overlapping regions. In random approach the ONUs are placed randomly in each region. In deterministic approach the ONUs are placed in the center region of each grid. Both approaches are not suitable for ONU placement because it do not provide proper connectivity in the network. In greedy approach, first ONUs is placed in center region and according to minimum distance of ONU from the users, primary ONU are identified for all users. Another approach called Simulated Annealing (SA) consists of five phases which are Initialization, Perturbation, Cost Calculation, Acceptance and Update phase. In initialization phase, ONU placement is done by greedy algorithm and perturbation phase relocates ONUs with small random distance. Author observed new cost changes with respect to old cost in Cost calculation phase. In acceptance phase, if new cost is lower than old cost then new cost is accepted otherwise remains with old cost. Same process repeated until no further cost improvement occurs in update phase. Among all the algorithms it is demonstrated that greedy algorithm outperforms than SA in terms of processing requirement. An analytical model viz. primal model is presented13, for optimum ONU placement. This model considers user assignment, capacity and interference constraint also. They used Lagrangean relaxation method to solve the primal model. The above algorithms discussed so far considered only internet traffic. Currently researchers are also considering p2p traffic along with internet traffic in FiWi network. Zeyu Zheng et. al.6 considered both type of traffic and proposed an algorithm called Tabu search algorithm. The objective of this algorithm is to minimize the total wireless hop number from routers to ONUs. Tabu search is used for this purpose. The author’s work remains untouched in minimizing the number of ONUs. With the concern of minimizing the number of ONUs in the network Yejun Liu et. al.7 proposed load balance ONU placement (LBOP) algorithm. They first place the ONUs in greedy manner and then find the best location for minimizing the number of ONUs. After that they applied the load balancing mechanism. 3. System Model Following consideration are taken to design system model. • Network area is considered as L×L. • Grid size is considered as ୱ × ୱ. • ONUs are placed in center region of each grid. • It is assumed that the maximum traffic on one ONU is equal to the traffic of their subordinate wireless router. Each router has two type of traffic i.e. internet traffic and p2p traffic. Internet traffic flows from user end to internet backbone. The traffic first goes to nearby wireless router, then router routs the traffic to its primary ONU via wireless hop way. ONU forward this traffic to OLT through distribution fiber, RN and feeder fiber. OLT inject this traffic to internet backbone. Since optical back-end has higher data rate than wireless front-end hence end-toend delay for internet traffic mainly depends on wireless end. p2p traffic originates from one router and goes to another router i.e. one user to other. It has mainly two paths pure wireless path and wireless-optical-wireless path. In pure wireless path the traffic goes through wireless routers. In wireless-optical-wireless path the traffic goes through routers, ONU and OLT. Initially router sends this traffic to its primary ONU. ONU forward this traffic to OLT through fibers and RN. Now OLT send this traffic to all the ONUs of the segment. Then ONUs determines either drop the packet or forward, according to destination address.

1306

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

Lastly ONU forward this traffic to destination router via wireless hops way which is the primary ONU of destination router. The end-to-end delay is less for wireless-optical-wireless path as compared to pure wireless path due to higher bandwidth of the optical back-end. Fig. 2. illustrate the routing in p2p traffic. If a user A want to communicate to user B then in pure wireless path it requires 5 hops whereas in wireless-optical-wireless path it only requires 4 wireless hops.

Fig. 2. Illustration of routing in p2p traffic7

4. Proposed Algorithm For making cost-efficient network we propose an efficient algorithm for ONU placement. In this algorithm, the ONUs are placed in the center region of each grid initially. To minimize the number of ONUs, the algorithm works in two stages. Stage 1: In first stage, we form a set of wireless routers in a grid with corresponding ONU. We identify the empty grid in which no wireless routers are present. We remove ONUs from empty grid. It is to be noted that the communication within a grid is on the basis of single hop communication. Stage 2: In the second stage, we further minimize the ONUs on the basis of its transmission range. For each ONU we form a superset of wireless router considering its maximum transmission power range in overall network area. This enables multi-hop communication among wireless routers and ONUs. We analyze the superset of each ONU and select that ONU in which highest number of routers can communicate. Then we take the ONU one by one according to their number of wireless routers. We try to remove those ONUs from the network who’s all the wireless routers are in transmission range of other ONUs. In this way we further reduce the number of ONU in the network.

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

1307

Main steps of proposed algorithm as follows: Step 1: Define Grid size. Step 2: Place ONUs at the center region of grid. Step 3: Allow wireless routers to communicate with ONUs within the grid. Step 4: Form the set of wireless routers within the grid for each ONU. Step 5: If there is no wireless routers present in particular grid, then remove that ONUs from the network. Step 6: Allow each ONU to communicate with wireless router of other grid areas by utilizing its maximum transmission power. Step 7: Form the superset of wireless routers for each ONU according to limited wireless hops. Step 8: Arrange ONUs in descending order according to number of wireless routers present in its range. Step 9: If all the wireless routers of particular ONU present in set of other ONUs then remove that ONU from the network otherwise let in remains in the network. We illustrate proposed algorithm with simple example. We divide L×L area into 5×5 grid size and 40 wireless routers are placed randomly in area. Initially we placed ONU at the center region in each grid. Therefore, initially we require 25 ONUs in the network as shown in Fig. 3. From fig. we may observe that from ‫ݑ݊݋‬ଶ - 1 wireless router is connected, from ‫ݑ݊݋‬ଷ - 3 wireless routers are connected and so on. It is also clear from figure that ‫ݑ݊݋‬ଵ ǡ ‫ ଺ݑ݊݋‬ǡ ‫ݑ݊݋‬ଵଷ ǡ ‫ݑ݊݋‬ଶଵ ǡ ‫ݑ݊݋‬ଶଶ ܽ݊݀‫ݑ݊݋‬ଶହ has no wireless routers to communicate. According to proposed algorithm these ONUs (‫ݑ݊݋‬ଵ ǡ ‫ ଺ݑ݊݋‬ǡ ‫ݑ݊݋‬ଵଷ ǡ ‫ݑ݊݋‬ଶଵ ǡ ‫ݑ݊݋‬ଶଶ ܽ݊݀‫ݑ݊݋‬ଶହ ) will be removed from the network. As a result, only 19 ONUs will remain in the network.

100

26

38 17 37 onu5

onu10 7

onu15 13

80

onu20

3 2

24 onu9

onu4

5

35 10 11 23 21 25 onu3 19 onu8

60

36 onu14

12 onu18 32 6 31 33

onu13

4 15 onu7

onu12

0

40 29 onu11 28 16

onu6

20

40

60

onu24 20

onu23 1

onu22

27

39 onu1

0

onu17

9

20

34

onu19 1822

40 onu2 14

onu25 8

30

onu16

onu21

80

100

Fig. 3. Initially placed ONUs at the center region of each grid

Now limit the number of hops for each wireless router. For internet traffic we consider this limit as 2 hops and for p2p traffic as 3 hops in considered transmission range. This means that all the wireless routers will communicate to other ONUs with limited number of hops. We compute the set of wireless routers of each ONUs and minimize the number of ONUs according to steps (7-9) of the algorithm. After execution of algorithm, 7 ONUs will remain in the network. These ONUs are ‫ݑ݊݋‬ଷ ǡ ‫ ଻ݑ݊݋‬ǡ ‫ݑ݊݋‬ଵ଴ ǡ ‫ݑ݊݋‬ଵଵ ǡ ‫ݑ݊݋‬ଵସ ǡ ‫ݑ݊݋‬ଵ଼ ܽ݊݀‫ݑ݊݋‬ଵଽ as shown in Fig. 4.

1308

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

Fig. 4. ONUs remaining in the network after execution of proposed algorithm

5. Performance Evaluation 5.1 Simulation Setting Setting for simulation of proposed algorithm is explained as follows: • Construct a FiWi network in a 100×100 square geographical area. • Take grid area as ‫ܩ‬௦ ×‫ܩ‬௦ , where ‫ܩ‬௦ is {4, 5, 6, and 7} in different cases. • Place ܰ௪ wireless routers in the network randomly, where ܰ௪ takes values {20, 30, 40, 50} in different cases. • Define transmission range of wireless router and ONUs. 5.2 Simulation Result Simulation has been carried out to analyze the performance of proposed algorithm. In this subsection we present the simulation result of proposed algorithm. Fig. 5. shows the result for number of ONUs vs. number of wireless router for 4×4 grid size. The Fig. shows required number of ONUs initially placed in the network and after the execution of proposed algorithm. It is clear from the figure that the requirement of number of ONUs has been drastically reduced using the proposed algorithm for all number of wireless routers considered in the network. Fig. 6. shows the result for number of ONUs vs. grid sizes. We set the number of wireless router ܰ௪  as 30. It is observed that fewer number of ONUs are required in the network for different grid sizes viz. 4×4, 5×5, 6×6 and 7×7. Therefore it can be concluded from Figs. (5 and 6), that the proposed algorithm is cost efficient in terms of fewer number of ONUs required in the network to setup FiWi network.

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

Required number of ONUs in the network

30 25 20 Initially placed ONUs in the network

15

Remaining ONUs in the network after execution of algorithm

10 5 0 20

30

40

50

60

Number of wireless routers Fig. 5. Required number of ONUs in the network vs. Number of wireless routers for 4×4 grid size

Required number of ONUs in the network

55 50 45 40 35 30

Initially placed ONU in the network

25 20

Remaining ONUs in the network after execution of algorithm

15 10 5 0 4×4

5×5

6×6

7×7

Grid Size Fig. 6. Required number of ONUs in the network vs. Grid Size for 30 wireless routers

1309

1310

Uma Rathore Bhatt et al. / Procedia Computer Science 46 (2015) 1303 – 1310

6. Conclusion In this paper, we proposed algorithm for ONU placement. Our aim is to reduce the network cost in terms of number of ONUs. It has been shown in simulation results that, the proposed algorithm requires fewer number of ONUs as compared to initially placed ONUs in the network, for varying grid sizes and wireless routers. Hence the proposed algorithm offers the cost efficient solution for deployment of FiWi network. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Ghazisaidi N, Maier M. Fiber-wireless (FiWi) access networks: a survey. IEEE Commun. Mag.;2007. 47: 160– 167. Ghazisaidi N, Maier M. Fiber-wireless (FiWi) access networks: challenges and opportunities. IEEE Network;2011. 25:36–42. Maier M, Ghazisaidi N, and Reisslein M. The audacity of fiber-wireless (FiWi) networks (invited paper). in Proc. ICST Int. Conf. Access Networks;2008.p. 16–35. Bhatt UR, Chouhan N. ONU placement in Fiber Wireless (FiWi) Network. Engineering (NUiCONE) Nirma University ;2013.p.1-6. Shaddad RQ et. al. A survey on access technologies for broadband optical and wireless network. journal of network and computer application; 2014,41: 459-472. Zheng Z, Wang J, Wang X. ONU placement in fiber-wireless (FiWi) networks considering peer-to-peer communications. In: Proc. GLOBECOM ;2009.p. 1–7. Liu Y, et al. Load balanced optical network unit (ONU) placement in cost-efficient fiber-wireless (FiWi) access network. Optik – Int.J. Light Electron Opt; 2013. Chowdhury P, Mukherjee B, Sarkar S. Hybrid wireless-optical broadband access network (WOBAN): prototype development and research Challenges. IEEE Network;2009.23: 41–48. Sarkar S, Mukherjee B, Dixit S. Towards global optimization of multiple ONU placement in hybrid optical-wireless broadband access networks. in: Proc. of COINNGNCON ;2006.p. 65-67. Sarkar S, Dixit S, Mukherjee B. Hybrid wireless-optical broadband-access network (WOBAN): a review of relevant challenges. J. Lightw. Technol.;2007.25: 3329–3340. Sarkar S, Yen H, Dixit S, et al. Hybrid wireless-optical broadband access net-work (WOBAN): network planning and setup. IEEE J. Sel. Areas Commun;2008. 26: 12–21. Sarkar S, Yen H, Dixit S, Mukherjee B. A mixed integer programming model for optimum placement of base stations and optical network units in a hybrid wireless-optical broadband access network (WOBAN). in: Proceedings of WCNC;2007.p. 3907–3911. Sarkar S, Yen H, Dixit S, et al. Hybrid wireless-optical broadband access network (WOBAN): network planning using Lagrangean relaxation. IEEE/ACM Trans. Network.;2009.17: 1094–1105.