HFC ARCHITECTURE IN THE MAKING
`
`Oleh Sniezko, Tony Werner, Doug Combs and Esteban Sandino
`AT&T Broadband & Internet Services
`
`Xiaolin Lu, Ted Darcie, Alan Gnauck, Sheryl Woodward, Bhavesh Desai and Xiaoxin Qiu
`AT&T Labs
`Rob Mclaughlin
`AT&T Broadband Services Engineering
`
`Abstract
`
`Many architectural realizations of the
`HFC network successfully support the services
`ranging from broadcast video (analog and
`digital), through pay per view and video on
`demand,
`to telephony and high-speed data.
`However, the HFC network and its architectural
`implementations continuously evolve to fulfill
`the
`increasing demand for reliability, high
`quality, and sufficient capacity to provide a
`broad array of services to an increasing number
`of customers.
`
`This paper presents an analysis of
`various architectural implementations of the
`HFC network. These implementations range
`from a
`traditional Fiber-to-the-Serving-Area
`approach,
`through fiber overlay for digital
`services, to fully-passive coaxial networks fed
`by dedicated fibers.
`
`INTRODUCTION
`
`In the last several years, cable network
`operators have embarked on extensive upgrades
`and on the transition to hybrid fiber coax (HFC)
`architectures. The main goal has been to evolve
`the infrastructure from a broadcast-type trunk(cid:173)
`and-branch plant to a high capacity, superior
`quality and reliability, and two-way network
`ready to deliver advanced telecommunications
`services.
`
`The need for this evolution is apparent
`the
`list of advanced services, which
`from
`demand increased network capacity and quality.
`Even for video services, the competition from
`
`alternative broadcast entertainment providers
`and transition from analog NTSC based video to
`digital (standard and high-definition) video
`require
`additional
`attention
`from
`cable
`operators.
`
`Architectures
`Requirements:
`• Superior reliability
`• Superior quality
`• Competitive price
`• Interactivity (two-
`way)
`Architectural changes:
`• Fiber supertrunking
`and backbone
`• Regional hub ring
`• Redundancy
`(secondary hub rings)
`• Deep fiber
`deployment &
`segmentation
`
`Services
`Traditional Services:
`• Broadcast TV
`• Broadcast Radio
`• Addressable Services:
`- Addressable tiering
`- PPV
`- Games
`- Digital radio
`• Home Shopping
`Advanced Services:
`• Targeted advertising
`• Targeted entertainment
`• Telephony
`• High speed data
`• Full multimedia
`Competition:
`• DBS
`• LEC
`• Multimedia mergers
`
`Table 1: Services, Requirements & Solutions
`
`m
`competitive
`cost
`Although
`comparison to the upgrades of other access
`networks to the same level of capacity and
`performance, upgrading from trunk-and-branch
`coax plant to HFC network requires significant
`financial effort from cable operators. Therefore,
`while satisfying the needs listed in Table 1, it is
`critical
`that
`the upgrades accomplish
`the
`following objectives:
`
`1. Establish a future-proof network
`2. Lower operating cost
`
`1999 NCT A Technical Papers -- page 20
`
`Cisco Systems, Inc., Exhibit 1035, Page 1
`
`

`
`Impact on Powering
`
`The analysis considered the impact of all
`architectures on powering options for the plant
`and for terminal equipment. The following
`issues were analyzed:
`
`network
`
`power
`
`1.
`
`overall
`
`on
`Impact
`consumption
`2. Terminal equipment powering alternatives
`3. Optimal network powering architecture
`(level of centralization)
`4. Power distribution in the existing plant and
`in plant extension or new-builds
`
`3. Significantly improve network reliability
`(M1TR: mean time to repair; and MTBF:
`mean time before failure)
`
`All these considerations lead to our
`continuous efforts in defining and re-defining
`architectural solutions for HFC networks to
`capture the ever-changing landscape of service
`demand and affordability (cost/benefit ratio) of
`new technological solutions.
`
`ARCHITECTURAL STUDY OBJECTIVES
`
`the scope and
`This section outlines
`objectives established and followed by our team
`for this advanced architecture study.
`
`Services Supported By the Network
`
`The network must be capable of
`supporting a wide variety of services:
`
`1. Analog video (basic, premium, PPV and
`IPPV
`2. Digital video (broadcast and IPPV)
`3.
`Interactive video (e.g., video on demand
`(VOD))
`4. High speed data access
`5. Telephony
`6. Telemetry
`
`Architectural Choices
`
`For practical
`concentrated
`on
`alternatives:
`
`reasons,
`several
`
`the analysis
`architecture
`
`1. Fiber-to-the-Serving-Area (FSA) with fiber
`node segmentation supporting forward and
`reverse communication services 12
`2. Trunk-and-branch coaxial architecture for
`traditional services with fiber overlay for
`advanced services3
`3. Deep fiber penetration to eliminate RF
`actives (Multiplexed Fiber Passive Coax(cid:173)
`MFPC)
`
`Acceptable Performance
`
`All the architecture alternatives were
`analyzed for the same levels of performance:
`
`1. Forward path end-of-line performance
`2. Reverse path performance
`3. Network availability
`
`Equipment
`
`The following issues were considered
`during the analysis:
`
`1. Technology feasibility
`2. Equipment availability
`3. Equipment price
`
`Fiber Count and Redundancy
`
`level of reliability
`the
`To provide
`required for the advanced services, the network
`must allow for cost-effective redundancy and
`low mean time to repair. To achieve these
`objectives, the following issues were analyzed:
`
`1. Number of fiber in single cable sheet
`2. Redundancy in different network sections
`3. Number of homes served per optical cable
`route without redundancy
`4. Restoration time
`
`1999 NCTA Technical Papers -- page 21
`
`Cisco Systems, Inc., Exhibit 1035, Page 2
`
`

`
`Network Characteristics
`
`While optimizing network topology to
`support different
`services with different
`characteristics,
`it
`is desirable
`to maintain
`transparency and flexibility in introducing new
`services. The
`team
`therefore defined and
`evaluated options and balances of:
`
`1. llitimate
`backward
`for
`topology
`compatibility and migration feasibility
`2. Network transparency between terminal unit
`locations
`3. Functionality of network terminals
`4. Centralized
`vs.
`distributed
`terminals
`
`network
`
`Cost Comparison
`
`the
`account
`into
`taking
`After
`the
`analysis
`above,
`listed
`considerations
`concentrated on cost comparison
`for
`the
`architecture alternatives. The initial capital
`costs were compared against benefits and life(cid:173)
`cycle savings. Some additional cost elements
`were also analyzed qualitatively (for example,
`terminal equipment cost).
`
`Additional Considerations
`
`The following additional items were
`considered:
`
`1. Terminal equipment standardization
`2. Time
`to market
`for
`the architectures
`(construction
`time
`and
`equipment
`development time)
`3. Suitability for MDUs and plant expansions
`
`This paper will provide a snap shot of
`this study with emphasis on the architecture
`alternatives,
`related
`features,
`and
`cost
`comparison. It will be seen throughout the paper
`that the differentiation among the alternatives
`are related to:
`
`1. Bandwidth per home passed
`2. Network power consumption
`
`3. Network availability (reliability and
`MTTR)
`4. Efforts of network alignment and
`certification
`5. Maintenance cost
`
`ARCHITECTURAL ALTERNATIVES
`
`Generic HFC Architecture
`
`Most of the existing HFC networks in
`markets exceeding 100,000 homes are based on
`ring configurations (single or dual). Traditional
`headends
`are
`consolidated,
`and
`are
`interconnected by the market rings (primary hub
`rings) with additional processing centers dubbed
`primary hubs (PH). The secondary hub rings
`are
`introduced
`to
`enable
`the
`headend
`consolidation. In this configuration, secondary
`hubs (SH) serve as signal concentration and
`distribution points to limit the number of fibers
`between primary hub ring and secondary hubs
`to achieve cost reduction, improve MTTR, and
`allow
`for cost-effective backup switching
`(redundancy). These two elements are common
`to most of the HFC networks used by cable
`operators.
`The differences are
`limited to
`.
`I h .
`. h
`.
`456
`technologtca c mces m t ese nngs.
`
`This generic architecture is depicted in
`Figure 1. The network sections marked as 'C'
`and 'D' (access network) provide the last links
`to the customers. The architectural solutions for
`this network section differ from operator to
`operator. However, most of them deploy fiber(cid:173)
`node-based configuration followed by active RF
`coaxial networks. The differences are reflected
`in the node sizes and in the level of redundancy.
`This access network is a subject of continuous
`architectural analysis by almost all major HFC
`network operators.
`
`1999 NCT A Technical Papers -- page 22
`
`Cisco Systems, Inc., Exhibit 1035, Page 3
`
`

`
`D
`
`50,000-
`300,000
`Homes Passed
`
`I
`lzo.~~~J
`I Homes
`!
`i
`
`Passed
`
`.
`
`~ Headend
`[fH]
`Primary Hub
`flW Secondary Hub
`0
`Node
`
`A
`B
`c
`D
`
`Primary Ring
`Secondary Ring
`Fiber Distribution
`Coax Network
`
`Figure 1: Modern HFC Network: Generic
`Configuration
`
`Access Network Alternatives
`
`Physically and logically, the last-mile
`access network can be categorized into two
`categories: Point-to-point (PTP) and point-to
`multi-point (PTMP). Examples of the PTP
`architecture include copper-pair based double
`star networks
`(traditional
`local
`telephone
`networks), certain FITC networks, and WDM(cid:173)
`based PON. Among the PTMP networks are
`such networks as HFC-based networks, certain
`TDMA-based PON, and wireless networks.
`Each network was historically defined
`to
`support certain type of services with their
`intrinsic characteristics. The challenge we are
`facing today is how to evolve the network
`architecture to support a wide variety of services
`with characteristics that the embedded system
`was not originally designed for.
`
`The HFC network was originally
`designed for broadcast services with PTMP
`architecture. Utilizing emerging
`lightwave
`technology with different
`levels of fiber
`deployment, our study concentrated on defining
`upgrade alternatives to support new service
`needs that may be optimized with certain degree
`of virtual PTP configuration. The alternatives
`were:
`
`1. Fiber to the Serving Area with fiber node
`segmentation
`supporting
`forward
`and
`reverse communication services
`2. Trunk-and-branch coaxial architecture for
`traditional services with fiber overlay for
`advanced services
`3. Deep fiber penetration to eliminate RF
`actives (Multiplexed Fiber Passive Coax -
`MFPC)
`
`FSA with FN Segmentation
`
`This network architecture (Fig. 2) has
`been used by many HFC network operators.
`The differences are mostly related to the node
`sizes, with particular emphasis on the design
`effort (optimization for power consumption,
`time-to-market, or end-of-line performance and
`bandwidth) and the level of redundancy (refer to
`Cox's ring-in-ring topology).
`In the analysis
`performed by the team, this architecture was
`used as a baseline 7 and was characterized by the
`following parameters:
`
`is between 600 and 1,200
`1. Node size
`household passed (HHP)
`2. Each FN can be segmented with up to four
`300 HHP buses
`3. Number of fibers from secondary hub to the
`node are between 4 and 6,
`4. Number of amplifiers
`between 5 and 8,
`5. Upstream is in 5-40 MHz, and downstream
`is in 50-750 MHz with 50-550 MHz being
`allocated for analog video and the rest of
`the bandwidth for digital services
`
`in cascade are
`
`Ill
`
`'Ill Segmentation
`DWDM Transport
`End-to-end Transparency
`4X capacity
`Figure 2: FSA with FN Segmentation
`
`•
`
`•
`
`1999 NCTA Technical Papers -- page 23
`
`Cisco Systems, Inc., Exhibit 1035, Page 4
`
`

`
`coax
`current
`beyond
`system bandwidth
`amplifier limitations, for new digital services,
`without replacing coax amplifiers and changing
`It
`amplifier
`spacing.
`also
`avoids
`the
`complexities of noise reduction (e.g., frequency
`agility) and related signal processing and RF
`techniques. This therefore simplifies system
`operation and reduces terminal cost.
`
`Also, because mFNs only carry digital
`subcarrier signals over a clean high-frequency
`band, low-cost, low power consumption and
`space-saving optical and RF components can be
`used in the mFN s and also at the headend.
`
`Mini Fiber Node for Fiber Overlay
`
`To resolve upstream limitations (ingress
`noise and bandwidth) and to simplify terminal
`operation, we proposed and evaluated the mini
`fiber
`node
`technology. Using
`emerging
`lightwave technology, the existing network is
`overlaid with a fiper-to-the-bridger architecture
`to exploit the large ingress-free bandwidth at
`high frequency for two-way digital services. As
`independent of existing
`shown
`in Fig. 3,
`systems, the mFNs couple directly into the
`passive coax legs (with drop taps) after each
`distribution coax amplifier (i.e. line extender).
`Each mFN contains a low-cost laser diode and a
`low-cost PIN diode, and is connected to the
`headend with separate fiber.
`
`Hubi:~--o--1
`
`Figure 3: Mini Fiber Node for digital overlay
`
`the mFNs
`this strategy,
`Based on
`subdivide the FN serving areas into small cells
`(typically 50 HHP/mFN) and exploit the clean
`and large bandwidth at high frequency for both
`upstream and downstream transmission. The
`mFN therefore creates a new path for digital
`services without affecting analog TV services
`canied by conventional FN/amplified-coax
`paths. All services are then merged over passive
`coax distribution legs.
`
`Features o(the mFN architecture:
`
`large
`the clean and
`By exploiting
`bandwidth,
`this
`strategy
`increases overall
`
`Figure 4: mFN Local Access Control Protocol
`
`The unique position of each mFN
`enables a considerable simplification in defining
`media access control (MAC) protocols. Each
`mFN can do
`local policing, and resolve
`upstream contention within its serving area
`without involving other parts of the networks
`can be
`accomplished by
`(Fig.4 ). This
`incorporating a simple out-of-band signaling
`loopback scheme such that users know the
`upstream channel status prior to transmission.
`This enables the use of standard, but full(cid:173)
`duplex, Ethernet protocol (CSMA/CD), and
`therefore the use of standard and low-cost
`terminals
`(modified Ethernet
`transceiver,
`Ethernet bridger and Ethernet card). No ranging
`is needed, and the headend becomes virtually
`
`1999 NCTA Technical Papers-- page 24
`
`Cisco Systems, Inc., Exhibit 1035, Page 5
`
`

`
`operation-free. The relatively small round-trip
`delay between each user and the mFN (-2000ft)
`also
`substantially
`increases
`bandwidth
`efficiency and reduces contention delay. This is
`appealing for VBR (variable bit rate) type of
`services. For CBR (constant bit rate) services,
`certain scheduling or priority provisioning may
`be necessary, and can be easily added to the
`above protocol (Fig. 5).
`
`6
`
`,..._....._. _
`
`_.___. Low Priority (20)
`
`...-11-t.....-.--.-.... Medimn Priority (10)
`
`High Priority (20)
`
`100
`200
`Request Packet Rate (Kbpslstation)
`
`300
`
`Figure 5: mFN-based priority provisioning
`protocol
`
`The large bandwidth supported by the
`mFN infrastructure also enables the use of
`efficient but much simpler modulation schemes
`such as multi-level FSK or even ASK. This
`therefore provides a
`low-cost,
`low-power(cid:173)
`consumption alternative to the current cable
`modem technique.
`
`Open Issues:
`
`The mFN technology explores a radical
`to
`resolve
`the network
`limitations.
`path
`Unfortunately, the existing system for broadcast
`video and DOCSIS-modem-based services is
`left behind with no benefits from the mFN
`strategy.
`
`Bringing fiber deeper into the network
`incurs certain incremental cost. Reducing the
`cost of fibering (material and
`labor)
`then
`
`becomes critical. The mFN strategy will
`simplify system operation for new services and
`reduce terminal cost. However, it will be more
`compelling if the front-end cost can be justified
`by operational savings over the entire network,
`both embedded and mFN overlay.
`
`Convergence: Multiplexed Fiber Passive
`Coax Networks
`
`To resolve those issues and to establish a
`platform that can improve the performance of
`the embedded system while also evolving to
`meet future needs and simplifying operations
`across the entire network, we proposed a new
`architecture called Multiplexed Fiber Passive
`Coax (MFPC). As shown in Fig. 6, instead of
`overlaying over existing coax amplifiers, mFNs
`eliminate all the coax amplifiers. (our design
`indicated that 2-3 coax amplifiers will be
`eliminated by one mFN). Between each mFN
`and customers, passive coax plant is used to
`carry both current and new services.
`
`Fibers connecting multiple mFNs will be
`terminated at the MuxNode that resides either at
`the original fiber node location or at location
`that "consolidate" multiple FNs. As its name
`implies,
`the MuxNode performs
`certain
`concentration and distribution
`functions.
`It
`"multiplexes" the upstream signals and sends
`them to the primary hub through the secondary
`hubs. It also "demultiplexes" the downstream
`signals received from the PH-SH fiber trunks
`and distributes them to mFNs.
`
`One of the interesting features of this
`architecture
`is
`that
`it maintains
`the
`characteristics of conventional HFC networks of
`being transparent to different signal formats and
`protocols, therefore fully supporting the existing
`operation for current services. To future-proof
`the network with more capacity and simple
`terminals, this architecture can also support a
`distributed-processing strategy for new, purely
`IP-based services enabled by the mFN and
`MuxNode, and maintain all the benefits of the
`
`1999 NCTA Technical Papers-- page 25
`
`Cisco Systems, Inc., Exhibit 1035, Page 6
`
`

`
`initial mFN strategy. The development can be
`partitioned into two phases.
`
`MuxNode. Other options, such as re-lasing at
`the MuxNode, are also promising.
`
`MuxNode
`
`mFN
`
`ITIJ.A:
`ITIJ.O:
`
`An81og ITU
`OlgllaiiTU
`
`RCV-A:
`ACV-D:
`
`Analog RCV
`Digit.! RCV
`
`Figure 6: Multiplexed Fiber Passive Coax
`(MFPC) architecture
`
`Phasel: Current Service Delivery
`
`One preferred approach is to transmit
`downstream broadcast signals, including analog
`and digital video, over dedicated fiber from the
`primary hub to the secondary hub. The existing
`narrowcast and switched signals, such as
`telephony and high-speed data using DOCSIS(cid:173)
`based cable modems and set top boxes will be
`DWDM multiplexed at the PH and transmitted
`over another fiber to the secondary hub. After
`DWDM demultiplexing,
`the narrowcast and
`switched signals will be optically combined
`with the broadcast signals (in a different RF
`band) and transmitted to the MuxNode over
`optical fiber.
`
`those combined
`the MuxNode,
`At
`downstream signals will be optically amplified
`and distributed to multiple mFNs over optical
`fiber. Given the short distance between the
`MuxNode and the mFN, it is also feasible to
`move the EDFA from the MuxNode to the
`secondary nub and only keep the optical splitter
`at the MuxNode,
`therefore simplifying the
`
`The mFN performs the same function as
`that of a typical FN. It receives downstream
`signals and distributes them to customers over
`passive coax cable. In the upstream direction,
`the mFN combines upstream signals from all the
`coax branches it serves and transmits them to
`the MuxNode over optical fiber.
`
`The MuxNode further performs 0/E
`conversion to those upstream signals from the
`mFNs, RF combines them, and transmits to the
`secondary hub using a wavelength specified
`laser. Upstream
`signals
`from multiple
`MuxNodes will be then DWDM multiplexed at
`the SH and transmitted to the PH. Besides
`DWDM multiplexing, another option is to re(cid:173)
`lase upstream signals at the SH.
`
`•
`
`To expand capacity of the system, one
`can frequency shift the upstream signals at each
`mFN such that when they are combined at the
`MuxNode they will be at separate RF bands. If
`the re-lasing option is used at the SH, the
`frequency stacking could also be deployed at the
`MuxNode.
`
`Phase 2: Add-on Distributed Processing
`Platform
`
`in Fig. 6, a distributed
`As shown
`processing platform can be transparently added
`when it is needed. The new IP-based services
`could be delivered to the MuxNode in baseband
`format over a DWDM link separate from the
`one carrying current services. As an alternative,
`the baseband signals and RF passband signals
`could be transmitted over the same DWDM
`link. At the MuxNode, the baseband signals are
`demultiplexed
`and
`further distributed
`to
`multiple mFNs.
`
`At each mFN, the downstream digital
`baseband signals are received and modulated
`onto RF carriers. They will then be combined
`
`1999 NCTA Technical Papers-- page 26
`
`Cisco Systems, Inc., Exhibit 1035, Page 7
`
`

`
`It is interesting to see that, with the help
`of the MuxNode and by eliminating all coax
`amplifiers, the cost of the passive coax design is
`comparable to that of initial mFN design. Of
`course, the value of this architecture is far
`greater than any other alternative.
`
`Figure 7: Front-end cost comparison
`
`Life Cycle Cost
`
`One of the biggest advantages of this
`architecture is the elimination of all the coax
`amplifiers. This results in substantial reduction
`in active components and therefore reduction in
`overall network power consumption (Fig. 8 and
`9). In addition, sweeping and maintenance of
`active components in the field will be reduced,
`which leads to further operation savings. It is
`estimated
`that
`annual operation
`savings,
`including reduction in customer calls. etc, could
`reach $11/HHP.
`
`with the received downstream RF signals and
`transmitted over the coax buses to customers.
`In the upstream direction, customers transmit
`the new IP-based upstream signals in a high
`frequency band (900 MHz - 1 GHz) to the mFN
`using simple FSK or QPSK modulation scheme.
`At the mFN, the MAC function is performed for
`those high frequency signals as discussed
`before. Those signals are also demodulated to
`baseband. They will then ASK modulate RF
`carriers at frequencies above the current 5-40
`MHz return band. These signals will be
`the 5-40 MHz band, and
`combined with
`transmitted to the MuxNode over a single fiber.
`Another option is to frequency shift the 5-40
`MHz band and keep the baseband signals
`untouched.
`
`Between the MuxNode and the mFN,
`coarse WDM (l.3J..trnll.5J..tm) is used for single
`fiber bi-directional
`transmission.
`At
`the
`MuxNode, the combined upstream signals are
`received and separated. The ASK signals are
`demodulated and multiplexed, and the 5-40
`MHz band signals are RF combined
`(or
`frequency stacked).
`Both signals will be
`transmitted to the PH over the same or separate
`DWDM trunks.
`
`COST COMPARISON
`
`Front-end Cost
`
`To evaluate implementation feasibility,
`we completed more than 600 miles network
`design of the above three architectures, and
`compared the front-end cost. An example is
`shown in Fig. 7. The design was over a
`medium-density system with 80 IlliP/mile. The
`analysis was based on the existing and emerging
`lightwave
`and RF
`technology,
`and
`the
`parameters used in the cost model, such as labor
`cost, were based on commonly used industry
`averages.
`
`1999 NCTA Technical Papers-- page 27
`
`Cisco Systems, Inc., Exhibit 1035, Page 8
`
`

`
`Network Availability
`
`increases
`availability
`The network
`dramatically,
`especially
`after
`the
`implementation of phase 2.
`In phase 1,
`elimination of the active cascade and the lower
`number of actives will allow for reduction in
`failures per MuxNode area.
`number of
`Moreover, due to the star configuration, the
`failure group size will be significantly lower
`(especially
`in
`the case of optoelectronics
`failure). Additional improvement in network
`availability will be realized thanks to lower
`MTTR (easier network troubleshooting). This
`feature will be significantly enhanced after
`of
`phase
`2
`(dedicated
`implementation
`smaller group of homes).
`bandwidth
`to
`Optimized powering will further improve the
`network availability.
`
`Flexibility
`
`The MFPC architecture maintains the
`transparency of the HFC network. In phase 1 of
`the implementation, the architecture does not
`differ
`from
`the
`traditional
`node-based
`architecture except for the fiber depth. When
`phase 2 is deployed, the network combines all
`the benefits of the HFC network and the cell(cid:173)
`based multistage multiplexing architecture. Any
`new service can be implemented by addition of
`the terminal equipment at the customer premises
`and in few network centers.
`
`This MFPC supports today' s terminals,
`while paving an easy migration path to enable
`high-performance
`terminals
`and
`simple
`operation scheme. The DOCSIS-based cable
`modems can operate normally with better
`performance due to reduced bandwidth sharing
`and potential elimination of the noisy 5-15 MHz
`band. The simple modems, based on simple
`modulation and local access control ("Ethernet(cid:173)
`like") protocol, can be added on when they are
`needed.
`
`• Current Network: 5.5 actives/mile
`
`Fig. 8 Current FSA HFC
`
`• 61% reduction In active components
`+ 21'% improvement in reliability
`
`Fig. 9 MFPC
`
`CHARACTERISTICS OF MFPC
`NETWORKS
`
`Powering
`
`Due to a significant reduction in the
`number of active devices and the progress in RF
`active component
`technology,
`the power
`consumption of the network elements will be
`decreased by at least 50%. This will allow for
`powering architecture optimization and for less
`challenging powering of the customer terminal
`devices.
`
`1999 NCTA Technical Papers-- page 28
`
`Cisco Systems, Inc., Exhibit 1035, Page 9
`
`

`
`architecture
`three
`studied
`solutions. We
`alternatives, analyzed their capabilities, and
`compared their costs. The results demonstrated
`that, by utilizing multi-stage multiplexing
`lightwave and RF
`topology and emerging
`technology, deep fiber penetration offers a
`future-proof
`network. The
`new MFPC
`architecture
`supports
`all
`current
`system
`operation with better performance, and enables
`abundant noise-free bandwidth
`for
`future
`growth. The advantageous architecture reduces
`the number of active components in the field,
`therefore simplifying network powering and
`operation. The incremental front-end cost can
`therefore be justified by life-cycle savings.
`More operation savings can be further realized
`with simplified terminal equipment. Yet the
`capacity, reliability, and performance of this
`new network are far better than that of other
`alternatives.
`
`ACKNOWLEDGMENT
`
`The authors would like to thank their
`colleagues: Adel Saleh,
`former division
`manager of AT&T Labs, and Mark Dzuban,
`Cameron Gough, Marty Davidson from AT&T
`Broadband Services Engineering
`for
`their
`support and assistance
`to
`the
`advanced
`architecture study.
`
`The authors also would like to express
`their gratitude to Bogdan Lomnicki and his team
`at Quasar Inc for designing the advanced
`architecture over our cable networks.
`
`Capacity Growth and Future-Proofing
`
`In phase 1, the network can deliver
`several (up to 10) TSD channels downstream
`with either 64 or 256 QAM modulation
`(presented in reference [1]).
`The reverse
`capacity per home passed can be increased by
`50% if 64QAM modulation can be used (the
`MFPC architecture will provide significantly
`better reverse performance from the point of
`view ingress and interference management).
`
`In addition, this architecture provides
`several unique paths for bandwidth expansion
`that other networks cannot match. In phase 1,
`this can be accomplished by:
`
`reverse
`traditional
`the
`1. Expansion of
`bandwidth
`to 45/48 MHz
`(no
`filter
`cascading)
`2. Wider implementation of concentration at
`the MuxNode (instead of conventional RF
`combining)
`
`More important, the l\1FPC ensures that
`the major part of the network stretching between
`the prim~y hub ring and the mini fiber node is
`future-proof and provides
`large bandwidth
`capacity (a bandwidth upgrade for this part will
`require upgrading some equipment
`in
`the
`primary hub, secondary hub, MuxNodes and
`mFN). In the passive coax plant, up to 100
`Mbps capacity can be added and shared among
`100-200 homes passed. This capacity can be
`provided with QPSK modulation and will be
`symmetrical. An increase in this capacity can
`be further achieved by deploying more efficient
`modulation schemes. Positioning
`the mFN
`closer to customers further paves a simple path
`to even bring fiber to the homes when it is
`needed.
`
`CONCLUSION
`
`HFC is no doubt the first economically
`viable means for broadband services. The need
`to further upgrade networks for future-proofing
`motivates industry to explore new architectural
`
`1999 NCTA Technical Papers -- page 29
`
`Cisco Systems, Inc., Exhibit 1035, Page 10
`
`

`
`1 Oleh J. Sniezko & Tony E. Werner, Invisible Hub or End-to-End Transparency, 1998
`NCT A Technical Papers.
`2 Tony E. Werner & Oleh J. Sniezko, HFC Optical Technology: Ten Years of Progress
`and Today's Status, Choices and Challenges, CED, September 1998.
`3 Xiaolin Lu, Ted Darcie, Alan Gnauck, Sheryl Woodward, Bhavesh Desai, Xiaoxin Qiu,
`Low-cost Cable Network Upgrade for Two-Way Broadband, SCTE 1998 Conference on
`Emerging Technologies.
`4 Oleh J. Sniezko, Video and Data Transmission in Evolving HFC Network, 1998 OFC
`Conference.
`5 Thomas G. Elliot and Oleh J. Sniezko, Transmission Technologies in Secondary Hub
`Rings- SONET versus FDM Once More, 1996 NCTA Technical Papers.
`6 Oleh J. Sniezko, Reverse Path for Advanced Services- Architecture and Technology,
`1999 NCTA Technical Papers.
`7 Network Architecture, Report TS-NETW-981001 (internal AT&T report)
`
`1999 NCT A Technical Papers -- page 30
`
`Cisco Systems, Inc., Exhibit 1035, Page 11