Transcript No Slide Title

Preparing Cable
for
Telephone Reliability
David Waks
System Dynamics Inc.
Spring 2000 Voice On the Net
Wednesday, March 29, 2000
System Dynamics Inc.
Copyright © 2000
Summary
• Modern “hybrid fiber-coax” (HFC) cable plant is much
improved, but has multiple potential points of failure
• Data network and physical cable plant are distinct
network management domains
• Present cable telephony solutions incorporate plant
management systems
• IP telephony over cable will require high availability to
be competitive with ILEC
• Several approaches to achieve high availability
• SNMP-based CPE such as cable modems will provide the
capability to monitor the physical cable plant
• Several vendors are developing solutions based on
SNMP
Copyright © 2000 System Dynamics Inc.
Slide 2
Plant Requirements for Primary Telephony
• Availability - both MTBF and MTTR
– Needs to match customer experience and expectations:
“always get dial tone”
• Cable plant is underpinning for all services
– Amplifiers in cascade, node size
– Powering - centralized at node versus distributed
– Many possible points of failure
• Different operational models
– Twisted pair telephone plant is not monitored — single
failure usually impacts only one subscriber line
– Cable plant has generally not been monitored — single
failure can impact many subscriber lines
• Three approaches to achieve high availability
– Redundancy - fiber, electronics
– Eliminate points of failure
– Proactive network management of physical cable plant
Copyright © 2000 System Dynamics Inc.
Slide 3
Traditional Cable Architecture Coax “Tree-and-Branch”
Amplifier
Headend
(or hub site)
Typically 25-30 amplifiers in cascade
Copyright © 2000 System Dynamics Inc.
Source: General Instrument
Slide 4
Modern Hybrid Fiber-Coax (HFC) Architecture
Coax
“Node” or
“optical-electrical
interface”
Fiber
Headend
(or hub site)
Neighborhood
Replaces trunk cascade
with fiber to “node” in
each neighborhood
•125 to 2000 home
“fiber serving area”
Broadcast “star-bus”
•“Star” from headend
to node
•“Bus” from node to
homes
Many fewer amplifiers
AT&T testing “fiber
passive coax” approach
•Logical extension of
HFC
•50 to 75 homes per
“mini node”
•Reduces or eliminates
trunk amplifiers
Source: General Instrument
Copyright © 2000 System Dynamics Inc.
Slide 5
Circuit-switched Cable Telephony
HFC distribution plant
Arris Interactive
Cornerstone equipment
Class 5
Switch
Host digital
terminal
Customer Home
PSTN
GR-303
or V5.2
(T-1/E-1)
“Voice port”
( or NIU)
Hub Site or Headend
• Current “HFC telephony” products incorporate plant
management systems
– NIU acts as status monitoring transponder
– Centralized analysis and reporting systems
– Detect failures and help isolate to specific home or small
segment of cable plant
Copyright © 2000 System Dynamics Inc.
Slide 6
Traditional “Status Monitoring” Systems
• Can monitor many possible points of failure
– Optical nodes
– Power supplies
– Amplifiers (rarely)
• Best implementation - Rogers Cable (largest in Canada)
– Transponders on all nodes, power supplies and amplifiers
– Data collection system at each headend
– Identifies trouble spots using weekly “figure of merit” for
each secondary hub (~5,000 passings)
– Isolates problem to specific amplifier using fault-detection
analysis of transponder data trends
– Substantial measurable improvement in data throughput
– Moved from R&D to system-wide implementation
• Most operators have not invested in status monitoring
– Monitor nodes and power supplies, not amplifiers
– Don’t want proprietary protocols, equipment and software
– Believe solution based on industry-standard protocols and
software would be much less expensive
Copyright © 2000 System Dynamics Inc.
Slide 7
New Digital CPE Offers Better Solution
• DOCSIS cable modems now industry standard
– Many certified vendors for modems and headend “CMTS”
– Rapidly replacing proprietary technologies
– Moving to retail distribution, customer installation
• All new digital cable devices support SNMP
– DOCSIS modems include robust SNMP MIB structure
– Digital set-top boxes
– IP telephony CPE based on PacketCable™ specs
• Digital devices will provide many more test points than
traditional “status monitoring”
– Many cable systems already have higher penetration of
customer-provided SNMP devices than possible with
traditional amplifier-mounted transponders
– Provides closer determination of failure location
Copyright © 2000 System Dynamics Inc.
Slide 8
Need New Software Systems
• DOCSIS modems provide two types of data
– Traditional measurement of signal levels
– Error counters on PHY layer of protocols
• Needs software to realize potential
– Map CPE measurements against model of plant topology
– Isolate problems to small section of plant
– Distinguish between in-home and plant problems
– Anticipate failures
• No “off-the-shelf” solutions yet, but several vendors
working toward one
– Cisco
– Com21
– Cheetah Technologies
– C-COR.net
Copyright © 2000 System Dynamics Inc.
Slide 9
Cable Operators Taking Several Approaches
• Rogers
– Substantial investment in status monitoring
• Cox
– Redundant fiber - “ring in ring” architecture
• AT&T BIS
– Reduce or eliminate points of failure with “fiber passive
coax” extension of HFC
• Operators will “mix-and-match” to achieve desired
reliability at reasonable cost
• CPE-based status monitoring highly promising
– Based on existing CPE and centralized software
– Much lower cost than rebuilding plant
– Fastest to deploy once proven
Copyright © 2000 System Dynamics Inc.
Slide 10
For More Information:
System Dynamics Inc.
18 Beaver Ridge Road, Morris Plains, NJ 07950-1901
(973) 644-4739 Fax (973) 538-6003
dave @ system-dynamics.com
sandy @ system-dynamics.com
http://www.system-dynamics.com