Patent Application
20070166000
The present disclosure is generally related to trick mode streams in digital video recorders, and more specifically, to trick mode streams in distributed digital video recorder systems.
Many consumers receive entertainment programming in their homes from a cable television operator. Many of today's cable offerings are broadcast using digital signals, which make more efficient use of communication bandwidth, and thus allow more programming to be carried on the same cable. In these cable systems, video programming (e.g., television programs, movies, etc.) is encoded at the cable head-end using a Motion Pictures Experts Group (MPEG) standard. The programming is transmitted from the head-end to the customer premises over a cable. At the customer premises, a digital home communication terminal (DHCT) decodes the programming and generates an analog picture signal. The analog picture is displayed by a television connected to the DHCT.
Some of today's DHCT units incorporate digital video recorder (DVR) functionality, which allows the DHCT to record video programming in digital form. These DHCTs can decode and display a video program in real-time from the head-end, or can decode and display a recorded program. A variation on the basic DHCT DVR is a distributed DVR system: a network of DHCT units, with one DHCT acting as the recorder and another acting as the player.
Popular DVR features include the ability to fast-forward, rewind, and pause a recorded program. These features are sometimes referred to as “trick modes.” Implementing trick modes often includes displaying frames at a faster or slower rate. Some trick modes also involve selecting only a subset of frames to decode and display. For example, fast-forwardx2 may choose frames 500 ms apart, and display them every 250 ms, while fast-forwardx4 may choose frames 500 ms apart and displaying them every 1000 ms. Pause can be implemented by decoding and displaying the same frame repeatedly.
There are several problems associated with implementing trick modes in a distributed DVR system. Video frames in the MPEG stream contain presentation timestamps that tell the decoder when to display a particular frame and/or decoder timestamps that tell the decoder when to decode a particular frame. The clock reference for these timestamps is provided by a program clock reference (PCR) that is also embedded in the MPEG stream. When the DVR records a video program, it records the stream as sent by the head-end, including the timestamps and the PCR. When this same stream is sent to the player DHCT in a trick mode, the clock reference provided by the PCR is incomplete, since many frames are skipped. Thus, the decoder in the player DHCT cannot rely on a clock recreated from the received PCR. Furthermore, even if the PCR was good, not every video frame in the MPEG stream has a timestamp. For at least these reasons, the distributed DVR system cannot use recorded PCR and timestamps for decoder timing.
Another approach is for the player DVR to generate a timestamp similar to a local PCR, and to transmit this timestamp in the stream. However, on many DHCTs, the recorded stream is stored in encrypted form, so that altering it requires first decryption and then re-encryption. The computational power required for this makes this approach infeasible.
In addition to these problems related to timing, there are other problems as well. When a state transition occurs from, for example, fast-forward mode to play mode, the decoder must know at exactly which frame this transition occurs. Otherwise, improper decoding will occur and the user will see artifacts. In a typical integrated DVR, this is easily accomplished. Since the video source resides in the same unit as the decoder, communication between the two occurs at relatively high bus speeds. Thus, the source can receive state change information back from the decoder before any further frames are communicated to the decoder. In contrast, communication between the recorder (source) and the player (decoder) in a distributed DVR system takes place at network speeds, which are much slower than bus speeds. This communication mechanism is often not fast enough to accurately communicate state transitions to the decoder.
In one embodiment, among others, the method supports trick modes in a distributed DVR system by inserting trick mode control packets into the MPEG stream sent from the recorder to the player. The decoder in the player relies on instructions in these control packets to determine the time at which frames are decoded rather than using timestamps in the originally received stream. Specifically, these instructions tell the decoder how many bytes should be buffered before decoding begins, and when a buffer of frames should be displayed. In one embodiment, the recorder sends a trick mode control packet, followed by the selected I-frame, followed by the picture header of the next frame, and then the sequence repeats with the next selected I-frame. In some embodiments, the control packet also contains additional information such as: ignore timestamps beginning with this frame; disable audio beginning with this frame; current trick mode (normal play, fast-forwardx2, fast-forwardx4, rewind, slow-motion, etc.); and whether the current stream contains I-frames or not.
The standard used by head-end 110 is the MPEG-2 standard, which describes how video and audio are compressed and coded to produce elementary streams. The MPEG-2 standard also describes how the elementary streams are multiplexed, transmitted, and demultiplexed, and how synchronization is achieved between elementary streams. Head-end 110 transmits multiplexed MPEG streams containing video, audio, and/or data to customer premises 120 over communication channel 130 (in the “downstream” direction). Typically, the downstream communication channel 130A contains one or more RF channels or frequencies, and each of these RF channels carries one MPEG transport stream. The MPEG transport stream is multiplexed to carry multiple elementary streams. For simplicity, the remainder of this discussion will discuss a single RF channel carrying an MPEG transport stream.
Customer premises 120 contains at least two DHCTs, 140A and 140B. In one embodiment, a DHCT is a standalone, integrated unit. In another embodiment, a DHCT is integrated into another consumer device, such as a television, among others. Server DHCT 140A receives streams over downstream channel 130A. Server DHCT 140A transmits commands, responses, and data to head-end 110 over upstream communication channel 130B. Client DHCT 140B is not in direct communication with head-end 110, but is coupled to server DHCT 140A via home network 150. Remote control 160 allows users to control one or more of the DHCT units.
Server DHCT 140A has the capability to record MPEG transport streams received from head-end 110 onto a storage medium 170. Server DHCT 140A can also transmit a recorded stream to client DHCT 140B over channel 150A, which is part of network 150. Channel 150A is also used to communicate commands and responses to client DHCT 140B. An MPEG decoder in client DHCT 140B decodes the stream and provides it to television 180 for display. Client DHCT 140B can transmit commands, responses, and status information to server DHCT 140A over channel 150B, which is part of network 150. In one embodiment, server DHCT 140A also includes a MPEG decoder and provides it to an attached television (not shown).
Network interface 260 is an interface for transmitting and/or receiving data to/from another DHCT. Network interface 260 may comprise, for example, an Ethernet interface, an IEEE-1394 interface, a USB (Universal Serial Bus) interface, a serial interface, a parallel interface, a wireless radio frequency (RF) interface, a telephone line interface, a power line interface, a coaxial cable interface, and/or an infrared (IR) interface, among others. In one possible implementation, the network interface 260 is an Ethernet interface, which is coupled one or more DHCTs via an Ethernet hub.
Memory 270, which may include volatile and/or non-volatile memory, stores one or more programmed software applications, herein referred to as applications, which contain instructions that may be executed by processor 220 under the direction of operating system 275. Input data used by an application is stored in memory 270 and read by processor 220 as needed during the course of the application's execution. This input data may be data stored in memory 270 by a secondary application or other source, either internal or external to DHCT 140, or may be data that was created with the application at the time it was generated as a software application program. Data transmitted by head-end 110 may be received via communications interface 210, whereas user input may be received from an input device via input system 250. Data generated by an application is stored in memory 270 by processor 220 during the course of the application's execution. Availability, location, and amount of data generated by one application for consumption by another application are communicated by messages through the services of operating system 280.
A navigator application 280 provides a navigation framework for services provided by DHCT 140. Navigator 280 registers for and in some cases reserves certain user inputs related to navigational keys such as channel increment/decrement, last channel, favorite channel, etc. Navigator 280 also provides users with television related menu options that correspond to DHCT functions such as, for example, providing an interactive program guide, blocking a channel or a group of channels from being displayed in a channel menu, and displaying a purchase list for a video-on-demand service.
Under user instruction, DVR application 285 records and/or and plays back received programs. DVR application 285 includes trick mode logic 290. When this system is used to implement server DHCT 140A, trick mode logic 290 creates a trick mode stream corresponding to the recorded stream, and provides it to client DHCT 140B using network interface 260. When this system is used to implement client DHCT 140B, trick mode logic 290 decodes and display the received trick mode stream.
Applications, such as navigator 280 and DVR 285, utilize services provided by window manager 295 and other graphics utilities provided by operating system 275 to draw menus, graphics, etc. for display on television 180. Window manager 295, which in one embodiment is part of operating system 275, contains functionality for allocating screen areas and managing screen use among multiple applications.
Applications executed by DHCT 140 comprise executable instructions for implementing logical functions. The applications can be embodied in any computer-readable medium for use by or in connection with an instruction execution system. The instruction execution system may be, for example, a computer-based system, a processor-containing system, or any other system capable of executing instructions. In the context of this document, a “computer-readable medium” can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
The computer-readable medium can be, for example, but is not limited to, an electronic, solid-state, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium, either internal to DHCT 140 or externally connected to DHCT 140 via one or more communication ports or network interfaces. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a hard drive storage device (magnetic), a random access memory (RAM) (solid-state device), a read-only memory (ROM) (solid-state device), an erasable programmable read-only memory (EPROM or Flash memory) (multiple devices), an optical fiber (optical), and a portable compact disc read-only memory (CDROM) (optical). Note that the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
The programming streams provided to DHCT 140 by head-end 110 preferably follow the MPEG-2 set of standards. This set of standards describes how video and audio are compressed and coded to produce elementary streams. The MPEG-2 standards also describe how the elementary streams are multiplexed, transmitted, and demultiplexed, and how synchronization is achieved between elementary streams.
The MPEG-2 standards use the three different picture types shown in
The arrows in
Although the word “frame” is used, an MPEG frame or picture is not equivalent to what is normally called a “movie frame” or “video frame,” since it may not contain enough information to decode and display an entire image. A single I-frame plus some number of P-frames and/or B-frames forms a group of pictures, or GOP, which is guaranteed to contain the information to properly decode and display an image. Thus, a GOP is equivalent to a “movie frame.” GOPs can be arranged in a sequence. All pictures in a sequence have the same picture size, aspect ratio, and frame rate.
Before a video stream is transmitted from a video source such as head-end 110, the frames are first packetized. This process is shown in
The MPEG standard allows the data portion of a PES packet to contain MPEG units described above, where each element begins with a start code. The example embodiments discussed here all contain one picture per PES packet. Video stream 400 contains two sequences, 430A and 430B, each beginning with a sequence start code (450A and 450B). Both sequences contain one I-frame and one B-frame. The I-frame of the first sequence 430 starts with picture start code 460A, followed by I-data 470. This I-frame is immediately followed by another B-frame, which starts with picture start code 460B, followed by B1-data 490.
There is no sequence end code. Instead, the start of the second sequence 430B is marked by a second sequence start code (450B). The format of the two frames in this sequence is the same as in the first sequence: picture start code followed by data.
The large size of PES packets makes them suitable for storage media, but not as suitable for transmission over error-prone communication channels. For transmission, PES packets are first segmented and then encapsulated into relatively small fixed-size packets called Transport Stream (TS) packets.
Server DHCT 140A receives an MPEG transport stream like the one of
When in trick mode, server DHCT 140A selects a subset of the TS packets in stored transport stream 600 for transmission to client DHCT 140B. (The selection process will be discussed in more detail later.) Picture start list 610 stored on storage medium 170 allows server DHCT 140A to efficiently locate TS packets containing picture frames. Picture start list 610 is a list of all MPEG pictures in the stored transport stream 600, along with a reference to the TS packet containing the start code for that picture. For example, stored transport stream 600 contains five MPEG pictures: I1; B1; I2; P1; I14; and I15. Picture start list 610 therefore contains the following entries: I1, starts in TS packet 530A; B1 starts in TS packet 530D; I2 starts in TS packet 530F; P1 starts in TS packet 530n; I14 starts in TS packet 540; and I15 starts in TS packet 550.
Picture start list 610 is useful when the recorded TS packets are encrypted and server DHCT 140A cannot examine the TS payload contents to look for picture start codes. In this case, picture start list 610 is created as stored transport stream 600 is recorded, which involves temporarily decrypting the TS packets. In another embodiment, stored transport stream 600 is not encrypted, and a separate picture start list 610 is unnecessary because server DHCT 140A can scan the TS payloads for picture start codes.
To implement trick mode, trick mode logic 290 in server DHCT 140A creates a trick mode stream corresponding to a recorded stream, by inserting trick mode command packets into the stream transmitted to client DHCT 140B. Trick mode logic 290 in server DHCT 140A also paces the transmission of picture frames according to the selected trick mode. For example, when the mode is fast-forwardx2, twice as many pictures frames are transmitted in the same time period, compared to normal play. Trick mode logic 290 in client DHCT 140B relies on instructions in these command packets to determine the time at which frames are decoded rather than using timestamps in the received stream. These instructions tell the decoder how many bytes should be buffered before decoding begins, and when a decode frame should be displayed. In some embodiments, the command packet also specifies additional information such as: ignore timestamps beginning with this frame; disable audio beginning with this frame; current trick mode (normal play, fast-forwardx2, fast-forwardx4, rewind, slow-motion, etc.); and whether the current stream contains I-frames or not.
On receipt of the trick mode request, server DHCT 140A determines the correct start position within the recorded stream. This is accomplished by first determining, in step 710, whether it is currently transmitting the recorded stream to client DHCT 140B. If No, the process continues at step 715, where the start position is initialized to the first I-frame in the recorded stream. If Yes, the process continues at step 720, where the start position is initialized to the most recently transmitted I-frame in the recorded stream.
After the start position is initialized, server DHCT 140A selects a series of I-frames to be transmitted to client DHCT 140B. At step 725, the first I-frame is selected, based on the start position and the specific mode selected. For example, if the start position is the last frame in the recorded stream and the mode is rewindx2, then the first selected I-frame is the I-frame 500 ms before the last frame. In another example, if the start position is the first frame and the mode is slow-motion, then the first selected I-frame is the first I-frame in the recorded stream. Further details about the selection criteria used by an example embodiment of the system for generating trick mode streams are found in Table 1.
Once an I-frame is selected, server DHCT 140A continues to step 730 where a trick mode control packet is created and transmitted to client DHCT 140B. The control packet contains the total size of the transport packet payloads following the control packet, and a decoder command. The decoder command instructs the decoder in client DHCT 140B how to handle the picture frame once it has been received in the decoder's buffer. The decoder command depends on the characteristics of the recorded stream. If the recorded stream contains I-frames, then the decoder command is Decode_and_Display. If the recorded stream contains no I-frames, the command is Decode_Only. In one embodiment, the command Normal_Play is also supported. Normal_Play tells the decoder to decode the stream normally.
Server DHCT 140A then continues to step 735 and transmits all the TS packets that make up the selected I-frame. In one embodiment, server DHCT 140A uses picture start list 510 to find the TS packet containing the start of a particular picture frame. Note that there may be PES packets with a non-zero length field. The last such PES packet containing I-frame data may be incomplete as a result of the selection. For those PES packets, additional stuffing TS packets are added so that the PES packets are complete. At step 740, after transmitting the entire I-frame, server DHCT 140A transmits an additional number (M) of TS packets following the last TS packet containing the selected I-frame. These additional TS packets allow the decoder to skip the data of the next picture frame in a clean manner. (This feature will be described in connection with
Next, server DHCT 140A determines if processing of this trick mode stream is complete. At step 745, server DHCT 140A determines if either a trick mode stop request has been received, or the end of the recorded stream has been reached. If one of the these conditions is TRUE, trick mode stream processing ends, at step 750. If client DHCT 140B has requested server DHCT 140A to stop the trick mode stream, server DHCT 140A transmits a trick mode control packet indicating the end of the trick mode stream.
If neither condition is true, trick mode stream processing continues at step 755, where the next I-frame is selected based on the current position and the mode. Next, at step 760, server DHCT 140A waits for an appropriate time interval before transmitting the next trick mode control packet and selected picture frame. The time interval is based on the trick mode (see Table 1). Next, the transmission of trick mode control packet and selected picture frame is repeated, starting at step 730. One skilled in the art will realize that the order of selecting the next frame, creating the control packet, and waiting for transmission can be varied. For example, the wait can occur before the selection, or the control packet can be created before the selection.
In some embodiments, control packet 810 also contains a trick string describing the trick mode. Examples of trick mode strings include “FFWD”, “PAUS”, “SRWD” (step rewind), “SLMF” (slow motion forward) and “END” (end of trick mode stream). In some embodiments, control packet 810 also contains a time code, which indicates the time at which server DHCT 140A sent the control packet 810. This is a relative time, and is an indication of the speed of the trick mode stream. For example, in a fast-forwardx2 stream, the time code for successive trick mode control packets would be 0, 500, 1000, 1500 (in milliseconds).
The next TS packet in the trick mode transport stream 800, following control packet 810, is TS packet 530A. TS packet 530A contains the start of the first selected picture frame. Next in the transmitted stream is TS packet 530B, which continues the data of the selected picture frame. The selected picture frame ends with TS packet 530D. Server DHCT 140A then transmits some number (M) of TS packets following this end, enough so that the next picture header is transmitted. Next in the stream is another trick mode control packet 860, followed by the TS packet 870 containing the start of the next selected picture frame.
When a TS packet is received, the process continues at step 915. The contents of the TS packet are examined to determine if the TS packet contains a trick mode control packet 810. If Yes, client DHCT 140B acts in step 920 to extract decoder command 840 and picture size 850 from the received control packet 810. Next, at step 925, client DHCT 140B receives the next TS packet, and buffers the picture frame contained within this TS packet. After receiving the next TS packet, client DHCT 140B determines, at step 930, if all TS packets following this control packet have been received. This determination is made using the picture size 850 in control packet 810, received before the picture frame. If No, client DHCT 140B waits for the next TS packet at step 925.
When the entire picture frame has been received, processing continues at step 935, where the picture frame is decoded. Step 940 then examines decoder command 840 from received control packet 810. If decoder command 840 indicates that the frame should be displayed, this occurs in Step 945. In Step 950, any partial frame received in the accumulated TS packets is discarded. Because picture frames are not required to align on a TS packet boundary, the TS packet containing the end of an I-frame may also contain the start of the next frame (usually a B-frame). In order to discard this partial frame, client DHCT 140B must receive the entire picture header of the partial frame. Having the server DHCT 140A transmit a relatively large number of extra TS packets makes it almost certain that the entire picture header is received. It has been empirically determined that transmitting 10 extra TS packets gives acceptable results (although it is possible that a long header would extend past the tenth packet). Next, client DHCT 140B determines at step 955 if the user has changed modes, for example, from one trick mode to another, or from trick mode to normal play. If Yes, the process ends at step 960. Otherwise, client DHCT 140B returns to step 910 to wait for another TS packet, containing either another control packet 810 or another picture frame.
In the embodiments describe above, the TS packets selected for trick mode transport stream 800 contained I-frames. I-frames are preferred over B-frames or P-frames, since I-frames can be decoded independently. However, it is possible that stored transport stream 600 will not contain any I-frames. An alternative embodiment handles a non-I-frame recorded stream by first instructing the decoder to decode, but not display, a number of P-frames, and later sending a Decode_and_Display instruction. First, a trick mode control packet 810 is constructed with the Decode_Only instruction. This Decode_Only control packet is sent, followed by a complete P-frame. This Decode_Only+P-frame sequence is repeated a number of times (in one embodiment, 12 times).
After receiving this sequence of P-frames, the decoder has buffered and decoded all the P-frames, but has not yet displayed any frames. Next, a Decode_And_Display control packet is sent, followed by an additional complete P-frame. After receiving this additional P-frame, the decoder displays the decoded frames built from the entire sequence of P-frames.
The foregoing description has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Obvious modifications or variations are possible in light of the above teachings. The embodiments discussed, however, were chosen, and described to illustrate the principles of the disclosure and its practical application to thereby enable one of ordinary skill in the art to utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. All such modifications and variation are within the scope of the disclosure as determined by the appended claims when interpreted in accordance with the breadth to which they are fairly and legally entitled.
