 |
Claims  |
|
|
What is claimed is:
1. A system for interactively viewing a selected one of a plurality of
videos, the selected video including a plurality of frames, each frame
including digitally encoded video data and a time stamp, the video data
representing time sequenced image signals and audio signals for play-back
on a viewing device, a communications network for transmitting the video
data, comprising:
means for requesting the transmission of the selected video on the
communications network;
a network interface, coupled to the communications network for receiving
the selected video as a first stream of video data;
a memory buffer, coupled to said network interface, for storing a segment
of the selected video, said segment including video data of a forward
moving time interval of the selected video;
a write pointer, coupled to said memory buffer, associated with said first
stream of video data to be written to said memory buffer;
a read pointer, coupled to said memory buffer, associated with a second
stream of video data to be read from said memory buffer;
means coupled to said write pointer, for writing said first stream of video
data to said memory buffer while receiving the selected video;
means, coupled to said read pointer for reading said second stream of video
data from said memory buffer while writing said first stream of video
data, said means for reading operating independently from said means for
writing, to display any video data of said segment of the selected video
on the viewing device while receiving said first stream of video data: and
an index for associating the time stamps of the frames with memory
addresses of said memory buffer where the frames are stored.
2. The system as in claim 1 wherein said memory buffer is a circular
buffer, said means for writing to advance said write pointer to the
beginning of said circular buffer upon reaching the end of said circular
buffer, said means for reading to advance said read pointer to the
beginning of said circular buffer upon reaching the end of said circular
buffer, and said means for writing to overwrite a particular video data
after said particular video data has been read.
3. The system as in claim 1 wherein said means for writing fills said
memory buffer with video data of the selected video at a different rate
than reading the video data.
4. The system as in claim 3 wherein said different rate is higher than said
rate of reading when said memory buffer is not full.
5. A method for interactively viewing a selected one of a plurality of
videos, the selected video including a plurality of frames, each frame
including digitally encoded video data and a time stamp, the video data
representing time sequenced image signals and audio signals for play-back
on a viewing device, transmitting the video data over a communications
network, comprising the steps of:
requesting the transmission of the selected video on the communications
network;
receiving the video data over the communications network as a first stream
of video data;
storing a segment of the selected video in a memory buffer, said segment
including a substantially forward moving time interval of said first
stream of video data of the selected video, said memory buffer including a
write pointer and at least one read pointer;
writing said first stream of video data of the selected video to said
memory buffer at said write pointer;
reading a second stream of video data from said memory buffer at said at
least one read pointer while writing said first stream of video data to
display any video data of said segment of the selected video on the
viewing device while receiving said first stream of video data; and
indexing the stored video data by the time stamps.
6. The method as in claim 5 further including the steps of selecting a
particular frame of said selected video, and writing the video data to
said memory buffer as said first stream of video data beginning with said
particular frame.
7. The method as in claim 5 wherein said memory buffer is a circular
buffer, said writing step to advance said write pointer to the beginning
of said circular buffer upon reaching the end of said circular buffer,
said reading step to advance said read pointer to the beginning of said
circular buffer upon reaching the end of said circular buffer, and
overwriting a particular video data after said particular video data has
been read.
8. The method as in claim 5 including filling said memory buffer with the
video data of the selected video at a higher rate than reading the video
data when said memory buffer is not full.
9. A method for viewing videos, comprising:
requesting distribution of a video over a communications network as a first
stream of video dam;
receiving said first stream of video data from said communications network;
writing, while receiving, said first stream of video data in a buffer as a
substantially continuously forward moving time interval of said video
data, and
reading, independent of said writing, a second stream of video data from
said buffer while receiving said first stream of video data;
converting said second stream of video data to video signals; and
displaying said video signals on a viewing device while receiving said
first stream of video data; and
reading said second stream of video dam from said buffer in a reverse
chronological order while receiving said first stream of video data in a
forward chronological order.
10. The method as in claim 9 further comprising reading said video data at
a different rate than writing said video data. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
FIELD OF THE INVENTION
This invention applies to the general area of video viewing systems, and
more particularly to video viewing systems which are interactive.
BACKGROUND OF THE INVENTION
There is a need to provide consumers with video services on-demand.
Desirable video on-demand services can include, for example, movies,
sporting events, interactive games, home shopping, textual information,
and educational and arts programs, hereinafter collectively referred to as
"videos". It should be understood that videos generally includes both
video and audio portions, although, a video may only have an image portion
as in textual information, or only an audio portion, as for example music.
Consumers would like videos of their choice to be available at times and
locations convenient for them. It would be an advantage if the videos
could be delivered by any transmission medium, such as commercial
telephone, cable, and satellite networks. The videos should be compatible
with readily available display systems, such as NTSC standard televisions,
or personal computers.
Furthermore, consumers would like to have real-time, interactive VCR-like
control of the videos, skipping, holding, or replaying portions at will.
Home shopping and video games require a much higher level of interaction
between the video and the game-player. In addition, the system used to
provide video on-demand services should be scaleable at a reasonable cost
to maximize the availability of the service to large populations of
consumers.
A major problem with known distribution systems for video services, such as
broadcast or cable television is that the consumer has no control over
program or time selection. Also, known video distribution systems are
limited by predetermined channel allocations in the number of different
videos that are available at any one time. And, most known video services
are generally not interactive.
Some video services allow for interactive videos by using specialized
high-speed fiber optic cable networks. Generally such services deliver the
video to the customer as one continuous uninterruptable stream. Therefore,
such services generally require expensive mass storage devices and special
play-back equipment at the consumer's location to provide interaction.
Alternative such services require duplicate system resources at the
distribution site for each of the customers, even if multiple customers
are viewing the same video.
SUMMARY OF THE INVENTION
In view of the foregoing, it should be apparent that there still exists a
need in the art for a method and system for interactively viewing videos
without a substantial incremental costs. In a system for viewing videos
interactively, a selected videos includes a plurality of frames. Each
frame including digitally encoded video data, the video data representing
a time sequenced image and audio signals for play-back on a viewing
device, the video data are transmitted over a communications network.
The system at the receiving end of the communications network includes a
memory buffer for storing a segment of a selected video. The segment
includes a predetermined time interval of the selected video. In addition,
the memory buffer including a write pointer and a read pointer.
Video data received over the communications network are written to the
memory buffer at memory locations indicated by the write pointer while
advancing the write pointer. Means, such as a remote controller, or
provided to interactively position the read pointer to any of the video
data stored in memory buffer. Play-back of the selected video commences by
reading the video data from the memory buffer at memory locations
indicated by the read pointer while advancing the read pointer as the
video data are read.
As an advantage, a customers can view different portions of the selected
video, even though only a single relatively small sized memory is required
for storing the segment. For example, if the memory buffer stores about 10
minutes of the video data, typically about 60 to 100 Megabytes, the
customer can interactively and independently view any portion of the 10.
In the preferred embodiment the memory buffer is a disk storage device
managed as a circular buffer. Initially, when the buffer does not contain
any video data to be transferred, the video data are written at a higher
rate than the rate at which the video are read for play-back. Thus, the
memory buffer is rapidly filled, so that requests to view different
portions of the video can be fulfilled soon after transfer of the video is
initiated.
In addition, an index to the memory buffer is included. Entries to the
index include the time stamps of the frames stored in the memory buffer.
The time stamps are associated with memory addresses of the frames when
the frames are stored in the memory buffer. Thus, the viewer can select a
particular portion of the video by time, and the index enables the rapid
positioning of the read pointer to the particular frame by associating the
time stamp with a memory address.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other features and advantages of the present invention will
become apparent from reading of the detailed description in conjunction
with the attached drawings, wherein:
FIG. 1 is a high-level schematic view of a system for providing videos
on-demand services;
FIG. 2 is a block diagram of a video on-demand system;
FIG. 3 is a block diagram of a video;
FIG. 4 is a block diagram of a procedure to request a video;
FIG. 5 is a block diagram of a procedure to transfer a video;
FIG. 6 is a timing diagram of a segmented video;
FIG. 7 is a block diagram of a segment cache;
FIG. 8 is a block diagram of a server segment cache;
FIGS. 9 and 10 are block schematics of the segment cache operation;
FIG. 11 is a block diagram of a procedure for managing segmented broadcast
streams; and
FIG. 12 is a block diagram of a customer segment cache.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
Referring to FIG. 1, there is shown a system for providing videos
on-demand. The system includes customer premises equipment (CPE) 10
located at sites distributed over a large geographical area, one or more
centralized video on-demand systems 20, and a communications network 30.
Generally, videos are transferred from the video on-demand systems 20 to
the CPE 10 over the network 30. However, videos can also be transferred
among the video on-demand systems 20.
The physical medium used for communicating between the video on-demand
system 20 and the CPE 10 can be, for example, a twisted pair of wires, a
co-axial or fiber optic cable, or a micro-wave or satellite link. The
continuous physical path used for communicating is called a circuit. The
circuit includes a relatively small bandwidth "bi-directional" channel for
communicating control signals, and a relatively large bandwidth
"downstream" channel for communicating video signals. The circuit may also
include additional channels, for example, a medium bandwidth channel may
be allocated for two-way communications such as plain old telephone
service.
The configuration of the CPE 10 at each customer location can vary with the
needs of the customer. The CPE 10 includes a network interface box 11, a
viewing device 12, and a video controller 13. The interface box can
include an optional customer segment cache 14.
The interface box 11 is for receiving videos to play-back on the viewing
device 12 via the downstream channel. The interface box 11 also
communicates commands with the video on-demand system 20 via the
bi-directional channel as will be explained hereinafter. Customer commands
can be demands for video services. While a video is being transferred to
the CPE 10, customer commands can include VCR-like control functions, such
as reverse, forward, and pause, generally not available for known
broadcast or cable-TV services. Moreover, customer commands for more
sophisticated functions, such as jump backward or jump forward, to skip
over entire portions of the video are also supported. Furthermore,
specialized customer commands for controlling highly interactive videos,
such as games or home shopping, are possible.
The viewing device 12 can be, for example, a monitor, a television, a VCR,
or data processing equipment such as a personal computer or a
work-station. The video controller 13 for entering customer commands can
be configured as a hand-held remote controller for communicating with the
interface box 11 by using radio or infrared signals. Alternatively, the
video controller 13 can be a telephone capable of generating audible tones
by pressing the dialing buttons. The optional customer segment cache 14,
which will be described in greater detail hereinafter, is for locally
storing portions of videos received by the CPE 10.
FIG. 2 is a more detailed depiction of one centralized video on-demand
system 20. The video on-demand system 20 includes a gateway server 21, a
video server 22, a library server 23, and an optional server segment cache
24. The video on-demand system 20 also includes a plurality of
communications ports 80 for interfacing with the network 30.
The gateway, video, and library servers 21-23 are connected to each other
for process control by a control bus 26. The video server 22 and the
library server 23 are connected to each other by a data bus 27. The
servers 21-23 can be implemented as workstations, each workstation having
disk and semiconductor memory for executing distributed software programs.
The optional server segment cache 24, which will be described in greater
detail hereinafter, is for locally storing portions of videos to be
transmitted to the CPE 10.
The gateway server 21 includes an interactive gateway unit (IGU) 31 and a
server management unit (SMU) 32. The IGU 31 is for communicating commands
with the CPE 10 over the network 30. The IGU 31 includes a customer
database 33 for containing administrative data identifying customers. The
SMU 32, in response to commands, coordinates the video server 22 and the
library server 23 to deliver the selected video to the CPE 10 for viewing
by the customer.
The library server 23 includes a bulk video storage system, such as a video
juke box 41. The juke box 41 can be in the form of high capacity disks or
tapes storing thousands of videos in encoded, compressed, and digitized
form. A typical two hour compressed VHS movie requires about 1.15
Gigabytes of storage. Of course videos having higher resolutions, such as
HDTV videos, may require greater amounts of storage. It should be
understood that videos can be acquired for storing on the juke box 41 via
the network 30 from, for example, another video on-demand system 20.
Each video server 22 includes a video server controller 51 and a disk
read/write controller 52 having disks 53. By using modern high capacity
disks, for example redundant arrays of inexpensive disks (RAID), a video
server 22 can store over 100 Gigabytes of video data, equivalent to some
one hundred full-length feature movies.
The video server 22 also includes one or more first-in, first-out (FIFO)
video buffers 60 made of, for example, semiconductor dynamic random access
memory (DRAM). Each buffer 60 is partitioned into a plurality of chunks
61. One chunk 61 for storing, for example, 64K bytes of video data. The
size of each FIFO buffer 60 is dynamically adjustable by allocating or
deallocating chunks 61 during operation.
Connected to each FIFO buffer 60 is one or more packet controllers 70. The
packet controllers 70 are for reading variable sized data packets from the
FIFO buffers 60.
The communications ports 80 provide the interface between the video
on-demand system 20 and the network 30. It should be understood that the
video on-demand system 20 can include ports 80 configured for
communicating signals using methods adapted for different types of
networks 30.
Two methods of communicating are used. In a first method, signals are
communicated over a fixed bandwidth, point-to-point circuit which is
established for the duration of the transfer between the source of the
video, the video on-demand system 20 and the destination, the CPE 10.
This method of communicating can be used, for example, with the world's
largest crossbar switch, the public telephone network carrying digital
broadband transmission. In the United States there are about 93 million
home phone connections, and about 43 million business phone connections
providing access to a substantial market for video on-demand services.
The telephone network uses T1-carriers having a bandwidth of 1.544 Megabits
per second (Mb/sec). T1-carriers can readily transfer compressed motion
picture videos. In North America, T2-carriers having a bandwidth of 6
Mb/sec are also available. And, T3-carriers, available at some locations,
have a bandwidth equivalent to 28 T1-carriers. In Europe and elsewhere,
E1-carriers support transmission rates of 2 Mb/sec.
Alternatively, in the United States, Bellcore's Asymetrical Digital
Subscriber Loop (ADSL) services permits limited distance, approximately 5
kilometers, communication of digital encoded signals at T1 speed using
ordinary twisted pair telephone lines. Higher data transfer rates are
possible over shorter distances. A typical ADSL circuit is generally
partitioned into three virtual channels, a 16 Kb/sec bi-directional
channel for carrying control signals between the CPE 10 and to the video
on-demand system 20, a 1.5 Mb/sec downstream channel for carrying video
data signals to the CPE 10, and a 64 Kb/sec two-way communications channel
for plain old telephone signals.
Alternatively, the dedicated point-to-point fixed bandwidth circuit can be
a channel, or a portion of a channel of a commercial cable-TV (CATV)
system. Channel here meaning a traditional 6 MHz broadcast TV channel. A
typical cable system can have a capacity equivalent to hundreds of
broadcast-TV channels. By using frequency and/or time division
multiplexing techniques, it is possible to partition a single traditional
6 MHz CATV channel into several sub-channels to provide two-way
communications and to increase the capacity of the CATV network.
In a second method, the video is transferred using a packet-switching wide
area, or local area network such as are used by computer systems in the
business sector. With this type of transfer, the communication signals are
transferred as packets between addresses on the network. In this case, the
network is shared by all potential sources (the video on-demand systems
20), and destinations (the CPE 10).
The transmission path of digital packet-switching networks typically has a
very high bandwidth and can accommodate burst transmission that are many
times that of the play-back speed of the compressed video. For example, an
Ethernet can accommodate digital signaling rates of 10 Mb/sec, and Fiber
Distributed Data Interconnect (FDDI) can accommodate rates of 100 Mb/sec.
Therefore, in these type of networks, many packets can be sent between
multiple sources and destinations over the same physical medium using time
division multiplexing techniques.
For example, a 10 Mb/sec Ethernet can be partitioned into six 1.5 Mb/sec
sub-channels. However, videos transferred over a full-capacity Ethernet
would probably appear to break-up or run at a slower speed, since a
minimum bandwidth can not always be guaranteed due to signaling
characteristics of the Ethernet such as collisions. Therefore, the network
has to be configured to operate at a lower than theoretical capacity in
order to sustain continuous video play-back speed.
Now with reference to FIG. 3 the internal data structure a video suitable
for transfer over the network 30 will be described. The video 100 is, for
example a 1.2 GB feature length movie, having a start and an end. The data
of the video 100 is in a form suitable for transfer over the network 30 by
using encoding and compressing techniques, for example, the industry
standard Motion Picture Expert Group (MPEG) compression algorithms. MPEG
compression can reduce the video data by as much as a factor of 200 while
achieving a quality comparable with known VCR videos.
The video 100 includes a plurality of packets 110, generally indicated by
the numerals 1-N. The number of packets in the video is dependent on the
"length" or viewing time of the video. Each packet 110 includes a packet
header 120, packet data 140, and an optional packet filler 160.
The packet header 120 includes an ID 121, a RATE 122, and a TIME-STAMP 123,
and a PACKET-SIZE 125. The ID 121 identifies the particular video. The
RATE 122 is determined by the level of compression that is used to
transform the analog signals of the video to digital data. MPEG supports
varying compression rates dependent on space, time, and motion factors of
the video. The TIME-STAMP 123 indicates the position of the packet data
140, time-wise, relative to the beginning of the video 100. The
PACKET-SIZE 125 is used to indicate the number of data bytes of the packet
110. The PACKET-SIZE 125 is used by software that manipulates the various
packets 110 to manage storage requirements.
If the video is being transferred over a packet-switched network having
addressable locations the packet header 120 also includes an ADDRESS 124
to determine the destination of the packet 110. The destination
corresponding to the geographical location of the CPE 10.
The packet data 140 of each packet 110 includes either digital video data
or digital audio data, but not both. In a typical video, there are about
seven "video" packets for every "audio" packet. The packet data 140
includes about 1 to 4 KB of encoded and compressed video or audio data.
The data in the consecutive packets 110 are treated as a continuous bit
stream defining the frames of the video. Typically, the frames 180 are
displayed on the viewing device at a fixed rate of, for example, 30 frames
per second.
Depending on the level of compression, a frame 180 can include 1 KB to 16
KB of digital data. In other words, for highly compressed portions of the
video 100, a single video packet 110 may contain several frames 180. For
images compressed to a lesser extent, several packet 110 may be required
to compose a single frame.
A frame 180 includes a time-stamp bit sequence 181, frame data bits 182,
and an end-of-frame bit sequence 183. Two types of time-stamps 181 are
used with MPEG compression, a program time-stamp and a temporal
time-stamp. A program time-stamp is the absolute time-wise position of the
frame 180 relative to the beginning of the video 100. A temporal
time-stamp indicates the time-wise offset of the frame 180 relative to a
previous frame. Typically every fourth frame will have program time-stamp.
The frequency of time stamping can be programmed at the time that the
video is encoded and compressed.
For MPEG, three types of frames are defined: an I-frame; a P-frame; and a
B-frame.
An I-frame is a "reference" frame depicting an image which is not dependent
on any other frame of the video. That is, the image of the I-frame is
wholly derived from the bits of the I-frame.
A P-frame is a "delta" frame representing an image derived from another
frame. The P-frame contains the bits of the image which are different than
the bits of the frame from which the image of the P-frame is to be
derived. The P-frame typically includes a temporal time-stamp indicating
the relative time-wise offset of the P-frame from the frame from which it
is derived.
A B-frame is an "interpolate" frame used to create images from previous
and/or following I or P frames. The B-frame can also include a temporal
time-stamp.
Typically an I-Frame is larger, that is, includes more data, than a
P-frame, which in turn is larger than a B-frame. The ratio of I-, to, P-
to B-frames in a particular video varies greatly, depending on the encoder
algorithm used, and the type of motion and detail of the images
represented by the frames.
The packet filler 160 is used to assure a constant bit transmission rate at
a predetermined bandwidth. For example, an MPEG compressed vide | | |
