SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Integration of EPON and WiMAX
Gangxiang Shen and Rodney S. Tucker

14.1 Introduction

The Internet today is characterized by a fast growth of bandwidth-intensive services, such as IPTV, video on demand (VoD), and peer-to-peer (P2P) services. To
Gangxiang Shen ARC Special Research Centre for Ultra-Broadband Information Networks (CUBIN), Department Electrical and Electronic Engineering, The University of Melbourne, Melbourne VIC 3010, Australia, e-mail: egxshen@gmail.com Rodney S. Tucker ARC Special Research Centre for Ultra-Broadband Information Networks (CUBIN), Department Electrical and Electronic Engineering, The University of Melbourne, Melbourne VIC 3010, Australia, e-mail: r.tucker@ee.unimelb.edu.au

A. Shami et al. (eds.), Broadband Access Networks, Optical Networks, DOI 10.1007/978-0-387-92131-0 14, c Springer Science+Business Media, LLC 2009

UN C

OR R

Abstract The integration of EPON and WiMAX is a novel research topic that has received extensive interest from both industry and academia. The major motivations behind the integration of EPON and WiMAX involve the potential benefits of fixed mobile convergence (FMC), which uses a single network infrastructure to provide both wired and wireless access services, and a good match of capacity hierarchy between EPON and WiMAX by using EPON as a backhaul (or feeder) to connect multiple disperse WiMAX base stations. This chapter recaps recent progress in the area of integration of EPON and WiMAX. Three different integration architectures, including independent architectures, hybrid architectures, and microwave-over-fiber (MoF) architectures, are described. Based on these architectures, a range of planning and operational issues are discussed, including optimal passive optical network deployment to connect disperse WiMAX base stations, packet forwarding, bandwidth allocation and QoS support, handover operation for mobile users, survivability, and cooperative downstream transmission for broadcast services. We hope that this will interest readers and stimulate further investigation in the area.

EC

TE

DP

RO
301

OF

Chapter 14

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

302
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Gangxiang Shen and Rodney S. Tucker

14.2 Literature Survey

The integration of EPON and WiMAX is a new topic in the literature. However, it has received extensive interest. Preliminary research on integration of EPON and WiMAX was described in [12, 13], under the banner of hybrid optical wireless

UN C

cater to the bandwidth demand of these types of services, access networks are evolving from traditional copper-based ADSL technology to more advanced fiber-based passive optical network technologies [1–3], including fiber to the home (FTTH), fiber to the node (FTTN), fiber to the curb (FTTC). Nowadays, there are two major passive optical network standards, including Ethernet passive optical networks (EPON) [1, 2] and Gigabit-capable Passive Optical Networks (GPON) [3]. EPON is based on the traditional Ethernet techniques, while GPON evolved from the traditional broadband passive optical network (BPON) technology. Meanwhile, in the wireless camp, we also see fast progresses of wireless access technologies, which are evolving from the traditional WiFi technology to more advanced technologies such as WiMAX [4–7] and long-term evolution (LTE) [8] technologies for wider cell coverage and higher access capacity. Driven by potentially significant cost reduction by converging network infrastructures and control systems of wired and wireless access networks, fixed mobile convergence (FMC) [9] is envisioned as a future generation of architecture for broadband access. Motivated by the potential benefits of fixed mobile convergence, this chapter focuses on integration of EPON and WiMAX. In this chapter, we specifically look at integration architectures and their related design and operations. Three types of integration architectures [10, 11] are considered, including (i) independent architectures, which represent the most intuitive way to interconnect an EPON optical network unit (ONU) and a WiMAX base station (BS) over a standardized interface, (ii) more advanced hybrid architectures that are based on an integrated ONU-BS device box, and (iii) microwave-over-fiber (MoF) architectures, which modulate WiMAX carrier signals onto an optical carrier and transmit the carrier signals over fibers together with the EPON baseband signal. Based on these architectures, a range of network design and operational issues are discussed, including (i) passive optical network deployment to optimally interconnect disperse WiMAX base stations to a Central Office (CO), (ii) upstream packet forwarding, bandwidth allocation, and QoS support, (iii) handover operation, (iv) survivability, and (v) downstream cooperative communication for broadcast services. Although the whole framework focuses on EPON and WiMAX, it is extensible to support other types of wired and wireless network combinations, such as GPON [3] and WiMAX, EPON and WiFi, and EPON and LTE. The rest of this chapter is organized as follows. Section 14.2 is a literature survey on integration of optical and wireless access networks. Section 14.3 introduces three integration architectures for EPON and WiMAX. Based on the integration architectures, Section 14.4 discusses related research issues from the perspectives of network design and operation. We conclude the chapter in Section 14.5.

OR R

EC

TE

DP

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

303

14.3 Integrated Architectures for EPON and WiMAX
In [10, 11], we proposed four integration architectures for EPON and WiMAX. This chapter recaps three of them, including (i) independent architectures, (ii) hybrid architectures, and (iii) microwave-over-fiber (MoF) architectures. For the fourth type of architecture, i.e., unified connection-oriented architectures, interesting readers may refer to [10, 11].

14.3.1 Independent Architectures

Independent architectures are the most intuitive case for integration of EPON and WiMAX. Figure 14.1 shows an example of this type of architecture, in which multiple ONUs are connected to a common OLT over a tree passive optical network. To in-terconnect the EPON and WiMAX network segments, the architectures connect a WiMAX base station (BS) to an EPON Optical Network Unit (ONU) over a standardized interface, e.g., an Ethernet interface. In this configuration, the WiMAX BS is a generic user of the ONU, and data from users are processed regularly in both of the devices. In the upstream direction, data from a wireless subscribed station (SS) are forwarded to the WiMAX BS and its connected ONU. The ONU further forward the data over the passive optical network to an OLT, located at a Central Office (CO). In contrast, in the downstream direction, a reverse packet forwarding

UN C

OR R

EC

TE

networks. The hybrid networks in [12, 13] employed a passive optical network as a backhaul to transmit data for a large WiMAX network, in which a single WiMAX base station collocated with an optical line terminal (OLT) functions as a central controller and coordinator to perform operations for the whole WiMAX network, including packet forwarding, bandwidth allocation, etc. To fully exploit the benefits of integration of EPON and WiMAX, more integration architectures were proposed in [10, 11]. These architectures include independent architectures, hybrid architectures, unified connection-oriented architectures, and microwave-over-fiber (MoF) architectures. Based on these architectures, a variety of design and operational issues including network design and planning, packet forwarding, bandwidth allocation and QoS support, handover operation, and network survivability were discussed in [10, 11]. In addition to integration of EPON and WiMAX, other types of optical wireless access networks were also proposed in [14, 15]. In addition to directly over an ONU as in the architectures proposed in [10, 12], a wireless BS, under the architectures in [14, 15], can also send data to gateways/ONUs over other intermediate wireless BSs by taking advantage of wireless mesh networking. For this type of optical wireless networks, research efforts were mainly concentrated on the placement of wireless BSs [14], and routing algorithms and load balancing for packet forwarding in wireless mesh networks [14, 15].

DP

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

304
01

Gangxiang Shen and Rodney S. Tucker

This 02 figure 03 will be 04 printed 05 in b/w
06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Fig. 14.1 Architectures for integration of EPON and WiMAX (adapted from [10], c 2007 IEEE)

14.3.2 Hybrid Architectures

To overcome the disadvantage of two device boxes in the independent architectures, as well as to provide better efficiency and flexibility of bandwidth allocation and packet scheduling in the integrated EPON and WiMAX systems, another type of integrated architecture, as shown in the bottom cell in Fig. 14.1, can be proposed, called hybrid architecture. The architectures integrate an EPON ONU and a WiMAX BS into a single system box, called ONU-BS. The new box realizes all the functionalities of both EPON ONU and WiMAX BS. Moreover, because ONU and WiMAX BS are integrated in a single device, conveniences are provided to promptly exchange network state information in both EPON and WiMAX network segments. Figure 14.2 shows the ASIC chips contained in the ONU-BS and the layout of the functional modules supported by each of the chips. In hardware as shown in the

UN C

OR R

process is performed to forward data from a central OLT to an intermediate ONU, through a WiMAX BS, and eventually to a targeted SS. The independent architectures have the advantage of incorporating standardized devices and interfaces. However, the physical boundary between a WiMAX BS and an EPON ONU may make it difficult to directly exchange control information between the two devices to quickly notify network state changes such as bandwidth allocation and packet scheduling in both upstream and downstream directions. Thus, the architectures may not fully exploit the potential benefits of integration of EPON and WiMAX. Moreover, two independent devices, i.e., a WiMAX BS and an EPON ONU, can be another disadvantage for the architectures, which are likely more expensive than a single integrated device as in other architectures.

EC

TE

DP

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

305

RO
Fig. 14.2 Functional modules and components of ONU-BS (adapted from [10], c 2007 IEEE)

upper part of the figure, the integrated ONU-BS composes three key ASIC chips. The first chip (i.e., ASIC-1) is in charge of the communications in the EPON segment, the third chip (i.e., ASIC-3) is in charge of the communications in the wireless segment, and the middle chip (i.e., ASIC-2) is responsible for the overall control and coordination of data communications between the EPON and WiMAX network segments. For cost saving, the three chips could be further integrated into a single large ASIC chip. The lower part of the figure shows the functional modules of the ASIC chips, which mainly include the modules for upstream data transmission. The modules of ASIC-1 and ASIC-3 are mainly related to the EPON and WiMAX network segments, respectively. The module ONU-BS central controller, which corresponds to ASIC-2, provides overall control and coordination functionalities for the whole integrated device. Hybrid architectures can provide better flexibility and efficiency in bandwidth allocation and capacity utilization than independent architectures. This is because the integrated ONU-BS understands the network state information in both EPON and WiMAX network segments, and it can therefore make the most efficient bandwidth allocation and packet scheduling for the whole integrated system. For example, for upstream data transmission because the ONU-BS understands how much upstream bandwidth has been allocated to associated wireless SSs in a WiMAX segment, though not receiving user data from the SSs yet, the ONU-BS can ask for more

UN C

OR R

EC

TE

DP

OF
This figure will be printed in b/w

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

306
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Gangxiang Shen and Rodney S. Tucker

14.3.3 Microwave-over-Fiber Architectures

38 39 40 41 42 43 44 45

Fig. 14.3 Microwave-over-fiber (MoF) integration architectures and carrier signal spectrum layout using WDM PON (adapted from [11])

UN C

This figure34 will 35 be printed 36 in b/w 37

33

OR R

To better utilize the fiber spectrum (in the previous architectures, both wired and wireless user data are transmitted over the fiber baseband), the third type of integrated architectures is based on microwave-over-fiber (MoF) technology [16], called microwave-over-fiber (MoF) architectures. Under this type of architecture, one can employ either TDM PON or WDM PON [17] for the integration. Specifically, under TDM PON, carrier signals from different WiMAX base stations are modulated onto a common optical carrier together with a baseband signal that carries EPON user data. To distinguish the wireless carrier frequencies, before modulation, frequency shifts are required to ensure that each WiMAX base station corresponds to a certain unique optical subcarrier in the PON system. Further details about this type of MoF architecture (based on TDM PON) can be found in [10]. Using a WDM PON arrangement, the integration does not need to distinguish the wireless carrier frequencies modulated within the PON network, as each wireless carrier signal (for each WiMAX BS) is modulated on a different wavelength in the WDM PON. Figure 14.3 shows a typical example for integrated MoF architectures using WDM PON technology [11]. The WDM PON carries 16 wavelengths and each wavelength carries a baseband signal for WDM PON and an optical subcarrier signal for WiMAX. The layout of carrier signals in the integrated architecture is also illustrated in Fig. 14.3. The hardware in each remote node is made up of an ONU and a dumb antenna. The ONU is responsible for data communication of the WDM PON, and the dumb antenna relays an analog WiMAX signal from and to its associated micro-cell. The

EC

TE

DP

RO

OF

bandwidth for the future user data from the SSs, when making an upstream bandwidth request in the EPON segment. By doing this, a shorter average packet delay and a better system throughput can be expected.

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

307

14.3.4 Multistage EPON and WiMAX Integration
Based on the previous two-stage architectures, we can further extend the integration of EPON and WiMAX. A four-stage integrated EPON and WIMAX system is shown in Fig. 14.4. Specifically, with the standardization of 10G EPON technology [19], it is possible to employ a 10G EPON to function as a backhaul to interconnect multiple EPON OLTs. Meanwhile, the WiMAX standard [4, 6] enables to establish point-to-point radio links between neighboring WiMAX BSs, thereby allowing for establishing a WiMAX star network as shown in Fig. 14.4. In the WiMAX star network, a WiMAX base station connected to an EPON ONU

UN C

OR R

user data of the WDM PON are transmitted in the baseband, which occupies 1.25 GHz bandwidth based on 1 Gb/s EPON data rate, and the user data of WiMAX are modulated on a wireless carrier frequency (e.g., 2.5 GHz). The two signals are multiplexed and modulated onto a common optical carrier frequency and transmitted to a central OLT node. The central node in the MoF architectures comprises two major modules: an array of OLTs and a central WiMAX BS as shown in Fig. 14.3. In the example, there are a total of 16 OLTs in the WDM PON and 16 WiMAX-BS units that, together with a macro-BS central controller/coordinator, make up a WiMAX macroBS. When a modulated optical signal enters the central node, the signal is first converted into an electronic format, and then the latter is demultiplexed into a baseband signal and a WiMAX signal. The baseband signal is further forwarded to its corresponding OLT for data processing, while the WiMAX signal is forwarded to its corresponding WiMAX-BS unit for data processing. Major advantages of the MoF architectures are as follows. First, it can provide efficient optical spectrum utilization as different optical spectrum frequencies are employed to transmit EPON and WiMAX user data. Second, the centralized feature of the WiMAX networks enables efficient bandwidth allocation and packet scheduling and forwarding among different WiMAX micro-cells. Third, the centralized WiMAX macro-BS can provide convenience in handover operation for mobile users. However, as shortcomings, the MoF architectures may suffer from the following disadvantages. First, despite efficiency of centralized control and operation in bandwidth utilization and handover operation, the central macro-BS can become a bottleneck of the whole system as it needs to process all the data packets and bandwidth requests from all the subscribed stations (SSs) (e.g., hundreds of SSs). In addition, the central macro-BS is critical, which can cause the whole system to be out of service if it incurs a failure. Second, with the growing maturity of the highfrequency (e.g., even up to 60 GHz [18]) CMOS technology, the MoF architectures may show disadvantages of higher system cost compared to other integrated architectures such as the hybrid architectures, which employ the high-frequency CMOS technology at the boundary of EPON and WiMAX networks.

EC

TE

DP

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

308
01 02 03

Gangxiang Shen and Rodney S. Tucker

09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Fig. 14.4 Multistage-integrated EPON and WiMAX systems (adapted from [11])

14.4 Design and Operation Issues
Based on the above integrated architectures for EPON and WiMAX, a range of research issues emerge as one looks for a better understanding of the technology. In the following, we discuss these research issues.

UN C

functions as a root node to relay data from all the other neighboring WiMAX base stations. The WiMAX star network can be cost-effective for the network scenarios in rural areas, where users are disperse and aggregate user bandwidth is low. Multistage architectures can cover much larger user service areas than the previous two-stage architectures. For example, if both of the 10G EPON and 1G EPON are equipped with 1:16 optical splitters and the star WiMAX network composes seven WiMAX micro-cells as shown in Fig. 14.4. The whole multistage architecture can cover a huge area, made up of a total of 1792 micro-cells. Moreover, due to the huge service area with numerous users, the benefit of statistical traffic multiplexing can be exploited to maximally improve the capacity utilization of the integrated system. In addition, the multistage architectures also provide convenience in handover operation for mobile users. We will give more detail on the handover operation later in Section 14.4. Finally, multistage architectures allow heterogeneousness of remote nodes. In addition to a WiMAX base station, a digital subscriber line access multiplexer (DSLAM) can be connected to an EPON ONU for DSL services. Also, rather than a 1G EPON ONU, a 10G EPON ONU can be directly connected to a WiMAX base station or a wired DSLAM for even higher bandwidth as shown in the bottom wireless cell in Fig. 14.4.

OR R

EC

TE

DP

RO

This figure05 will 06 be 07 printed in b/w 08

04

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

309

14.4.2 Packet Forwarding

Under the integrated EPON and WiMAX architectures, there are two operational modes for user packet forwarding. User packets can be forwarded in either IP layer or MAC layer. As shown in Fig. 14.5, the IP layer forwarding mode does not require intermediate ONU-BSs to have any packet switching capability. All the user packets are forwarded in the upstream direction to a central OLT and then the latter sends the packets to an edge access IP router. The router makes routing decisions to find where the packets should be further sent, either to the public Internet or downstream back to another local SS. Such a packet forwarding model requires a simple intermediate ONU-BS. Moreover, because each user data packet goes through an edge IP router, better network security can be expected. However, as a shortcoming, we can see that the forwarding mode suffers from a bandwidth waste due to the loopback of the packets forwarded between the SSs that are subscribed to a common ONU-BS as shown in Fig. 14.5.

UN C

OR R

Efficient planning and deployment of the physical layout of integrated EPON and WiMAX systems is an important issue. For this, we need to consider various costs, including hardware costs and operational costs. Hardware costs include the cost of EPON deployment and the cost of WiMAX BSs. In general, the cost of EPON deployment is much higher than that of WiMAX BSs due to the dominance of labor cost for laying fibers. Thus, minimizing the total cost for EPON deployment is a key optimization problem. In the context of integrated architectures, the problem of optimal EPON deployment can be defined as given a set of disperse WiMAX BSs, minimize the total deployment cost for EPON networks that connect all the WiMAX base stations (BS) to a Central Office (CO), subject to a set of EPON system constraints, including (i) maximal transmission distance, (ii) maximal differential distance, and (iii) optical split ratio. According to the EPON standard, the typical value for the maximal transmission distance between an ONU and an OLT is 20 km [1]. The maximal differential distance, which is defined as the maximal distance difference from different ONUs to a central OLT within a common EPON, is 20 km [1]. Finally, EPON can support a range of optical split ratios such as 1:16 and 1:32. For the above research problem, we have developed optimization models and efficient heuristics to find solutions for EPON deployment without considering geographic constraints such as road maps. The heuristics are scalable to find suboptimal solutions for network planning scenarios with hundreds of, or even up to several thousand ONUs. Interested readers may refer to our papers [20, 21]. Subsequent research is required to consider optimal PON deployment taking the geographic constraints such as road maps into consideration.

EC

TE

DP

RO

OF

14.4.1 Optimal Passive Optical Network Deployment

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

310
01 02 03

Gangxiang Shen and Rodney S. Tucker

08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Fig. 14.5 IP layer packet forwarding in integrated EPON and WiMAX networks (adapted from [10], c 2007 IEEE)

This figure35 will 36 be printed 37 in b/w 38
39 40 41 42 43 44 45

34

Fig. 14.6 MAC layer frame forwarding in integrated EPON and WiMAX networks (adapted from [10], c 2007 IEEE)

UN C

OR R

The MAC layer forwarding mode is an alternative to the IP layer mode, which can avoid the bandwidth waste due to the loopback of packet forwarding. This is at the cost of an extra MAC layer switch by each intermediate ONU-BS as shown in Fig. 14.6. Specifically, user packets are first encapsulated into Ethernet frames in the WiMAX network. Then the Ethernet frames are forwarded to an intermediate ONUBS. An Ethernet switch attached to the ONU-BS then directly switches the frames based on their MAC addresses. By doing this, the packets do not experience a loopback forwarding process. Under the MAC layer switching mode, the IEEE 802.3D STP protocol [22] can be employed to establish a simple minimum spanning tree (MST) for the whole integrated system with the Ethernet switch attached to an OLT as a root node. In addition, the virtual LAN (VLAN) (IEEE 802.3Q) protocol [23] can also be deployed for better network security.

EC

TE

DP

RO

This 04 figure05 will 06 be printed 07 in b/w

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

311

Bandwidth allocation is important for both EPON [2, 24] and WiMAX [25] networks. EPON and WiMAX show similarity in bandwidth allocation in both downstream and upstream directions. In the downstream direction, the communications in both EPON and WiMAX follow the point-to-multipoint mode; that is, user data are sent from a central point, such as an EPON OLT and a WiMAX BS, to multiple disperse subscribe stations, such as EPON ONUs and WiMAX SSs. Under this type of transmission mode, no network resource competition or transmission collision occurs because when a central node transmits data to one of its subscribed users, the downstream channel is always uniquely reserved for the transmission. In contrast, bandwidth allocation in the upstream direction is generally more complicated. This is because the transmission in the upstream direction follows the multipoint-to-point mode, under which collisions can occur when multiple subscribed users send data in the upstream direction simultaneously to the central node. In order to avoid the collisions, both EPON and WiMAX have developed a set of mechanisms specially for the upstream bandwidth allocation [1, 4]. A generic poll/request/grant process is a typical example of this type of bandwidth allocation mechanisms. For example, in EPON networks, a central OLT first sends a polling message to each of the subscribed ONUs to check if the ONU has any data to transmit in the upstream direction in the next transmission cycle. The ONUs respond with bandwidth request messages based on their currently accumulated user data stored in their local buffers. Upon receiving such bandwidth request messages, the central OLT makes an optimal time-slot allocation for the different ONUs by taking into consideration of various optimization aspects such as maximizing system capacity utilization, minimizing overall system packet delay, and ensuring the QoS requirement of each ONU user service. Then the OLT notifies each of the ONUs on when they can start transmitting data in the upstream direction in the next transmission cycle. Similar bandwidth allocation mechanisms are devised for the WiMAX networks, in which a central WiMAX BS plays a central role like an OLT in an EPON network. The WiMAX BS polls each of subscribed stations (SSs) enquiring whether they have data to transmit in the upstream direction in the next transmission cycle. The SSs respond with their bandwidth requests and then the central WiMAX BS makes an optimal bandwidth allocation among the different SSs. Moreover, in addition to the above poll/request/grant mechanisms, the WiMAX technology also supports other bandwidth allocation modes such as unsolicited bandwidth allocation [25], which does not need a bandwidth request message from an SS to a central WiMAX BS. The similarities in bandwidth allocation in the upstream direction provide convenience and opportunities of performance optimization for overall bandwidth allocation within the integrated systems. Such advantages can be fully exploited when the hybrid integrated architectures are implemented. Figures 14.7 and 14.8 show examples of upstream data transmission under the independent architectures and the hybrid architectures, respectively, which illustrate how the hybrid architectures can

UN C

OR R

EC

TE

DP

RO

OF

14.4.3 Bandwidth Allocation

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

312
01 02 03 04

Gangxiang Shen and Rodney S. Tucker

09 10 11 12 13 14 15 16 17

Fig. 14.7 Upstream bandwidth allocation in independent architectures

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Fig. 14.8 Upstream bandwidth allocation in hybrid architectures

improve average packet delay in the upstream direction compared to the independent architectures. Figure 14.7 shows an example of upstream bandwidth allocation for the independent architectures. There are several sequential and parallel steps in the example. A polling message is first sent from an OLT to an ONU in Step 1. Meanwhile, a WiMAX BS connected to the ONU grants y units of bandwidth to its SS users in Step 2. Upon receiving the polling message, the ONU responds with a bandwidth request message asking for x units of bandwidth for upstream transmission in Step 3. Under the independent architectures, the ONU has no idea on how much bandwidth that the WiMAX BS has granted for its SS users. Thus, the ONU requests the x-unit bandwidth simply based on the accumulated user data in the ONU local buffer as shown in Fig. 14.7. In Step 4, the WiMAX SS users transmit data in the upstream direction to the BS, and the latter forward the data directly to the ONU over the ONU-BS interface. Upon receiving the data, the ONU stores the data into its local buffer. Now, as shown in Fig. 14.7, after Step 4 there are a total of x+y units of user data stored in the ONU local buffer. In the EPON segment, upon receiving

UN C

OR R

EC

This 18 figure 19 will be 20 printed 21 in b/w

TE

DP

RO

This 05 figure will 06 be 07 printed in b/w 08

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

313

14.4.4 QoS Support and Mapping

14.4.5 Handover Operation
User mobility is an important feature that should be supported by integrated EPON and WiMAX systems. To enable user mobility, integrated systems should be able

UN C

EPON and WiMAX have similar QoS support, which facilitates QoS support in integrated systems. EPON classifies user packets into up to eight priority levels (i.e., eight priority queues) [1]. Likewise, WiMAX classifies user packets into up to five different QoS levels, including unsolicited grant services (UGS), extended real-time polling services (ertPS), real-time polling services (rtPS), non-real-time polling services (nrtPS), and best effort services (BE) [25]. To enable integration, an effective mapping mechanism is required between EPON priority queues and WiMAX different levels of QoS. The mapping should decide which WiMAX flow packets should be stored in which EPON priority queues for equivalent QoS levels. In addition, because EPON classifies packets based on priority queues, it actually supports QoS in a differentiated services (DiffServ) mode [26]. In contrast, because WiMAX supports QoS based on connection-oriented service flows, it actually follows an integrated services (IntServ) mode [27]. When connecting these two types of access networks in an integrated system, it is interesting to understand how QoS-level conversion can be carried out between DiffServ services and IntServ services. Also, it is interesting to understand how end-to-end QoS can be supported in the integrated systems.

OR R

EC

TE

DP

the bandwidth request message from the ONU, the OLT grants x units of bandwidth for the ONU to transmit data in the upstream direction in Step 5. The grant message triggers the ONU to send x units of data in the upstream direction in Step 6, with y units of data remaining in the buffer. In contrast, Fig. 14.8 shows an example of upstream bandwidth allocation under the hybrid architectures. Due to the integration of ONU and WiMAX BS, the ONUBS understands how much bandwidth has been granted for WiMAX SS users in Step 2. The ONU-BS can predict that y units of user data will arrive in the near future from the SS users in Step 4. Thus, when requesting for upstream bandwidth from the OLT, the ONU-BS can ask for y more units of bandwidth, i.e., a total of x+y units of bandwidth, in Step 3. Upon this request, if the OLT accordingly grants x+y units of bandwidth for upstream transmission in Step 5, then in Step 6, x+y units of user data are transmitted without any data remaining in the ONU-BS local buffer. It is clear that the hybrid architectures can transmit upstream user data faster and more efficiently, which means a shorter average packet delay and a better system throughput.

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

314
01 02 03 04 05 06 07 08 09 10 11 12 13 14

Gangxiang Shen and Rodney S. Tucker

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Fig. 14.9 Handover operation in a multistage-integrated system

Under the multistage-integrated architectures as shown in Fig. 14.9, there are four types of wireless cells. The first type of cell is a micro-cell within a WiMAX star network, defined as cluster micro-cell. The second type of cell is a micro-cell covered by an ONU-BS, defined as ONU-BS micro-cell. The third type of cell is called macro-cell, which composes all the micro-cells that are associated with a common central EPON OLT. The last type of cell is the largest, called macro-macrocell, which is made up of all macro-cells associated with a common 10G EPON OLT. Based on these different types of wireless cells, there are four different types of handover. The first type of handover can occur between two cluster micro-cells. For this, a central ONU-BS can function as a central controller to coordinate the handover. The second type of handover can occur between two ONU-BS microcells. In this case, a central OLT that connects to the two ONU-BSs can function as a central controller to coordinate the handover. Handover can also occur between two macro-cells, for example, between macro-cells associated with two different EPON OLTs. In this case, a central 10G EPON OLT that connects the two EPON OLTs can function as a coordinator for the handover. Finally, the last type of handover can occur between two macro-macro-cells that correspond to two different 10G EPON OLTs. In this case, a higher level of coordinator is required for the operation.

UN C

OR R

EC

TE

This figure 16 17 will be printed18 in b/w 19

15

DP

RO

OF

to handle handover operation. The multistage feature of the integrated architectures provides convenience for user handover operation. In each stage, we can always find a central coordinator that can handle and coordinate handover operation. In the context of multistage integration architectures as shown in Fig. 14.9, we next explain how user handover can be realized in the different stages.

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

315

OR R

An integrated EPON and WiMAX system may contain many WiMAX micro-cells. In the example as shown in Fig. 14.9, there are up to 1792 micro-cells associated with a single integrated system. Moreover, within each micro-cell, a large number of WiMAX subscribed users may be served. Thus, for an integrated system, a fatal failure, like a failure at a central OLT or a cut of the fiber link between an OLT and an optical splitter, can disrupt the services of all the users in the system. Survivability is therefore important to the integrated EPON and WiMAX systems. For the integrated architectures, we mainly consider the network failures due to fiber cut (because of its relatively high occurring probability). As shown in Fig. 14.10, there are two types of fiber cut. The first one is a fiber cut on the link between a central OLT and an optical splitter, and the second one is a fiber cut on the link between an optical splitter and an ONU-BS. In general, the first type of fiber cut is more serious, as it disconnects all the ONU-BSs from the central OLT, thereby disrupting the services of all the users. In contrast, the second type of fiber cut will only affect a small group of users that are associated with an ONU-BS. Different protection strategies can be adopted to recover from the above two types of fiber cut. For the first type of fiber cut, it is cost-effective to deploy an extra backup fiber along with the first fiber as shown in Fig. 14.10, such that if the first fiber gets cut, the backup fiber can take over the responsibility to carry user data. This is a type of 1+1/1:1 failure protection. Because the backup fiber is shared by all the users subscribed to the integrated system, the extra cost for this second fiber is generally acceptable for each of the subscribed users. However, for the second type of fiber cut, the deployment of 1+1/1:1 protection scheme can be expensive as only a small group of users share an extra fiber segment from an optical splitter to an ONU-BS. Under the integrated architecture, a more attractive solution can be adopted, which employs point-to-point WiMAX radio links between neighboring

EC

TE

DP

RO
This figure will be printed in b/w

Fig. 14.10 Survivability of integrated access systems (adapted from [11], c 2007 SPIE)

UN C

OF

14.4.6 Survivability

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

316
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Gangxiang Shen and Rodney S. Tucker

14.4.7 Cooperative Transmission for Broadcast Services

This figure36 will 37 be printed 38 in b/w 39
40 41 42 43 44 45

35

Fig. 14.11 Downstream data transmission for broadcast services: without cooperative communication

UN C

OR R

Under the multi-cellular wireless communications, interference from neighboring wireless cells is always troublesome for a user attempting to receive high-quality signals. Figure 14.11 shows an example of conventional multi-cell interfering communication. Assume that there are multiple WiMAX base stations (BS) and each WiMAX BS covers a wireless micro-cell, in which multiple subscribed users are associated with the WiMAX BS and receiving data from it. Due to independent user communications within each WiMAX micro-cell, any wireless power leakage from one WiMAX cell to another is considered as interference for the latter, which is harmful to signal to interference and noise ratio (SINR) of the users within the cell. A user within region A overlaid by three neighboring micro-cells as shown in Fig. 14.11 can suffer from strong signal interference. Specifically, if the user is associated with WiMAX BS 1 and receives a signal with power P1 from base station 1, and meanwhile, it also receives interfering signals leaked from the other two neighboring WiMAX BSs (ONU-BS 2 and ONU-BS 3) with power P2 and P3, respectively, then the user has an SINR P1/(N+P2+P3), where N is the power of noise. The SINR is poor if P1, P2, and P3 are close. With such a poor SINR, only

EC

TE

DP

RO

WiMAX base stations to realize failure recovery for the second type of fiber cut. As shown in Fig. 14.10, when the fiber between an optical splitter and an ONU-BS gets cut, the ONU-BS can relay its user data to its neighboring WiMAX bases stations over the two point-to-point radio links. Then the neighboring WiMAX base stations further forward the user data to the central office. This type of failure protection technique is expected to be cost-effective for the second types of fiber cut.

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

317

OR R

a low-level modulation scheme such as BPSK can be employed for data transmission from ONU-BS 1 to the user. However, integrated EPON and WiMAX systems show good potential to overcome the above interfering situation by employing a transmission scheme called cooperative communications [28–30]. This transmission scheme is especially suitable to transmit broadcast services such as IPTV over the integrated systems. Based on the infrastructure as shown in Fig. 14.1, a single copy of user data can be forwarded from a central OLT down to all the connected ONU-BSs and then the latter duplicate the data copies to all the SSs within associated WiMAX micro-cells. Under broadcast services, each user wishes to receive the same copy of data, no matter the users are in the same micro-cell or in different micro-cells. Thus, in this case the signals leaked from neighboring WiMAX micro-cells should not be considered as interference as the signals carry the same information; rather, they are useful signals. By properly manipulating these signals (e.g., based on some space–time coding technique [31]), they can be used to enhance the SINR of the signal received by each user, as multiple WiMAX BSs are transmitting multiple copies of information from different directions to the user. This is a type of transmitter macro-diversity, which is helpful to improve SINR in wireless communications. To realize the above cooperative communication, space–time coding techniques are required [31]. Specifically, each WiMAX BS is allocated with a unique orthogonal code. The BS uses the code to encode user data received from the central OLT and broadcasts the encoded signal to SSs. An SS receives multiple copies of signals (or superposed signals) [32] from different WiMAX BSs. The SS then decodes all the orthogonal signals to construct a stronger signal, which shows a better SINR. With a good SINR, an advanced modulation scheme such as 16 QAM can be applied for a high data transmission rate. We use an example as shown in Fig. 14.12 to illustrate the cooperative transmission mechanism. Assume that we have a data stream to broadcast to all the SSs

EC

TE

DP

RO
This figure will be printed in b/w

Fig. 14.12 Downstream data transmission for broadcast services: with cooperative communication

UN C

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

318
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Gangxiang Shen and Rodney S. Tucker

Acknowledgments The authors would like to thank the Australia Research Council (ARC) for supporting this work.

References

1. IEEE 802.3ah Task Force: http://www.ieee802.org/3/efm/ [Online on April 4, 2008]. 2. G. Kramer, B. Mukherjee, and G. Pesavento, “IPACT: A Dynamic Protocol for an Ethernet PON (EPON),” IEEE Communications Magazine, vol. 40, no. 2, pp. 74–80, Feb. 2002. 3. ITU-T G.984.4, SG 15, “Gigabit-capable Passive Optical Networks (G-PON): Transmission Convergence Layer Specification,” July 2005. 4. IEEE 802.16-2004, “Air Interface for Fixed Broadband Wireless Access Systems,” October 2004.

UN C

OR R

We have considered integration of EPON and WiMAX. Three integration architectures have been described and elaborated on. These architectures include the simplest independent architectures, more advanced hybrid architectures, and more spectrum-efficient microwave-over-fiber (MoF) architectures. Based on these architectures, the issues ranging from optimal planning to efficient operation were discussed. These issues include optimal PON network deployment to interconnect disperse ONU-BSs, packet forwarding modes, bandwidth allocation and QoS support, handover operation for mobile users, survivability, and downstream cooperative transmission in the integrated architectures. Due to the novelty of integrated EPON and WiMAX broadband access networks, these architectures and issues are open for further investigation. We hope that this chapter has stimulated the readers interest in integrated EPON-WiMAX systems.

EC

TE

14.5 Summary

DP

within an integrated access network. The stream is first broadcast from the central OLT to all connected ONU-BSs. Then the ONU-BSs (e.g., ONU-BS 1, 2, and 3), of which each is allocated with a unique orthogonal code (e.g., orthogonal codes 1, 2, and 3, respectively), encode the received data stream into an orthogonal signal and broadcasts the signal to all the surrounding SSs. In particular, for an SS user located within the region overlaid by the three neighboring WiMAX micro-cells, it receives superposed orthogonal signals from the three neighboring WiMAX BSs. The SINR of the user in this case is (P1+P2+P3)/N, which is significantly enhanced over the previous SINR P1/(N+P2+P3) which is under the conventional interfering communication mode. Based on this example, we can see that the integrated architectures are talented to realize cooperative communication, which can help to eliminate the interferences suffered by the conventional multi-cellular wireless communication systems and provide a good SINR for a high transmission data rate.

RO

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

14 Integration of EPON and WiMAX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

319

5. WiMAX Forum, “Mobile WiMAX – Part I: A Technical Overview and Performance Evaluation,” Aug. 2006. 6. IEEE 802.16e/D12, “Air Interface for Fixed and Mobile Broadband Wireless Access Systems,” Feb. 2005. 7. IEEE 802.16 Working Group: http://www.ieee802.org/16/ [Online on April 4, 2008]. 8. H. Ekstrom et al., “Technical Solutions for the 3G Long-term Evolution,” IEEE Communications Magazine, vol. 44, no. 3, pp. 38–45, March 2006. 9. M. Vrdoljak, S. I. Vrdoljak, and G. Skugor, “Fixed-Mobile Convergence Strategy: Technologies and Market Opportunities,” IEEE Communications Magazine, vol. 38, no. 2, pp. 116–121, Feb. 2000. 10. G. Shen, R. S. Tucker, and T. Chae, “Fixed Mobile Convergence (FMC) Architectures for Broadband Access: Integration of EPON and WiMAX,” IEEE Communications Magazine, vol. 45, no. 8, pp. 44–50, Aug. 2007. 11. G. Shen and R. S. Tucker, “Fixed Mobile Convergence (FMC) Architectures for Broadband Access: Integration of EPON and WiMAX (invited),” in Proc., SPIE Network Architectures, Management, and Application V, APOC, Wuhan, China, vol. 6784, pp. 678403-1-678403-13, Nov. 2007. 12. Y. Lou et al., “Integrating Optical and Wireless Services in the Access Network,” in Proc., OFC, paper NThG1, Anaheim, CA, March 2006. 13. Y. Lou et al., “QoS-aware Scheduling over Hybrid Optical Wireless Networks,” in Proc., OFC, paper NThB1, Anaheim, CA, March 2007. 14. S. Sarkar, S. Dixit, and B. Mukherjee, “Hybrid Wireless-Optical Broadband-Access Network (WOBAN): A Review of Relevant Challenges,” IEEE Journal of Lightwave Technology, vol. 25, no. 11, pp. 3329–3340, Nov. 2007. 15. W. T. Shaw, S. W. Wong, N. Cheng, K. Balasubramanian, X. Zhu, M. Maier, and L. G. Kazovsky, “Hybrid Architecture and Integrated Routing in a Scalable Optical-Wireless Access Network,” IEEE Journal of Lightwave Technology, vol. 25, no. 11, pp. 3443–3451, Nov. 2007. 16. A. Nirmalathas, D. Novak, C. Lim, and R. Waterhouse, “Wavelength Reuse in the WDM Optical Interface of a Millimetre-wave Fibre-wireless Antenna Base Station,” IEEE Transactions on Microwave Theory, vol. 49, no. 10, pp. 2006–2012, Oct. 2001. 17. M. P. McGarry, M. Reisslein, and M. Maier, “WDM Ethernet Passive Optical Networks,” IEEE Communications Magazine, vol. 44, no. 2, pp. 15–22, Feb. 2006. 18. C. H. Doan et al., “Design Considerations for 60 GHz CMOS Radio,” IEEE Communications Magazine, vol. 42, no. 12, pp. 132–140, Dec. 2004. 19. 10G EPON IEEE Working Group: http://www.ieee802.org/3/av/index.html/ [Online on September 18, 2007]. 20. J. Li and G. Shen, “Cost Minimization Planning for Passive Optical Networks,” in Proc., OFC/NFOEC, paper NThD1, San Diego, CA, March 2008. 21. J. Li and G. Shen, “Cost Minimization Planning for Passive Optical Networks,” submitted to IEEE Journal on Selected Areas in Communications, March 2008. 22. IEEE 802.1D, “Media Access Control (MAC) Bridges,” June 2004. 23. IEEE 802.1Q, “Virtual Bridged Local Area Networks,” May 2006. 24. C. M. Assi, Y. Ye, S. Dixit, and M. A. Ali, “Dynamic Bandwidth Allocation for Quality-ofService over Ethernet PONs,” IEEE Journal on Selected Areas in Communications, vol. 21, no. 9, pp. 1467–1477, Nov. 2003. 25. G. Nair et al., “IEEE 802.16 Medium Access Control and Service Provisioning,” Intel Technology Journal, vol. 8, no. 3, pp. 213–228, Aug. 2004. 26. S. Blake et al., “An Architecture of Differentiated Services,” IETF RFC2475, Dec. 1998. 27. R. Braden et al., “Integrated Services in the Internet Architecture: An Overview,” IETF RFC1633, June 1994.

UN C

OR R

EC

TE

DP

RO
AQ1

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

320
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Gangxiang Shen and Rodney S. Tucker

UN C

OR R

EC

TE

DP

RO

28. H. Zhang and H. Dai, “Cochannel Interference Mitigation and Cooperative Processing in Downlink Multicell Multiuser MIMO Networks,” EURASIP Journal on Wireless Communications and Networking, vol. 2004, no. 2, pp. 222–235, Dec. 2004. 29. I. D. Garcia, K. Sakaguchi, and K. Araki, “Cell Planning for Cooperative Transmission,” in Proc., IEEE WCNC, Las Vegas, March/April 2008. 30. P. Ho. B. Lin, J. Tapolcai, and G. Shen, “Cooperative Service Provisioning in Integrated EPON-WiMAX Networks,” submitted to IEEE Communications Magazine, March 2008. 31. V. Tarokh, H. Jafarhani, and A. R. Calderbank, “Space-time Block Codes from Orthogonal Designs,” IEEE Transactions on Information Theory, vol. 45, no. 5, pp. 1456–1467, July 1999. 32. IEEE 802.16 Task Group m (TGm): http://www.ieee802.org/16/tgm/ [Online on April 4, 2008].

OF

SPB-158632

14

March 24, 2009

Time: 10:57

Proof 1

01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Chapter-14
Query No. AQ1 Page No. 319 Line No. 34 Query

Please provide the volume and page range for Ref. [21].

UN C

OR R

EC

TE

DP

RO

OF