Mini-Fiber Node Technology


Published on

First publication of the mFN to Cable TV Industry: 1998 Emerging Technology Conference and 1999 CED Magazine.

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Mini-Fiber Node Technology

  1. 1. Low-Cost Cable Network Upgrade for Two-Way Broadband Xiaolin Lu, Ted Darcie, Alan Gnauck, Sheryl Woodward, Bhavesh Desai, and Xiaoxin Qiu AT&T Labs-Research Newman Spring Lab 100 Schulz Dr. Red Bank, NJ 07701 (732)345-3217 xl@research.att.comIntroduction One of the most challenging issues facing the cable industry today is the need tocost-effectively transform a traditional one-way cable system, which was designed forbroadcast services with tree-and-branch architecture, into a two-way broadband digitalplatform. Without clairvoyance to foresee specific future service requirements, we shouldwant the platform to be flexible enough to support different service needs withoutincurring operation complexities and cost. Therefore, operators should carefully balanceinfrastructure upgrade strategy, service uncertainty, financial constraints and theinevitable evolution of technology. The industry has been feverishly following the traditional upgrade strategy inwhich 350MHz or 550MHz coax amplifiers are upgraded to 750MHz, and two-waycapability is enabled using low-frequency (5-40 MHz) upstream technology. A regrettableconsequence of this traditional technology is that the upstream channel performance islimited. This may be adequate for existing applications like web browsing, but is likely tobecome problematic with emerging service requirements. Also, ingress noise necessitatesperforming complicated signal processing and spectrum management, which thentranslates into higher terminal cost. Further, expanding the system bandwidth to higherfrequency requires more amplifiers to overcome the increased coaxial loss. Noise andreliability concerns, on the other hand, demand the use of fewer amplifiers in cascade.Resolution of this conflict requires network re-branching and re-engineering. This leadsto not only more amplifiers in the field but also a more complicated coax plant. Thesedifficulties then translate into higher cost and operation complexities, and raise seriousquestions about the adequacy, quality and reliability of the resulting transport capability.Operators hope that shrinking the fiber node (FN) serving area would create more 1
  2. 2. bandwidth per user and reduce ingress noise accumulation. However, these fundamentalproblems remain and the required costly network re-engineering is a bitter pill that is hardto swallow. It is a gamble, not a given, that a traditional upgrade approach would providean operationally and economically viable platform for next-generation services. Without alternatives, operators have little choice but to take this gamble, or watchwhile new opportunities (high-speed data, digital video, telephony, etc.) pass them by.However, the current upgrade path is based on traditional network technology which hasevolved from the notion of providing broadcast analog TV services. Further, instead ofresolving the infrastructure limitations, it relies on complex signal processing to attemptto make the resulting poor channels serviceable. In this paper, we will describe an alternative approach that aims at fixing theinfrastructure first. Using emerging lightwave technology, we overlay the existingnetwork with an economically-viable fiber-to-the-bridger architecture. This providesabundant ingress-free bandwidth and radically simplifies service provisioning. Theresulting architecture enables simple and standard-compatible MAC protocols, easyoperation schemes, low-cost terminals and a wide spectrum of new service opportunitieswhile minimizing the cost of network re-design and re-engineering. All thosetechnologies have been prototyped, evaluated and demonstrated over an HFC test bed atAT&T Laboratories.Basic Idea: Mini-Fiber Node (mFN) The mFN-based cable upgrade strategy is shown in Fig.1. Independent of existingsystems, the mFNs couple directly into the passive coax legs (with drop taps) after eachdistribution coax amplifier (i.e. line extender). Each mFN contains a low-cost laser diodeand a low-cost PIN diode, and is connected to the head-end with separate fiber. Based on this strategy, the mFNs subdivide the FN serving areas into small cells(typically 50 household-passed/mFN) and exploit the clean and large bandwidth at highfrequency for both upstream and downstream transmission. The mFN therefore creates anew path for digital services without affecting analog TV services carried by conventionalFN/amplified-coax paths. All services are then merged over passive coax distributionlegs. Indeed this approach brings fiber to each coax distribution amplifier. At this pointreaders may recall economic studies that illustrate clearly that shrinking node sizes to 100or fewer household-passed (HHP) is uneconomical. However, these studies do not applyto the mFN architecture, which separates analog video services from those delivered overthis new “digital only” path. Hence the expensive high-quality linear lasers used intraditional systems are replaced by low-cost lasers, and the mFN can be extremely simplewhile serving a small group of users. The mFN is purely an O/E converter without anyA/D, modulation/demodulation or mux/demux functions. Our analysis and prototyping 2
  3. 3. experience show that the cost of each mFN is not a concern. The cost of fiber dominates,but is still under control as discussed later.Advantages: By exploiting the clean and large bandwidth, this strategy avoids the complexitiesof traditional RF techniques (e.g., frequency agility) and related signal processing. Thistherefore simplifies system operation and reduces terminal cost. It also increases theoverall system bandwidth beyond the current coax amplifier limitations, for new digitalservices, without replacing coax amplifiers and changing amplifier spacing. For example,a system with 750 MHz bandwidth and 1 GHz passives (taps) cannot be operated above750 MHz with the traditional approach. The mFN can use the 750 MHz to 1 GHzspectrum directly, independently of the coax amplifier capabilities. Using this approach, we separate the upgrade for two-way digital services fromthe improvement of analog broadcasting service. We can then establish a simplebroadband digital platform without affecting existing analog services. No network re-engineering is needed and the additional fiber can be installed along the existing strands.This strategy can be easily incorporated into an on-going “analog-service-enhancement”upgrade, thus minimizing incremental cost. On the other hand, if the analog service workssatisfactorily, the mFN technology could be transparently implemented over embeddedsystems for new services, therefore eliminating the need of system re-design and re-engineering. Also, by bypassing the coax amplifier cascade, the new mFN/passive-coaxarchitecture has substantially higher reliability and smaller failure group size (50 HHPcompared with 500-2000 HHP) than conventional cascade-coax architecture. Thepositioning of mFNs in the network further enables easy network monitoring. Because the mFN is transparent to two-way traffic, the system capacity can beflexibly provisioned to meet both today’s service needs and the need for future growth.No further infrastructure investment is needed.Challenges: The major challenge is, of course, bringing lightwave components and fiberdeeper into the network, at a reasonable cost. Unlike the conventional FN-splittingapproach, mFNs only carry digital subcarrier signals over a clean high-frequency band.Therefore, low-cost, low power consumption and space-saving optical and RFcomponents can be used in the mFNs and also at the head-end. Fig.2 illustrates the capability of a $100 uncooled Fabry-Perot laser used in themFN. This device requires no thermo-electric cooler and no optical isolator to maintainacceptable performance, in strong contrast to the high-performance lasers used to deliveranalog video. Our prototype mFN, including power converter, laser, PIN diode androughly 50 dB of amplifier gain in each direction, fits into a typical tap box (Fig.3). 3
  4. 4. Given the cost of the laser, minimal power consumption (<1W/mFN) and functionalsimplicity, the projected cost of each mFN is small enough not to be relevant. The dominant cost of this mFN-HFC architecture is that of the fiber deployment.Operators have enough experience deploying fiber to 500-2000 HHP nodes to be able topredict reasonable costs for a mFN system deployment. Our estimates over several cablesystems show that, depending on system specifics like demographics and topography, acomplete mFN deployment, including fiber deployment, costs $100-200/HHP. Atraditional upgrade, based on operators’ experience, costs $200-400/HHP. This of coursedepends strongly on the amount of coax re-engineering required to complete thetraditional upgrade. In some cases, a mini-coax node, which is a diplexer-based jumper toshunt high frequency signals around a coax amplifier, could also be used to reduce theamount of mFNs and therefore fiber cost. Also, if mFNs are clustered to some local hubs,as favored by many cable operators, the cost of mFN strategy could be substantiallylower. Yet the performance and capacity of the resultant mFN infrastructure could be farsuperior to that of the traditional system. Especially when we are moving into thebroadband arena, the capacity cost (dollars per MHz) is probably a bigger concern tooperators (Table 1). Most important, the mFN technology provides a one-stop upgradefor two-way broadband and creates more value and opportunities that cannot be providedotherwise.Implications: New service opportunities If one breaks from tradition and fixes the infrastructure to establish a noise-freebroadband digital platform rather than using complex signal processing over a capacity-limited network, then numerous opportunities arise.Data services: The unique position of each mFN enables a considerable simplification indefining medium access control (MAC) protocols. Each mFN can do local policing, andresolve upstream contention within its serving area without involving other parts of thenetworks (Fig.4). This can be accomplished by incorporating a simple out-of-bandsignaling loopback scheme such that users know the upstream channel status prior totransmission. This enables the use of standard, but full-duplex, Ethernet protocol(CSMA/CD), and therefore the use of standard and low-cost terminals (modified Ethernettransceiver, Ethernet bridger and Ethernet card). No ranging is needed, and the head-endbecomes virtually operation-free. The relative small round-trip delay between each userand the mFN (~2000ft) also substantially increases bandwidth efficiency and reducescontention delay (Fig.5). This is appealing for VBR (variable bit rate) type of services.For CBR (constant bit rate) services, certain scheduling or priority provisioning may benecessary, which can be added easily to the above protocol. The local access control in the clustered small mFN serving area further simplifiessystem operation, enables QoS and multi-tier services capability, and makes overall costindependent of usage rate. In contrast, the large round trip delay of traditional cablesystems makes this an unfavorable or unacceptable protocol, forcing the industry tostandardize complex head-end mediated protocols. Such protocols rely on central- 4
  5. 5. controlled broadcasting schemes over large serving areas with limited channel capacity.They work well in lightly loaded systems, but fall short and become expensive if capacitydemand increases. The large and clean bandwidth supported by the mFN infrastructure also enablesthe use of efficient but much simpler modulation schemes such as multi-level FSK oreven ASK. This therefore provides a low-cost alternative to the custom RF techniquesincorporated into current cable modems.Telephony services: The mFN local access control protocol with QoS capability caneasily support packet voice and other synchronous and mixed synchronous/asynchronousservices. Alternatively, the large bandwidth and small serving group (50 HHP/mFN)provisioned by each mFN allow dedicating one or more RF channels to each user forsynchronous telephony services. Whereas the traditional approach uses time-sharedchannels to increase spectral efficiency, the mFN system can use a less efficient butsimpler approach, like FDM/FDMA (Fig.4). Freedom from ingress noise eliminates the need to have complex ingress-avoidance features built into the call-control software. One can then consider leveragingcordless phone technology, with simple modifications suited to the mFN architecture.Normally, cordless phones convert the voice signal into an RF signal (digital or analog)for transmission over the air between the base and the handset. We have demonstratedthat we can instead transmit the RF channels directly over the mFN system between thebase, located at the head end, and the handset at home. By reconfiguring the parts of acordless phone handset to make it also interface with regular phone sets, we can utilizeoff-the-shelf technology for telephony services without incurring the cost andcomplexities of traditional NIUs. The low power consumption of cordless phonetechnology also simplifies powering scheme and makes home-powering more attractive.At the CO/HE, instead of using custom HDT, the standard but simplified Telco interface,such as the channel bank (but without line cards and battery), could then be used.Digital TV: Independent of analog TV services carried by the existing amplified-coaxpath, multichannel digital video can be directly broadcasted over the mFN path. Thiseliminates the analog-digital interference arising from clipping and nonlinearity of lasersand coax amplifiers, and also substantially increases the cable network channel capacitywithout network re-engineering. Because digital video signals do not have the stringentrequirement as the AM-VSB does, low-cost lasers (Fig.2) and electronics can be used.The small serving area provisioned by each mFN also enables easy narrowcastingscheme.PCS: By including a remote antenna at each mFN, the mFN-HFC system provides anadvantageous backhaul for wireless PCS application. The unique location of each mFNallows the use of various distributed antenna or simulcasting techniques. The directlightwave backhaul also allows PCS signals to be directly transmitted between the basestation at the CO/HE and the remote antenna at each mFN without competing for coax 5
  6. 6. bandwidth. This effectively extends the overall cable network capacity beyond coaxlimitations. The above technology has been prototyped and evaluated over an HFC test bed atAT&T Labs. We demonstrated simultaneously delivering 80 channels of MPEG-2 digitalvideo, 10Mbps bi-directional data (full-duplex Ethernet), and telephony service over themFN system, in addition to the existing analog video carried by the traditional HFC.Bottom Line HFC is no doubt the first economically viable means for broadband services.However, well-known network problems limit the value it can provide and traditionalsolutions create unresolved challenges and complexities. There is a risk, a gamble thatoperators must take, that the problems associated with the traditional cable upgrade mayrender the resulting systems incapable of supporting, or cost-effectively supporting, theservices that operators will demand in the near future. We have described an alternative approach, based on new technology, that offers ameans to a future-proof HFC architecture. We advocate aggressive fiber deployment, andshow how the resulting abundant ingress-free bandwidth and the advantageousarchitecture can reduce operation and terminal complexities and create moreopportunities that the traditional strategy could not provide (Table 2). Economic studiesproject that the cost of this new system is, in many cases, comparable to that of thetraditional approach. Yet the mFN infrastructure enables significant cost reduction onsystem operation and terminal equipment. Fiber has had a profound impact on theindustry since the advent of the linear laser, and will continue to drive the evolution to thetwo-way broadband era. 6
  7. 7. Table 1. Cable upgrade comparison mFN Upgrade Strategy Conventional Upgrade Strategy Architecture Clustered fiber-to-the-amplifier More tree-and-branch Bandwidth 250-450MHz/50 HHP noise-free two-way 35MHz/500-2000 HHP noisy upstream Capacity bandwidth 250-400MHz/500-2000 HHP downstream, limited by coax amplifier RF Allow simple and easy transmission Bandwidth efficiency and robustness to Transmission noise are necessary Operation Simple operation (no frequency agility) Complex operation (spectrum management) Standard access protocol Custom access protocol Easy network monitoring Reliability Small failure group (50 HHP) Large failure group (500-2000 HHP) Better reliability Limited reliability Upgrade More fiber usage Less fiber usage No network re-engineering Extensive network re-engineering One-stop upgrade, no need for further Architecture and cost barriers for future infrastructure investment growth Upgrade Cost: $100-200/HHP Upgrade Cost: $200-400/HHP Capacity cost: Capacity cost: $12-40/MHz two-way $1.4K-11.4K/MHz upstream $125-1,600/MHz downstreamTable 2. Service capability comparison mFN-HFC Conventional HFC Data Standard Protocol (IEEE802.3/Ethernet) Custom protocol (MCNS, IEEE802.14) Simple terminals, off-the-shelf components Complex terminals High-efficiency, low-delay and easy to scale Large overhead and delay, expensive to scale POTS Simple FDM/FDMA Complex TDM/TDMA Standard Telco interfaces Custom interfaces Digital Low-cost laser for broadcasting Need further network upgrade TV Automatic capacity enhancement May need to resolve analog-digital interference Easy narrowcasting PCS Remote antenna at mFN Competing for coax bandwidth Advantageous lightwave backhaul 7
  8. 8. CO/HE Analog video FN mFN mFN New Services New Services TV Analog video Modem 5 50 500 750 1G Fig. 1. Mini-Fiber Node (mFN) for cable upgrade -2 T heory Room T emperature T >80C -3 Reflec tion log(BER) -4 -5 -6 -7 -8 -9 -10 2 4 6 8 2 4 6 8 2 0.1 1 10 OMD(%) Fig. 2. Capacity of an uncooled FP laser. 60 channels QPSK at 2Mbps/ch were applied to the laser, and the BER was measured in the temperature ranging from 200C to 800C. 8
  9. 9. Fig. 3. Early mFN prototype. The upper part consists of power convertor and lower part contains all the photonic and RF components. This version uses two fiber and the latest version uses only single fiber for bi-directional transmission. 9
  10. 10. CO/HE mFN Analog TV FN Digital Video PCS Base Station mFN-HFC mFN Channel Bank5E Switch (w/o line cards) Local TI Signaling ST TV Ethernet Bridger TI Data NAD P Network DATA POTS PCS NAD Ethernet card C Digital Analog TV Video … 5 50 500 750 1G … 2G NAD: Network Access Device - modified Ethernet transceiver Fig. 4. mFN-HFC for multiple services TI: Telephony Interface - modified cordless phone handset or base ST: Set-top box 1000 100 Average delay (ms) 10 mFN MCNS 1 0.1 0.01 10 30 50 70 90 Number of active users Fig. 5. Upstream delay comparison between mFN based NAD and MCNS standard based cable modem. It was assumed that the average data rate is 120 kbps/user, with total speed of 10Mbps, and the packet size is 600 bits. In the case of MCNS modem, the “super frame” size is 24 Kb, in which 10% is allocated for contention. The assumptions for MCNS modem will change in real implementation. 10