The Networked Infrastructure: Genlock and 10 Gb Ethernet

Paul Briscoe

Harris Broadcast

Networked Infrastructures • Evolution

–Past - minor threads of connectivity –Today - parallel systems –Future - full infrastructure

• Inverting the broadcast system cloud

Networks Past • Broadcast interconnect “cloud”

– Video, Audio, Control • Application specific network islands

– Computer graphics – Automation – Control and monitoring – Intercom

• Isolated, independent, non-standard

Broadcast “Cloud”

Network Islands

Networks Present • Parallel Clouds

– Media lives in both – Streaming distribution – File-based workflows – Contribution – Allied services

• Greater integration • Standardized methods, protocols

Broadcast Cloud

Network Cloud

Equipment

Networks Future • IT becomes the infrastructure • Islands of legacy broadcast

– Gradually replaced by IT • Support all services

– Homogenized environment – Live baseband – Production

IT Cloud

Broadcast Islands

Network Ubiquity • Broadcast equipment is specialized

– Limited selection of vendors – Price floor - volume and specialty

• IT equipment is everywhere – ‘Enterprise grade’ = ‘broadcast quality’ – Vast vendor selection (incl. quality!) – Pricing vs. volume: generic HW and SW

• Continuing trend

The File Environment • Non-realtime workflow • Large data • Simple protocols (FTP, etc.) • Constrained span • Same as a ‘regular’ IT network

The Evolutionary Facility • Hybrid operations today • Migration to live IP

–Baseband, compressed • Integrated services on the network

• Requires strong enablers

Two Network Enablers • Sync and Time (Genlock / Timecode)

–Distribution over network • 10 Gb ethernet over copper

(10GBase-T) –SDI-like baseband interconnect –Comparable physical layer

Legacy Synchronization • Video

– Switching, Mixing – Use Genlock – Black Burst or TLS

• What about audio? – DARS (AES3 / AES11)

• What about time? – Timecode SMPTE ST-12

• All are at end of life

Legacy Synchronization

Legacy Synchronization

Studio A

Studio B Edit 1

OB Truck

Edit 2 Graphics

MCR Feeds

Master Generators

�

Legacy Synchronization

Studio A

Studio B Edit 1

OB Truck

Edit 2 Graphics

MCR Feeds

Master Generators

�

Legacy Synchronization

Studio A

Studio B Edit 1

OB Truck

Edit 2 Graphics

MCR Feeds

Master Generators

�

SynchronizationTopology Master SPG

Autochangeover

First Floor Second Floor Third Floor Fourth Floor Fifth Floor Sixth Floor Seventh Floor

NW Quad SW Quad SE Quad SW Quad

STD 12 STD 13 STD 14 STD 15 STD 16

Master Control / Central Equipment Fifth Floor SW Quad STD 14

Studio Gear

Studio Gear

Slave SPG

Master SPG

SynchronizationTopology X2 Master SPG

Autochangeover

First Floor Second Floor Third Floor Fourth Floor Fifth Floor Sixth Floor Seventh Floor

NW Quad SW Quad SE Quad SW Quad

STD 12 STD 13 STD 14 STD 15 STD 16

Master Control / Central Equipment Fifth Floor SW Quad STD 14

Studio Gear

Studio Gear

Slave SPG

Master SPG

Autochangeover

Slave SPG

The Opportunity

Studio A

Studio B Edit 1

OB Truck

Edit 2 Graphics

MCR Feeds

Master Generators

� �

IP Network

Legacy Limitations • Ancient technologies (from a simpler time)

– Mix of analog and digital-ish signals

• Serious modern functional limitations • Don’t support > 60 fps, VFR • BlackBurst ≠ HDTV • Require multiple nailed-down distributions

– CAPEX, OPEX

Requirements Today • Deliver references over IP network • Provide capability of legacy systems • Provide performance for today and future • Deterministic behaviour • Provide open platform for future • Signals (transports) • Formats (essence) • Use COTS technology wherever possible

Network Characteristics • Not so good for reference delivery

– Non-deterministic • Can’t predict when things happen • Traffic all shares the road (wire, fabric)

• Packet delays are variable • Jitter!

– Things get lost

Genlock Requirements • Deliver Frequency

– Different for each signal type used – Varying degrees of tolerance

• Deliver Phase – VSync, HSync ST 318 – Audio block, frame AES3 / AES11 – Time of Day ST-12 – Date ST-309

Frequency from Time • Transferring time to a clock transfers frequency • Consider a pendulum clock • “At the tone the time is…” • Set the time • Wait a day • “At the tone the time is…” • Set the time, adjust the pendulum • Repeat

Signals from Time • Generate required frequencies • Pick an Epoch where time=0 • Define signal alignment at Epoch • Start time counter • Calculate signal phases from time

A Candidate Solution? • NTP

• Transfers time • Works on IP networks • Deterministic • Delivers time • Not enough performance

• millisecond-class time accuracy

Enter IEEE1588 PTP • Network-based Precision Time Protocol • Delivers precision time to many slave clocks • Spans hundreds of years (2177) • Sub-nanosecond granularity (12.5 ps) • Delivered over IP network • Can be globally locked (GPS, GLONASS, etc.) • Can co-exist with other traffic

Enter IEEE1588 PTP • Master and each slave

measure path delays • Slave corrects for

measured delay • Ensures all slaves are

in sync with master

Enter IEEE1588 PTP • Master sends a packet

down the stack • At the MAC layer,

timestamp is inserted • Receiving MAC

extracts timestamp • Sends it up to wait for

the packet to arrive at the top of the stack

IEEE 1588 PTP Time Count

1 Se

cond

1 M

inut

e

1 H

our

1 D

ay

1 M

onth

1 Ye

ar

1 M

illise

cond

1 M

icro

seco

nd

1 N

anos

econ

d

SDI Video

12M Timecode

Calendar

NTP

AES Audio

Composite Video

IEEE1588

GPS Native

Nanoseconds – 32 bits (1 ns)Whole Seconds - 32 bits (~136 years)

Time Counter

LSB

1 Hz(PPS)

1588 Applications

Switches get Involved • Switches introduce variable

latency • PTP switches actively

manage delay – Transparent Clock switches

recalculate time – Boundary Clock switches ‘re-

master’ time

Switches get Involved PTP

Grandmaster

Boundary Clock Switch

GPS Rx

Same time in every clock Slave

Boundary Clock Switch

Slave

Switches look like a Slave to upstream Master

Slave Clock

Master Clock

Slave Clock

Master Clock Switches look like a Master to downstream Slaves

Slave Clock

Slave Clock

Master Clock

Net Result? • Can deliver all references over IP network • Can have multiple independent systems in phase • Can share other traffic on same network

• Control, monitoring • File-based workflows • Compressed streaming • Live Baseband

One Other Thing • Timecode (= time labelling)

• 1588 offers an opportunity

• Greater granularity / span than timecode • Can label media with it

• ST12 equivalent labels possible • Any framerate / date span possible • It’s already in the equipment for synchronization

purposes • Common timestamping technology across industries

SMPTE Activities • 33TS Synchronization and Time Labeling

Standardization Committee • ST 2059 Standard Suite

– Includes EGs and RPs • ST 2059-0 Overview of Standard • ST-2059-1 SMPTE PTP Profile • ST-2059-2 Generation of Signals

SMPTE Activities • Two 33TS Working Groups • 33TS-20 is developing sync system

• Work nearing completion • 33TS-10 developing new Time Label

• Work actively underway

In Summary • Reference signal distribution – coming soon to a

network near you! • No need for independent distribution of multiple

reference signals • Can be used on same network as media • Can support all present and future types of video

and audio signals • Delivers timecode “+” for time labeling

Now,What About the Essence?

�

IP Network

Live Networked Media • ‘IT’ Networks are not ‘live’ friendly

– Reasons noted previously! • Not ‘time-sensitive’ nor traffic aware • Use retransmission to fix errors and loss

– Or • Happily accept loss

Live Networked Media • Compelling solution

– Flexibility – CAPEX / OPEX – Ubiquitous infrastructure equipment – Multifunctional

• Video, Audio, Control, Monitoring, ICM, FTP, Web, email, etc.

Networks Future • IT becomes the infrastructure • Islands of legacy broadcast

– Gradually replaced by IT • Support all services

– Homogenized environment – Live baseband – Production

IT Cloud

Broadcast Islands

The Live Environment • It’s Live! • Cannot tolerate:

– Latency (variable or too much) – Errors (no time to fix) – Outage

• Equivalent to hard-wiring • Same operational capabilities as today

Serial Digital Interface History • SD-SDI (ST 259) – Mid ‘80s • HD-SDI (ST 292) – Mid ‘90s • 3G-SDI (ST 424) – Early ’00s • Where next?

– 6 Gb? – no standard activity – yet – 12 Gb? On Coax?

Why So Popular • Simple evolutionary coax interface • Useful reach – comparable to analog • Robust and “error free” signal • Audio included • Ancillary services

– AFD, PID, Captioning..

SDI Limitations • Unidirectional • Reach drops as bitrate increases

• Technology can mitigate to some degree

• Dedicated wiring, one to one connectivity • Pipe is mostly full

– ANC data space is small!

Where to Next? • Equally capable physical layer • Available headroom • Bidirectional • Same or better error performance • Not industry-specific technology

A Possible Solution • 10 Gigabit ethernet

– Fiber or copper • Lots of bandwidth (vs. SDI) • Scalable through bonding • Becoming ubiquitous in IT world

IT Network Characteristics • Not so good for essence delivery

– Non-deterministic • Can’t predict when things happen • Traffic all shares the road (wire, fabric)

• Packet delays can be long, and are variable • Latency, Jitter

– Things get lost or resent

Protocols to the Rescue! • Protocols are coming which solve the

problems – AVB – SMPTE 2022 suite – IETF

What can 10 Gb Support? • ~33 SDI (720 x 480 / i30) • ~ 5 HDSDI (1080i30 / 720p60) • 3 3Gb SDI (1080p30) • 1 4Kp30

– Varying degrees of overhead – Bidirectional

Transport Methods • AVB

– Adjacent space to broadcasting – Developed for media over small networks – Initially compressed media only – Now being extended to high bitrates – Network Layer 2 (MAC)

Transport Methods • SMPTE 2022

– Encapsulation and transport – Provides Forward Error Correction (FEC) – Initially for compressed streams – Being updated to handle high bitrate – Network Layer 3 (IP)

SMPTE 2022 Suite

The nBase-x standards • Ethernet IEEE 802.3 root • (Speed)Base-(Media)

– 10Base-2, 5 – 10 Mb over coax, shared media – 10BaseT - 10 Mb over UTP, hub / router / switch

• Introduces the RJ-45 Connector • Extended to 100 and 1000 Mb versions • Added fiber types

Ethernet at 10 Gb • Full duplex / point to point links only

– Switched topology • no shared media = no collision loss

• Many PHY types – Copper, fiber (modes), backplane, InfiniBand – Differing media, reach, coding, modulation

10GBase-T • “IEEE 802.3an-2006” • RJ-45 (8p8c) connectors

– World’s most deployed non-power-connector type

• UTP / STP wiring – Differing reach

• Power hungry interfaces

On The Wire • Data is pre-coded for interference

avoidance • Signal on wire is PAM-16 modulated • 55M / 180 feet on Cat-5 cabling • 100M / 330 feet on Cat 6 / 6e / 7 cabling

– Existing infrastructure wiring can be used. • Can throttle down to 1 Gb

10GBase-T Power • Higher than SDI • Comparable when number of

channels considered • Declining with process geometry • Mitigation strategies

– Power tuning (length) – Green ethernet (idling) – Wake-on-LAN (snoozing)

Cabling Considerations • Standard structured IT cabling

– Cost effective • 1694A coax = ~$1.00 / foot • BNC = ~$1.25 • Cat6e cable = ~$0.36 / foot • RJ45M = ~$0.40

• Friendly legacy cabling technology

Connector Considerations • BNCs / XLRs are tough • RJ-45’s are pretty good • Ruggedization systems exist

– Reinforcement – Housing / shell (e.g. XLR body)

• Robust, weatherproof, failsafe

Power Over Ethernet (POE) • Phantom Power

– 15W / 25W / non-std (60 today) • Power management

– Sensing – Protocol

• Enables ‘single connect’ equipment

Moving the Baseband Media • SMPTE 2022

– Provides encapsulation of SMTPE standardized formats (x-SDI)

– Provides FEC protection – Specifies use of RTP for transport – Works at IP layer (layer 3)

Moving the Baseband Media • AVB (IEEE)

– Provides mapping of SMTPE standardized formats

– Utilizes protocol-aware switches • Allocate bandwidth, meter traffic

– Works at MAC layer (layer 2)

Moving the Baseband Media • AES X-192 (AES)

– Provides transport of audio – Maps AES3 – Works at IP layer (layer 3)

Compressed vs Baseband • Application

– Essence quality – Available / desired bandwidth – Tolerance of latency – Power, cost considerations

Other Services • SIP Sessions (VOIP)

– IFB, ICM, orderwire • MDCP

– SMPTE control protocol development • FTP • Email, chat, cloud editing

What about 4K • 4K res demands higher framerates

– 4Kp30 = 6 Gb/s – 4Kp60 = 12 Gb/s – 4Kp120 = 24 Gb/s

• Link bonding can achieve summed aggregate bitrates – Logically bound together

System Architectures • Will remain router-centric

– Video router becomes IP switch fabric – Everything goes through the fabric

• Connectivity reduction • One type of physical plant wiring • Hardwired equivalency by management

Timeframe? • Critical mass

– PHY power, cost – Ubiquity – Broadcast protocols, methods

• Takeoff in IT – 2013-2014 • Takeoff in Broadcast – 2014+

Standards Activities • SMPTE

– SMPTE 2022-6/7 HBRMT WG – Study Group – Networked Media Architectures – AHG – Media Device Control – SMPTE 2059 – Sync and Timing WG

• AES – X-192 audio

Standards Activities • IEEE - AVB

– 802.1AS – synchronization – 802.1 Qat – Stream Reservation Protocol – 802.1 Qav – Forwarding and Queueing for

Time Sensitive Streams – 802.1BA – Audio and Video Bridging Systems

Future Standards Activities • IEEE

– Wifi transport of AVB – AVB transport of SMPTE 2022 payloads – Interoperability between AVB / 2022

• IETF – New RTP / RTCP / SIP protocols

Summary • It’s all moving to the network.

– Get used to it. • It’s not your granddad’s network anymore.

– Network capabilities are now beginning to exceed our requirements. It’s time to strike!