ECOC Technical Digest © 2011 OSA

Technical options for NGPON2 beyond 10G PON P. Chanclou, F. Bourgart, B. Landousies, S. Gosselin, B. Charbonnier, N. Genay, A. Pizzinat, F. Saliou, B. Le Guyader, B. Capelle, Q. T. Le, F. Raharimanitra, A. Gharba, L. Anet Neto, J. Guillory, Q. Deniel, S. Deniel Orange Lab – France Telecom R&D, 2 avenue. Pierre Marzin, 22307 Lannion, France, philippe.chanclou@orange–ftgroup.com

Abstract: This paper presents a view of the trends and perspectives in the next generation of optical access solutions following the 10Gbit/s Passive Optical Networks. OCIS codes: (060.2330) Fiber optics communications, (060.4250) Networks

1. Introduction Gigabit Passive Optical Networks (PONs) are currently deployed by many operators in the world and are currently one of the fastest available access technologies which highlights its key role in delivering high bandwidth services to any type of users. Current PON solutions are standardized by ITU-T G.984 under the name of "G-PON", and by IEEE 802.3ah under the name of "GE-PON". The Gigabit solution has now a follow-up in standards, a new PON generation driven by the enhancement of the delivered bandwidth. This next coming wave [1] of PON systems is based on a 10 Gbit/s transmission technology known as XG-PON (ITU G.987) and 10GE-PON (IEEE p802.3). These systems still operate under the same technical basis as the previous PON generation, since they use Time Division Multiplexing (TDM) / Time Division Multiple Access (TDMA) to manage the connectivity of "N" terminations to one optical port of the central office equipment. Essentially, these 10Gbit/s PON solutions multiply by four the line bit rate and use a wavelength plan designed to allow a smooth migration through an overlay with current Gigabit PON. The major benefits provided by this migration are: Co-existence between the Gigabit PON and 10G PON on the same Optical Distribution Network (ODN), Service interruption for subscribers who are not migrated to the 10G PON should be minimized, The 10G PON must support/emulate all legacy services of Gigabit PON in case of full migration. The XG-PON standardization is almost ready and some initial field trials have been already done in 2010 to confirm the feasibility of XG-PON's coexistence with commercially deployed G-PON systems. We are now starting the discussions of what would be a new era of access networks, known as the “Next Generation PON2”. This paper presents some investigations on what new optical access solutions should be preferably based on. 2. What NG-PON2 could be…? The goal of an operator's access network has always been to propose optical access solutions to support broadband services such as the digital home and mobile or DSL backhauls. This is still valid for all possible NGPON2 technical candidates. In addition, we might have to deal with the fact that NG-PON2 could imply the end of DSL access and consequently the start of copper de-commissioning operations in ducts and central offices. Furthermore, fiber connectivity will address the whole telecommunication set and not only standard triple play services (Internet, TVoIP, VoIP). The continuity of G.703's physical interfaces such as E1 and T1 as well as wide area networks based on protocols such as ISDN, X.25, Frame Relay and ATM ought to be guaranteed. This issue could be solved by an appropriated encapsulation which, however, will challenge new systems to have versatile and cost-effective Optical Network Terminal adapted to "low" bandwidth services. In essence, NG-PON2 solutions must be able to satisfy the needs of telecom industry players by providing a new broadband market not only capable of generating financial saving but also prepared to a migration in which we should probably need some integration with former deployed generations such as the G-PON and the XG-PON. In addition, to bring a significant benefit compared to "older" technologies, NG-PON2 shall become a "musthave" solution by delivering a bundle of services over 1Gbit/s per user. The line rate or aggregate bit rate for such a solution will depend on the multiplexing choices. Current optical access solutions are based on a PON architecture that uses time resource allocation, an ODN based on passive and achromatic optical splitters and a wide wavelength multiplexing (diplex) for up stream and downstream transmission. Such properties arose from the necessity of deploying fiber cables inside ducts where the space is limited due to the coexistence with copper cables. NG-PON2’ basic requirements could be based on a similar passive optical infrastructure with 64 or more terminations sharing over 40 Gbit/s of aggregate capacity at the central office equipment interface. Apart from the appropriated migration and bit rate increase, the new generation is also conditioned by a “central office consolidation". An extended optical budget remains an essential option in NG-PON2 to allow the increase of network’s reach and thus its eligibility area and to reduce the number of central offices based on G-PON performances. The large variety of potential network "users" (residential, business, SME, backhaul) over a

We.9.C.3.pdf 1

7/27/2011 4:38:39 PM

ECOC Technical Digest © 2011 OSA

consolidated network should require solutions capable of assuring a higher service resiliency. An enhanced security plane and high service availability should also be pointed as necessary needs of a cost-effective resilient solution. These options will allow each operator to build the service architecture that best suits their specific market and geography. The NG-PON2 preliminary work is considering a panel of high speed technologies and is open for major changes, meaning that the TDM/TDMA PON standard with two-wavelengths across the optical splitters could be reconsidered. … and what it could not be? Firstly, NG-PON2 could not be a solution that forces operators to install new ducts or to change the existing fiber cables and passive devices. It must be preferably deployed in pre-existing ducts and should minimize operational interruptions of former access services. Nevertheless, it could not be constrained by the needs of such co-existing PON ODNs. Secondly, NG-PON2 could not be a wavelength stacking of un-connected XG-PONs. If wavelengths are used as a resource in NG-PON2, the Optical Line Terminal should be capable of guaranteeing services to any customer with an agile and unstressed wavelength allocation. Thirdly, when it comes to questions such as spectrum availability or optical budget, NG-PON2 could not be defined without taking previous generations into consideration. Different scenarios should be considered by operators when implementing NG-PON2. For the deployment in "optical" green fields (areas where G-PON and XG-PON are not deployed) since such areas are likely to be thinly populated, the choices of the fiber plant architecture are still open. In other cases, we should expect multiplying XG-PON’s bit rate by four. This scenario will be more restrictive in terms of service disruption if coexisting with G-PON and XG-PON in the same ODN. Last but not least, the convergence of fixed and mobile access networks is often presented as a possible driver of NG-PON2. Under the term convergence, not only should we consider the mobile backhaul, which is already supported by G-PON (IP traffic), but also the notion of remote antennas based on Digital-Radio of Fiber [2] (Analog-RoF is most an option for remote indoor antennas). It should be remembered that current mobile networks use optical point to point fiber infrastructure given that the timeline and location of fiber is not necessarily identical to FTTH, and point to point or daisy chain topologies are more current. Thanks to all this considerations, the fiber infrastructure convergence based on A-RoF could not be a driver of NG-PON2; D-RoF should be considered as framed traffic inside the NG-PON2 connectivities. To sum up, in order to define NG-PON2 for a mass market, new optical access solutions should be flexible in terms of architecture. As a consequence, newly and already deployed ODN could be unlighted by new generation equipments. 3. Possible guidelines Some NG-PON2 guidelines follow the same premises that leaded the requirements for the XG-PON. Power saving, QoS and traffic managements and synchronization for the mobile backhaul remain prerequisites for the NGPON2. In this section we assess three aspects of such guidelines: generic optical access interfaces, wavelengthagility and Frequency Division Multiplexing/Multiple Access (FDM(A)). a. Generic optical access interfaces: Generic interfaces are a prerequisite to help interoperability. In addition, they must be power efficient, simple and compatible with high integration scales. A generic interface would be necessary mainly due to: The multiplication of pluggable interfaces in the telecommunication and data market. The ability of pluggable interfaces to support more and more advanced functions on board. The necessity to be open for different optical front ends while being generic in terms of Medium Access Control (MAC). NG-PON2 could be based on a meta-MAC on the OLT board plus an on-board MAC inside a pluggable device which would be designed for different connections, physical layers, optical budgets and wavelengths. The MetaMAC would be capable of fulfilling all network functionalities such as per user resource allocation, dynamic bandwidth allocation, congestion policy, QoS policies, OAM and OMCI within one chipset endowed with powerful silicon resources. The input-output signal of the Meta-MAC chipset is a signal based on generic framing and capable of connecting the generic OLT host to the pluggable transceiver. In this transceiver, we will find an on-board MAC which transposes the generic meta-MAC signals into the dedicated physical connectivity for downstream and upstream. This MAC chipset is a slave device concerning resource allocation, QoS and congestion policies and etc. It is transparent in terms of data flow processing but could be associated to physical layer operations and to administration and maintenance proposes. The last element is the physical layer interface which is one of the possible multi-connectivities. For example, the pluggable transceiver could be designed with a slave on-board MAC

We.9.C.3.pdf 2

7/27/2011 4:38:39 PM

ECOC Technical Digest © 2011 OSA

to execute a signal translation to standard TDM(A)PON or maybe several point-to-points or wavelengths. Next section will present an example of what could be done based on a front end with several wavelengths. A generic optical access interface leaves open the choice of the optical connectivity and standardizes instead the protocol exchanges (electrical, mechanical and management) to the OLT where the traffic engineering is done. b. Wavelength-agile PON : A wavelength-agile PON can be an option for TDM(A) PONs since a second sharing resource based on wavelength broadcast & select is added to the network. The wavelength could be considered as a static (quasi-static) dimension (circuit mode) while the time is considered as the dynamic dimension (packet mode). One possible solution is to manage the wavelength resource at the border of an untouched ODN (cf. Figure1). Typically, Fig.1 Wavelength-Agile PON several wavelengths are broadcasted by the OLT through a splitter based ODN and selected by the ONT receivers, in the downstream. In the upstream, each ONT or ONT cluster emits a specific wavelength that will be received at the OLT by a specific receiver. The wavelength agility could be done on a quasi-static or dynamic mode. This solution could use tunable (or selectable) wavelength emitters [3] and receivers [4] at ONT. A possible target is to achieve several 10Gbit/s wavelength links using low-cost wavelength agile devices. Evolution of photonic integrated circuits and existing potentially low-cost technologies for tunable receivers and emitters stimulate the evolution towards wavelength-agile TDM(A) PON. Other candidate solutions also exist in the same context by combining wavelength agility to code or frequency allocation rather than time. The next section will focus on frequency allocation to increase ONT’s versatility. c. FDM(A) : The evolution of Digital/Analogue Converters (DAC/ADC) sampling rates combined with the advances of Digital Signal Processing (DSP), are stimulating investigations on new multiplexing techniques for the optical access network, one of the most interesting of which is FDMA. In this technique, multiple electrical Radio Frequency (RF) carriers are transported together over one optical carrier. For the downstream, one or more carriers are dedicated to a customer. The allocation of carriers per subscriber can be static or slowly-dynamic (quasi-static) as a function of time. For the upstream, each subscriber emits its traffic over one radio-frequency carrier which has been dynamically allocated. This solution allows to solve the main drawback of pure TDM(A) PON in which customers need a MAC and PHY layers working at the same bit rate of the line eventhough they actually use only a part of this bit rate at the UNI. The FDM(A) ONT processes only its own allocated physical channel (radio frequency), which is a fraction of the global bit rate. This solution could also help to design versatile ONT in terms of cost and potential bit rate. Nevertheless, it requires a selective RF band designator after the optical receiver (tuneable radio frequency local oscillator). In the upstream, in order to avoid optical carrier beating noise, the wavelengths must be strictly identical or sufficiently different (with a separation equal or greater than the total RF bandwidth of all customers). Such requirement involves an emitter based on either a reflective scheme or sufficiently separated wavelengths at the emitter. The emitter could also need a tuneable RF band front end. FDM(A) adopts an approach based on the utilization of a large number of sub-carriers which are individually modulated using On-Off Keying or M-ary Quadrature Amplitude Modulation (M-QAM). The use of Orthogonal Frequency Division Multiplexing (OFDM) in each allocated RF sub-band can also be envisaged. 4. Conclusion The technological and architectural choices for the NG-PON2 will be driven by the state of the FTTH ecosystem ate the time the decisions are made. If NG-PON2 is proposed for a timescale or area when/where previous PON solutions are considerably deployed, technical choices could possibly be driven by the coexistence with splitterbased ODNs. Versatility should be a good term to define the quest of NG-PON2 as far as issues like cost, bandwidth, sporadic request of important traffic, wavelength allocation and chip set capabilities are concerned. The definition of an OLT optical backplane capable of aggregating the traffic of consolidated access networks (>10 000 users) with 1Gbit/s per user should also be taken into account. After all, would it be green and cost-effective to have around 10 Tbit/s of electrical switching capacity in the next generation OLT backplanes? References [1] J.-i, Kani, et al., "Next-Generation PON Part I – Technology Roadmap and General Requirements", IEEE Commun. Mag., vol. 47, n°11, Nov. 2009, pp. 43-49 [2] G. Kardaras et al. "Analysis of Control and Management Plane for Hybrid Fiber Radio Architectures", ICCT 2010, pp.281-284, 2010 [3] S.H. Oh et al., "2.5Gbps WDM-PON Tunable Light Source Hybrid Integrated with Superluminescent Diode and Polymeric Waveguide Bragg Reflector", ECOC 2009, Paper P2.11, 2009. [4] R. Murano et al. "Tunable GPON receivers enable phased migration to 1Gb/s per subscriber", OFC-NFOEC 2009, paper NME4, 2009. [5] B. Charbonnier et al., "(O)FDMA PON over a legacy 30dB ODN", OFC 2011, paper OTuK1, 2011

We.9.C.3.pdf 3

7/27/2011 4:38:39 PM