|
Switched
Digital Video (SDV) White Paper
Building a new foundation for efficient delivery of video content
By: Bruce Bahlmann - Contributing Author (your
feedback
is important to us!)
Created: July 12, 2001
Note: For help designing your provisioning system or developing tools to help test, automate, and deploy your system contact Birds-Eye.Net.
“The more things change, the more they remain the same”.
There is a lot of truth to this statement especially in the business of delivering video
content to subscribers over broadband transports. Although there have been significant
improvements in video delivery technology over the years most of the basic video
distribution methods (e.g. broadcasting) remain the same. Broadcasting built the
foundation of the broadband industry and is the cornerstone of most of its service
delivery systems. However, this foundation has become increasingly costly to host new
services and as a result has developed cracks from which more innovative solutions are
sprouting. Remote switching, one of the most promising of these solutions, will challenge
the broadcast foothold in the delivery of video content. This white paper will explore the
depths of remote switching, also called switched digital video (SDV), and many of the
possible benefits that it would bring.
Broadcasting Video Content:
Broadcasting video content is nothing new (see Figure 1.0). It
involves pulling video content (HBO, CNN, ESPN, etc.) off satellites, assigning them a
carrier frequency (upconverting), combining all these carriers on to a single transport,
and then pushing them down a broadband pipe using amplifiers to the set top box (STB) or
traditional television set belonging to subscribers. Broadband pipes are long stretches of
coaxial cable strategically placed in close range of a collection of residential homes
that in turn provide easy/cheap connectivity to the signals they carry. Modern day
broadband operators have combined the use of coaxial cable and fiber optic cable (known as
hybrid fiber coax or HFC) to increase the range they can transmit quality analog video
signals thus increasing the number of communities any one-facility support.
Figure 1.0 Broadcast
Model
Subscribers of this broadband service are able to gain access to this
content by changing channels. The process of changing channels, instructs their STB or
television to tune to a different carrier that is associated with some other content (e.g.
different television station such as TNN). While broadcasting signals down coax is
relatively easy to do, it is an extremely inefficient use of the vast broadband transport.
This is because in order to use the broadcast model to supply a number of video content
sources one would need to send all of this video content down the broadband transport.
Having all this content readily available on the broadband transport enables
subscribers’ instant access to it via changing channels.
In summary, the broadcast model sends all available content down the
broadband transport to enable subscriber placed channel changing to instantly access this
content. The broadcast model sends this content continuously in the presence or absence of
subscribers. The broadcast model is also, by its nature a one dimensional delivery
mechanism. It attempts to deliver the same thing to everyone no matter who they are, where
they live, or what kind of content they would like to receive, or what kind of viewing
device (STB, television, etc.) they have. Lets look at a few of the more inherent
inefficiencies of the broadcasting model.
Broadcast must be all things to all places, all people, and all
devices. For example when stereo sound was introduced to broadcast it had to support mono,
Color had to support black and white, second audio program (SAP) had to support a primary,
etc. Thus improvements in this technology was required by the FCC to be backwards
compatible with older (existing systems). Today, Apple Computer Corporation’s new
operating system does not support all its older versions – only those 5-10 years
back. However, today’s newest broadcast standard still supports black and white
reception that is more than twice as old as Apple. Innovation in television content has
been slowed as a result of this constant requirement of backward compatibility. In the day
and age of disposable television sets where it is generally cheaper to replace than fix
why do we still insist on such rigorous backwards compatibility?
The broadcast model also limits new content from being offered. For a
broadband operator to offer new content they must provide that content to everyone all the
time. For example if a broadband operator wanted to offer some Spanish content to its
Spanish speaking subscribers it must send it everywhere – broadcast it! Eventually,
broadcasting leaves its broadband operators with nearly all of its available
spectrum/bandwidth consumed by broadcasted content. Any remaining spectrum is usually
earmarked for revenue generating content or future services – that leaves the
broadband operator with little (if any) unused or unallocated spectrum. If these broadband
operators want to offer something new they must either take something away (which is hard
to do) or add more spectrum (which is extremely costly).
Broadcast is also a waste of advertising opportunities. There are
only so many advertising slots available. Since each is broadcasted that means each slot
is also one-dimensional. That is, once filled it goes out the same to everyone.
Perhaps the most unfortunate thing about the broadcast model is that
it has set an ominous trend or arbitrary standard for new technologies to follow. The best
example of this is digital video (DV) distribution system. While DV boasts significant
efficiencies in transmission of high quality digital video content its use of the
broadcast model infects this new technology with the same broadcast inefficiencies –
thus DV consumes a significant portion of bandwidth/spectrum (albeit more efficient than
that of analog video content) and further limits the availability of unallocated
spectrum.
These limitations of the broadcast model paint an ugly picture of the
state of affairs for a majority of broadband operators today. In fact, most were unable to
provide the likes of high-speed data services over cable modems, telephone service, and
high quality DV without investing millions in their infrastructures to increase the
available spectrum required to provide these new services – which is a testament of
how the broadcast model limits new content. However, because DV uses the broadcast model
most of this newly created spectrum has already been consumed – all but perhaps a
small fraction of their available spectrum (made possible by completed upgrades) remains.
As a result these broadband operators find themselves back in an all too familiar position
of not having enough spectrum to provide the increasing array of new content being
developed (such as video on demand, high definition television, etc.) which their
competition is egger to offer. For these broadband operators, the thought of upgrading
their networks once again is not an option – especially in light of the fact that
most are still paying for the last upgrade and plan to keep doing so for years to come.
These broadband operators, which represent the vast majority, need to find a better way to
distribute video content than that used by the broadcast model to free up significant
chunks of spectrum for new services. That better way is remote switching!
Unicasting Video Content:
Unicast has the ability to deliver unique content to subscribers
where they want, what they want, when they want it, and how they want it. Essentially,
unicasting is delivering content to the subscriber in a manner in which it only consumes
the spectrum required to transmit the specific content requested. For example, if a
subscriber wanted to watch Randy Moss and the Minnesota Vikings play the Greenbay Packers,
they would only consume the spectrum required to receive that content.
How can this be applied to a broadband transport that is for the most
part considered a broadcast transport medium? The answer lies in the ability to perform
remote switching. Remote switching has also been around for a long time. One of the
simplest examples of remote switching is one’s VCR as shown in Figure 2.0.
Figure 2.0 VCR Example
VCRs have the ability to switch between incoming video content coming
from an antenna or broadband operator and video content coming from playing VCR tapes. The
VCR upconverts what ever content source that is selected to VHF channel 2 or 3. If the
television that is connected to the VCR is tuned to this channel the VCR tape can be
viewed. By toggling this VCR’s remote switch, the television viewer is able to view
either VCR tapes or the incoming video content from the antenna or broadband operator
using the VCR’s built in tuner.
The significance of the VCR example is that it shifts video content
switching or channel changing from the television set upstream to the VCR with no
noticeable impact to the viewing experience. Shifting this switching upstream enables
greater choice of video content and increased flexibility by offering video content the
television does not have the ability to receive or produce on its own. In this way, the
television becomes less a channel changer and more a monitor that has its tuner positioned
upstream. The STB is but an extension of this model giving access to individual program
call letters, program descriptions, on-line program guides, etc. – none of which
televisions can provide by themselves.
Interestingly, the video content upstream from either the VCR or STB
is all broadcasted. However, the video content between the VCR or STB and the television
is unicast. Since whichever video content selected is sent to the television (i.e.
upconverted to either VHF channel 2 or 3), that is the only content being passed to the
television and this content only consumes the spectrum of one broadcast channel (6
MHz).
Pushing the remote switching further upstream permits traffic between
the STB and the distribution hub to be unicast. Figure 3.0 shows a simplistic example of
how one STB would remotely switch the content being sent down to it from the distribution
hub.
Figure 3.0 Distribution Hub
Placed Remote Switching
Here the STB controls the switch and what video content it is
connected to. The result of this technology enables each STB to only consume the spectrum
required to view the video content selected. This approach has other benefits as well, for
example:
If the STB is shut off, the spectrum previously used by the STB is
now freed up for use by some other STB. In fact this model closely resembles that of how a
Remote Authentication Dial In User Service (RADIUS) works. Dialup Internet customers
authenticate with RADIUS to obtain temporary access criteria (IP address, DNS, etc.) that
they return upon disconnect (or logging off the service). Similarly, remote switching can
allocate spectrum for the STB to operate and then release these resources once they are
powered down. In this way the amount of spectrum in use would be determined by the actual
number of subscribers using the service rather than (in the case of broadcast) the amount
of content being offered.
Remote switching can also provide the flexibility that would enable
more of a telephony type of model in terms of how it is designed and built to handle peek
load. Thus, only a portion (typically regional bells only allocate enough bandwidth to
permit 20-25% of their customers to access the network at the same time) of the overall
spectrum is allocated – e.g. just enough to support peek demand. The remaining
spectrum is completely freed up or placed in reserve for deploying additional services.
Remote switching can work with either analog or digital video
content. In the case of using it with digital video one can realize a savings of up to 83%
of the existing spectrum used for delivering this content. That equates to a savings of
about 80MHz of spectrum for digital video services alone. Adding new services to unicast
does not double spectrum consumption but only slightly increases it. This is because it is
all based on usage patterns and average load.
Analog video represents the worst cast implementation of remote
switching. An analog video channel consumes 6 megahertz worth of spectrum. This would
permit about 90 channels worth of content in the broadcast model over a 550 MHz transport
or ~90 simultaneous subscribers if remote switching were used. Depending on how many homes
each segment (also known as a node) of the HFC passed, which ranged from 1000 to 2000 in
the early days of cable television, 90 simultaneous represented 5-10% of these homes
passed which was likely to be exceeded. However, with today’s 250 to 500 homes
passed, 90 simultaneous subscribers represent 18-36% of the possible homes passed. What
this means is that a 550 MHz HFC node with 250-500 homes passed count could utilize remote
switching using analog video. These subscribers (up to 90 simultaneous users) could then
have access to hundreds of channels of video content that would only be possible via
broadcast model by upgrading spectrum beyond a gigahertz.
Digital video changes everything! The use of digital video with
remote switching pushes the envelope of what is possible within a 550 MHz system. Within a
single 6 MHz analog video channel, digital video can transmit 12-16 different high quality
video channels*. What digital video allows a broadband operator to do is achieve 12 to 16
times the number of simultaneous users per segment of HFC. So, instead of supporting only
90 simultaneous users per segment, digital video will support 1,080 to 1,440 simultaneous
subscribers. This would permit broadband operators to maintain their previous investment
in 550MHz transmission hardware and merely add some two-way capability. Not having to
upgrade to 750 MHz or 860 MHz would mean a significant savings to the broadband operator
while permitting them to keep existing HFC segment sizes anywhere from 1000 to 1500 –
also a significant savings. Statistical multiplexing digital content spectrum (or
bandwidth) utilization across multiple subscribers would allow certain individuals to use
more or less spectrum based on need or application.
*Note use of the word “channel” becomes arbitrary within
the digital video domain as it no longer differentiates video content by frequency but
rather by transports and streams.
In summary, remote switching extends the life of a 550 MHz system and
permits a 750-860 MHz system to be nearly future proof. Once more, it would finally put a
stop to all this nonsense about reducing node sizes (the number of homes passed per node).
Since this approach significantly increases efficiency (especially when used with
statistical multiplexing) the possibilities become endless and the demands for broadband
transports with increased spectrum are finally laid to rest.
Use of the broadcast model for all content deliver certainly does not
promote the best efficiency because with lots of different content to choose from it is
unlikely that all subscribers will want to see the same thing. However, there are frequent
cases where a number of subscribers want the same content at the same time – for
example Monday Night Football. Thus it would be advantageous to enable one stream of
content to satisfy this group of subscribers rather than each requesting the same stream.
Multicast provides a means of accomplishing this.
Multicasting Video Content:
Broadcasting the Super Bowl is an extremely efficient means of
transmission but how many subscribers (viewers) constitutes a good enough reason to
broadcast something? If you said, “more than one” you win the prize for the day.
However, once a video delivery system is converted to use unicast it does not make sense
to bring back broadcast. Instead, one can use multicast as shown in Figure 4.0. Multicast
is a way of providing the same content to multiple selected endpoints. Unlike broadcast
(where everyone receives it), a multicast is targeted to select endpoints – thus
leveraging a broadcast-like delivery mechanism over a purely unicast delivery system.
Figure 4.0 Multicasting
Remote Switched Content
In this way, the spectrum required to send video content to something
that is in high demand (required by 2 or more subscribers) can be multicasted to those
endpoints. The result is an increase in the number of active subscribers on the broadband
HFC who select the same content do not represent an increase in the spectrum consumed. The
drawback of multicast is that these endpoints all must receive the same content. Thus in
terms of targeted advertising one would either use the profile for the original subscriber
or a combination of the profiles for all subscribers.
Dealing with Contention
If you have a finite resource and too many people want it, you have
contention. Thus contention is when the resources you have allocated for something become
exhausted and cannot satisfy demand. In a delivery system contention is what you want and
contrary to what others might say, it’s a great problem to have. If you didn’t
have to worry about contention your not squeezing everything you can out of your delivery
medium. Multicasting is a terrific example of optimizing resources and helping reduce
contention. By satisfying multiple requests for the same content with one response (or
stream) you can reduce the number of individual subscribers who each are requesting the
same content. Using the RADIUS model also relieves contention, as only those subscribers
who are using the content delivery system should consume spectrum, the rest of the
inactive subscribers should not contribute to contention.
Other Implications of Remote Switching:
Operation of the STB should migrate from interactive software to
managing an MPEG stream that is interactive. This pushes application control to the edge
in terms of management traffic but data/content generation such as program guides, service
packages, and access control are pushed back to the distribution hub.
Individualized content would enable smart televisions to enter the
market and push out aging STBs. If a television is simply an MPEG monitor and this MPEG
stream were standardized, all televisions could once again become individualized rather
than being relegated to simply an over achieving monitor. These days, any television set
over $300 that is paired with a STB constitutes a waste of technological capability.
However, if televisions were able to take direct MPEG feeds from a broadband operator
using a smart card or some other unique identifier to constitute the incremental gain in
revenue per subscriber television – the need for STBs would greatly diminish. This
would improve the bottom line of broadband operators (as they would not have to purchase
STBs to deliver digital content), reduce the cost of digital television for consumers (as
they would not have to rent a STB), and stimulate innovation among the consumer television
industry. These next generation consumer televisions could take these direct feeds and
promote an open MPEG driven marketplace that could potentially subsume IP and the Internet
or perhaps even combine the two somehow.
This whole process may well represent an application instance being
run remotely. Perhaps similar to an ASP model but not all content will resemble
applications (interactive mainly). To date, everyone pays for a certain level of quality
in his or her picture. What if for a cheaper price they could get a slightly lower quality
picture – like ok analogy or high quality or even HDTV. In this way, quality as well
as stream type could be configurable on an individual basis. If a subscriber has an HDTV
she could get all HDTV content. That which is not HDTV compatible might be converted
– much like today’s VCR movies are made to fit the screen size of the
television. This could also manage or individualize aspect ratio and could make that an
option in further selling individualized content. Think of never again watching another
movie that only partially covered your ¾ aspect ratio television screen.
There is no reason the local switch control could not be a Network
Interface Unit (NIU). In this way it could act as a local connection point for multiple
services not just primarily interface with a television.
Handling Dynamic Content
How would remote switching work with dynamic or on the fly content
– that which is not calandarized like traditional broadcast channels? In the case of
broadcast, it is streamed down to the STB, cached, and displayed. The problem with
broadcast is to handle spectrum allocation bottlenecks in the last mile – a very
difficult space to conquer. The remote switching model does not worry about restrictions
within the last mile but rather the addressability of the subscriber’s remote switch.
One possible solution would be to create a switch position that is a file pointer as well
as an MPEG stream of content. The file pointer would keep track of where the movie is in
terms of its progress in traversing (streaming) the movie file. Management traffic could
manipulate this pointer ahead (Fast Forward) back (rewind) or freeze it (stop or pause).
Contention would be at the distribution hub for the resources rather than for spectrum on
the last mile.
Who can take advantage of this?
Remote switching uniquely applies to broadband HFC infrastructures as
they can segment their HFC delivery medium. Segmenting allows the broadband operator to
potentially treat different deliver different content to different segments. DSS is by its
nature only a single dimensional transmission medium thus it cannot segment its subscriber
base or the resources that are used to provide it. Everything is common to all –
satellite, content, etc. thus it can only leverage the broadcast model – which it
does quite well.
Starband Internet service is perhaps one exception to this as it uses
regional satellites. This provides some means of segmentation required to provide unicast.
However spectrum must be at least in the gigahertz range and ground stations must be
capable of negotiating onto non-overcrowded segments.
Summary
Remote switching seems to have an answer for the most challenging
aspects of operating a broadband delivery system based on the broadcast model. Although it
is in the early stages of development and acceptance, the benefits of this model can
permit even the smallest of broadband operators to deliver voice, video, and data services
over a 550MHz broadband transport. The ability of remote switching to deliver this
capability cheaply and efficiently could very well permit broadband operators to offer
high quality voice, video, and data services all for under the $100 price tag while still
maintaining comfortable profit margins.
Can Birds-Eye.Net help you or your Company?
Receive your Birds-Eye.Net articles and white
papers hot off
the presses by adding our RSS feed to your reader.
|