Channel Bonding Synchronization

Abstract

Different data communication architectures deliver a wide variety of content, including audio and video content, to consumers. The architectures employ channel bonding to deliver more bandwidth than any single communication channel can carry. In some implementations, the communication architectures distribute data streams to bonded channels that are clocked independently. A system is provided for synchronizing the bonded channels.

Description

2. TECHNICAL FIELD

This disclosure relates to communication techniques. In particular, this disclosure relates to channel bonding synchronization.

3. BACKGROUND

Rapid advances in electronics and communication technologies, driven by immense private and public sector demand, have resulted in the widespread adoption of smart phones, personal computers, internet ready televisions and media players, and many other devices in every part of society, whether in homes, in business, or in government. These devices have the potential to consume significant amounts of audio and video content. At the same time, data networks have been developed that attempt to deliver the content to the devices in many different ways. Further improvements in the delivery of content to the devices will help continue to drive demand for not only the devices, but for the content delivery services that feed the devices.

BRIEF DESCRIPTION OF THE DRAWINGS

The innovation may be better understood with reference to the following drawings and description. In the figures, like reference numerals designate corresponding parts throughout the different views.

FIG. 1 shows an example of a content delivery architecture that employs channel bonding.

FIG. 2 shows an example of logic for content delivery using channel bonding.

FIG. 3 shows an example of a content delivery architecture that employs channel bonding.

FIG. 4 shows an example of logic for content delivery using channel bonding.

FIG. 5 shows a timing example.

FIG. 6 shows an example of a content delivery architecture that employs channel bonding.

FIG. 7 shows an example of logic for content delivery using channel bonding.

FIG. 8 shows an example implementation of a distributor.

FIG. 9 shows an example implementation of a collator.

FIG. 10 shows an example of a content delivery architecture that performs channel bonding below the transport layer, e.g., at the data-link layer.

FIG. 11 shows an example of channel bonding at the data-link layer.

FIG. 12 shows an example of channel bonding at the data-link layer.

FIG. 13 shows an example of logic that a data-link layer may implement for channel bonding at the data-link layer.

FIG. 14 shows an example of logic that a data-link layer may implement for channel debonding at the data-link layer.

FIG. 15 shows an example variant of the content delivery architecture of FIG. 6.

FIG. 16 shows an example variant of the content delivery architecture of FIG. 7.

FIG. 17 is a schematic diagram of a system for synchronizing channels in a channel bonding system.

FIG. 18 is a schematic diagram of another system for synchronizing channels in a channel bonding system.

FIG. 19 is a schematic diagram of another system for synchronizing channels in a channel bonding system.

FIG. 20 is a schematic diagram of yet another system for synchronizing channels in a channel bonding system.

DETAILED DESCRIPTION

FIG. 1 shows an example content delivery architecture 100. The architecture 100 delivers data (e.g., audio streams and video programs) from a source 102 to a destination 104. The source 102 may include satellite, cable, or other media providers, and may represent, for example, a head-end distribution center that delivers content to consumers. The source 102 may receive the data in the form of Motion Picture Expert Group 2 (MPEG2) Transport Stream (TS) packets 128, when the data is audio/visual programming, for example. The destination 104 may be a home, business, or other location, where, for example, a set top box processes the data sent by and received from the source 102.

The source 102 may include a statistical multiplexer 106 and a distributor 108. The statistical multiplexer 106 helps make data transmission efficient by reducing idle time in the source transport stream (STS) 110. In that regard, the statistical multiplexer 106 may interleave data from multiple input sources together to form the transport stream 110. For example, the statistical multiplexer 106 may allocate additional STS 110 bandwidth among high bit rate program channels and relatively less bandwidth among low bit rate program channels to provide the bandwidth needed to convey widely varying types of content at varying bit rates to the destination 104 at any desired quality level. Thus, the statistical multiplexer 106 very flexibly divides the bandwidth of the STS 110 among any number of input sources.

Several input sources are present in FIG. 1: Source 1, Source 2, . . . Source n. There may be any number of such input sources carrying any type of audio, video, or other type of data (e.g., web pages or file transfer data). Specific examples of source data include MPEG or MPEG2 TS packets for digital television (e.g., individual television programs or stations), and 4K×2K High Efficiency Video Coding (HVEC) video (e.g., H.265/MPEG-H) data, but the input sources may provide any type of input data. The source data (e.g., the MPEG 2 packets) may include program identifiers (PIDs) that indicate a specific program (e.g., which television station) to which the data in the packets belongs.

The STS 110 may have a data rate that exceeds the transport capability of any one or more communication links between the source 102 and the destination 104. For example, the STS 110 data rate may exceed the data rate supported by a particular cable communication channel exiting the source 102. To help deliver the aggregate bandwidth of the STS 110 to the destination 104, the source 102 includes a distributor 108 and modulators 130 that feed a bonded channel group 112 of multiple individual communication channels. In other words, the source 102 distributes the aggregate bandwidth of the STS 110 across multiple outgoing communication channels that form a bonded channel group 112, and that together provide the bandwidth for communicating the data in the STS 110 to the destination 104.

The distributor 108 may be implemented in hardware, software, or both. The distributor 108 may determine which data in the STS 110 to send on which communication channel. As will be explained in more detail below, the distributor 108 may divide the STS 110 into chunks of one or more packets. The chunks may vary in size over time, based on the communication channel that will carry the chunk, the program content in the chunk, or based on any other desired chunk decision factors implemented in the distributor 108. The distributor 108 may forward any particular chunk to the modulator for the channel that the distributor 108 has decided will convey that particular chunk to the destination 104.

In that regard, the multiple individual communication channels within the bonded channel group 112 provide an aggregate amount of bandwidth, which may be less than, equal to, or in excess of the aggregate bandwidth of the STS 110. As just one example, there may be three 30 Mbs physical cable channels running from the source 102 to the destination 104 that handle, in the aggregate, up to 90 Mbs. The communication channels in the bonded channel group 112 may be any type of communication channel, including dial-up (e.g., 56 Kbps) channels, ADSL or ADSL 2 channels, coaxial cable channels, wireless channels such as 802.11a/b/g/n channels or 60 GHz WiGig channels, Cable TV channels, WiMAX/IEEE 802.16 channels, Fiber optic, 10 Base T, 100 Base T, 1000 Base T, power lines, or other types of communication channels.

The bonded channel group 112 travels to the destination 104 over any number of transport mechanisms 114 suitable for the communication channels within the bonded channel group 112. The transport mechanisms 144 may include physical cabling (e.g., fiber optic or cable TV cabling), wireless connections (e.g., satellite, microwave connections, 802.11 a/b/g/n connections), or any combination of such connections.

At the destination 104, the bonded channel group 112 is input into individual channel demodulators 116. The channel demodulators 116 recover the data sent by the source 102 in each communication channel. A collator 118 collects the data recovered by the demodulators 116, and may create a destination transport stream (DTS) 120. The DTS 120 may be one or more streams of packets recovered from the individual communication channels as sequenced by the collator 118.

The destination 104 also includes a transport inbound processor (TIP) 122. The TIP 122 processes the DTS 120. For example, the TIP 122 may execute program identifier (PID) filtering for each channel independently of other channels. To that end, the TIP 122 may identify, select, and output packets from a selected program (e.g., a selected program ‘j’) that are present in the DTS 120, and drop or discard packets for other programs. In the example shown in FIG. 1, the TIP 122 has recovered program ‘j’, which corresponds to the program originally provided by Source 1. The TIP 122 provides the recovered program to any desired endpoints 124, such as televisions, laptops, mobile phones, and personal computers. The destination 104 may be a set top box, for example, and some or all of the demodulators 116, collator 118, and TIP 122 may be implemented as hardware, software, or both in the set top box.

The source 102 and the destination 104 may exchange configuration communications 126. The configuration communications 126 may travel over an out-of-band or in-band channel between the source 102 and the destination 104, for example in the same or a similar way as program channel guide information, and using any of the communication channel types identified above. One example of a configuration communication is a message from the source 102 to the destination 104 that conveys the parameters of the bonded channel group 112 to the destination 104. More specifically, the configuration communication 126 may specify the number of communication channels bonded together; identifiers of the bonded communication channels; the types of programs that the bonded communication channels will carry; marker packet format; chunk, program packet, or marker packet size; chunk, program packet, or marker packet PID or sequence number information, or any other chunk or bonding configuration information that facilitates processing of the bonded channel group 112 at the destination 104. One example of a configuration communication message from the destination 104 to the source 102 is a configuration communication that specifies the number of communication channels that the destination 104 may process as eligible bonded channels; identifiers of the eligible bonded channels; status information concerning status of the demodulators 116, e.g., that a demodulator is not functioning and that its corresponding communication channel should not be included in a bonded channel group; channel conditions that affect bit rate or bandwidth; or any other information that the source 102 and the distributor 108 may consider that affects processing of the data from the sources into a bonded channel group.

FIG. 2 shows an example of logic 200 for content delivery using channel bonding that the architecture 100 described above may implement in hardware, software, or both. Additional detailed examples are provided below, particularly with regard to marker packets and other options.

In FIG. 2, input sources receive program data (202). The program data may be received from any content provider, and may include any desired audio, visual, or data content, including cable television programming, streaming music, file transfer data, as just three examples. The input sources provide the program data to the statistical multiplexer 106 (204), which multiplexes the program data to generate the source transport stream (STS) 110 (206).

The source 102 provides the STS 110 to the distributor 108 (208). The distributor 108 reads bonding configuration parameters (210). The bonding configuration parameters may specify the number of communication channels in the bonded channel group 112, the communication channels that may be included in the bonded channel group 112, the type of communication channels that may be included in the bonding channel group 112, the program sources eligible for bonding, when and for how long communication channels and program sources are available for channel bonding, bonding adaptation criteria, and any other parameters that may influence how and when the distributor 108 pushes program data across the communication channels in the bonded channel group 112. The distributor 108 sends the program data to the communication channels in the bonded channel group 112 (212). Specific examples of how the distributor 108 accomplishes this are provided below. The source 102 thereby communicates program data to the destination 104 across the multiple communication channels in the bonded channel group 112 (214).

At the destination 104, the demodulators 116 receive the program data over the communication channels (218). The demodulators 116 provide the recovered program data (optionally after buffering) to the collator 118. The collator 118 analyzes group information, sequence information, PIDs, and any other desired information obtained from the data packets arriving on the communication channels and creates a destination transport stream (DTS) 120 from the recovered program data (220). The DTS 120 may convey the program packets in the same sequence as the STS 110, for example.

The collator 118 provides the DTS 120 to the TIP 122 (222). The TIP 122 reads data selection parameters (224). The data selection parameters may specify, for example, which audio/visual program is desired, and may be obtained from viewer input, from automated selection programs or processes (e.g., in a digital video recorder), or in other ways. Accordingly, the TIP 122 filters the DTS 120 to recover the program packets that match the data selection parameters (e.g., by PID filtering) (226). The TIP 122 thereby generates a content output that includes an output packet stream for the selected program. The TIP 122 delivers the generated content to any desired device 124 that consumes the content, such as televisions, smart phones, personal computers, or any other device.

Several channel bonding processing options are discussed next. Some options make reference to marker packets (MPs) inserted into the data streams going to the destination 104 over the communication channels. The marker packets may be MPEG2 TS packets, for example, with an identifier that flags them as MPs. In the first option, the distributor 108 adds marker packets on a per-channel basis, for example in a round-robin manner. In the second option, the distributor 108 generates and adds markers on a per-chunk basis, for example in a round-robin manner at chunk boundaries. In the third option, when packets from the same program will be routed to multiple communication channels, each packet receives a program ID and a sequence ID, and no marker packets are needed. In the fourth option, spare bits in network frames defined below the network layer, e.g., at the data-link layer, carry channel bonding information to the source 104.

Regarding the first option, FIG. 3 shows another example of a content delivery architecture 300 that employs channel bonding. In the architecture 300, a marker packet (MP) source 302 feeds MPs to the statistical multiplexer 106. The MP source 302 may provide marker packets at any frequency. For example, the MP source 302 may provide a marker packet for each communication channel in the bonded channel group 112 for every ‘n’ non-marker packets received from the sources, every ‘k’ ms, or at some other time or packet spacing frequency. The time or packet spacing, ‘n’ or ‘k’ may take any desired value, e.g., from n=1 packet to tens of thousands of packets, or k=1 ms to 1 second. In other implementations, the distributor 108 generates the MPs, rather than receiving them in the STS 110.

Fewer marker packets consume less channel bandwidth, leaving more room for program data. However, more marker packets increase the ability of the destination 104 to adapt to changes in the program data, including allowing the collator 108 to more quickly synchronize the multiple data streams across the bonded channel group 112, allowing faster program channel changes through the TIP 122, and facilitating faster adaptation to changes in the configuration of the bonded channel group 112. Marker packet insertion may vary depending on any desired parameters. Examples of such parameters include available buffer sizes; target, average, or worst case recovery time for recovering from transmission errors or other transmission issues; target program channel change latency or other types of latency; and target program frame size.

In FIG. 3, the distributor 108 pushes packets to the modulators 130 on a round-robin basis, starting with any desired modulator 130. More specifically, the distributor 108 may communicate packets on a round-robin basis to each communication channel in the bonded channel group 112, one packet at a time. In other implementations described below, the round-robin distribution may be done n-packets at a time, where ‘n’ is greater than 1. However, for the example shown in FIG. 3, the distributor 108 pushes one packet at a time in a round-robin manner across the communication channels that compose the bonded channel group 112. Accordingly, given the example STS 110 packet stream of {MP-0, MP-1, MP-2, 1-0, 1-1, 2-1, 2-2, n-0}, the distributor 108 pushes:

MP-0 to channel 1, MP-1 to channel 2, MP-2 to channel 3; then

pkt 1-0 to channel 1, pkt 1-1 to channel 2, pkt 2-1 to channel 3; then

pkt 2-2 to channel 1, pkt n-0 to channel 2, and so on.

The MP source 302 may provide MPs to the statistical multiplexer 106 at a selected priority level, such as a highest available priority level, or a higher priority level than any other packets arriving from the program sources. Furthermore, the number of MPs in a set of MPs may match the number of communication channels in the bonded channel group 112. For example, when there are seven (7) communication channels in the bonded channel group 112, the MP source 302 may provide seven highest priority MPs to the statistical multiplexer 106. The statistical multiplexer 106 may then output the high priority MPs immediately next in the STS 110, so that the group of seven MPs arrive in sequence at the distributor 108. As a result of the packet-by-packet round-robin distribution, one of each of the seven marker packets correctly is pushed to one of the seven communication channels in the bonded channel group 112 to flag a stream of program packets that follow each MP.

The statistical multiplexer 106 or the distributor 108 or the MP source 302 may give the MPs a special identifier, such as a unique PID (e.g., MARKER PID) that flags the MPs as marker packets. Any other desired content may be present in the MPs. As examples, the MPs may include a channel number and group number. The channel number may identify the communication channel that sent that MP (e.g., channel 0, 1, or 2 for a bonded channel group 112 of three communication channels). The channel number provides a type of sequence number that identifies, the first, second, third, and so on, communication channel in sequence to which the distributor 108 has sent program packets. The channel number, in other words, identifies a bonded channel sequence of distribution of program packets to the communication channels in the bonded channel group.

The group number may identify which set of MPs any particular MP belongs to, and the source 102 may increment the group number with each new set of MPs (e.g., every three MPs when there are three communication channels in the bonded channel group 112). The group number may also facilitate packet alignment, when, for example, jitter or skew is larger than the gap between inserted packets.

Note that the distributor 108 need not have any special knowledge of the MPs. Instead, the distributor 108 may push packets on a round-robin basis to the communication channels, without knowing or understanding what types of packets it is sending. However, in other implementations, the distributor 108 may in fact analyze and manipulate the packets that it distributes, to insert or modify fields in the MPS, for example. Additionally, the distributor 108 may generate the MPs, rather than receiving them in the STS 110.

The destination 104 processes the marker packets, and may align packets in a fixed order from the demodulators 116 to form the DTS 120. The destination 104 may include First In First Out (FIFO) buffers 304, or other types of memory, to counter jitter/skew on the communication channels, and the resultant mis-alignment in reception of packets across the various communication channels. The FIFOs 304 may be part of the collator 118 or may be implemented separately. A FIFO may be provided for each communication channel, to provide a set of parallel buffers on the receive side, for example.

At the destination 104, the collator 118 may drop all packets before a MP from each channel is received. The collator 118 checks the group number of the marker packet in each channel, and drops packets until the collator 118 has found marker packets with matching group numbers on each communication channel. When the group numbers do not match, this may be an indicator to the collator 118 that the skew is larger than the gap between marker packets.

The channel number in the marker packets specifies the sequence of communication channels from which the collator 118 will obtain packets. The collator 118 obtains packets in a round-robin manner that matches the round-robin distribution at the source 102. In an example with three communication channels in the bonded channel group 112, the collator 118 may start by obtaining a packet from the communication channel carrying MP sequence number zero, then moving to the communication channel carrying MP sequence number one and obtaining a packet, then moving to the communication channel carrying MP sequence number two and obtaining a packet, then back to the sequence number zero communication channel in a round-robin manner. The collator 118 thereby produces a DTS 120 that corresponds to the STS 110. The TIP 122 may then extract the selected program from the DTS 120.

FIG. 4 shows an example of logic 400 for content delivery using channel bonding that may be implemented in hardware, software, or both in the example architecture 300 described above. Input sources receive program data (402), and in addition, a MP source 302 may provide MPs (404). The input sources and MP source provide the program data and the MPs to the statistical multiplexer 106 (406), which multiplexes the program data and MPs to generate the source transport stream (STS) 110 (408).

In particular, the MPs may have a high priority, so that the statistical multiplexer 106 inserts them into the STS 110 sequentially without gaps before other program data packets. The STS 110 is provided to the distributor 108 (410). The distributor 108 reads bonding configuration parameters (412). The bonding configuration parameters may specify that the distributor 108 should take the round-robin distribution approach, and may specify round-robin distribution parameters. Examples of such parameters include the round-robin distribution chunk size, e.g., ‘r’ packets at a time per communication channel (e.g., ‘r’=1), in what situations the distributor 108 should execute the round-robin technique, or any other round-robin parameter. As noted above, the bonding configuration parameters may also specify the number of communication channels in the bonding channel group 112, the communication channels that may be included in the bonding channel group 112, the type of communication channels that may be included in the bonding channel group 112, the program sources eligible for bonding, when and for how long communication channels and program sources are available for channel bonding, and any other parameters that may influence how the distributor 108 pushes program data across the communication channels in the bonded channel group 112.

The distributor 108 pushes the program data to the communication channels in the bonded channel group 112 (414). The source 102 thereby communicates program data to the destination 104 across the multiple communication channels in the bonded channel group 112 (416).

More particularly, the distributor 108 may push the program packets to the communication channels in round-robin manner. In one implementation, the round-robin approach is a one packet at a time approach. In other words, the distributor 108 may take each packet (when ‘r’=1) from the STS 110 and push it to the next communication channel in sequence. As such, the in-order sequence of MPs from the STS 110 is distributed one MP per communication channel, and is followed by one or more program packets. The MPs thereby effectively flag for the destination 104 the program packets that follow the MPs. After a predetermined number of program packets, the MP source provides another group of MPs that are then distributed across the communication channels, and the cycle repeats.

At the destination 104, the demodulators 116 receive the program data over the communication channels (420). The demodulators 116 provide the recovered program data to buffers (e.g., the FIFOs 304) to help address jitter/skew (422) on the communication channels. The buffered data is provided to the collator 118, which may pull packets from the buffers to synchronize on MPs. The collator 118 analyzes group information, sequence information, PIDs, and any other desired information obtained from the MPs and program packets to synchronize on MPs. The synchronization may include finding sequential MPs of the same group number across each communication channel in the bonded channel group 112. The collator 118 may then create a destination transport stream (DTS) 120 from the recovered program data (424) by adding packets to the DTS 120 in a round-robin manner across the communication channels in the bonded channel group 112, going in order specified by the channel numbers specified in the MPs. The DTS 120 may convey the program packets in the same sequence as the STS 110, for example.

The collator 118 provides the DTS 120 to the TIP 122 (426). The TIP 122 reads channel selection parameters (428). The channel selection parameters may specify, for example, which program is desired, and may be obtained from viewer input, from automated selection programs or processes (e.g., in a digital video recorder), or in other ways. Accordingly, the TIP 122 filters the DTS 120 to recover the program packets that match the channel selection parameters (e.g., by PID filtering) (430). The TIP 122 thereby generates a content output that includes an output packet stream for the selected program. The TIP 122 delivers the generated content to any desired device 124 that consumes the content, such as televisions, smart phones, personal computers, or any other device.

FIG. 5 shows a timing example 500 which shows that in some implementations, the source 102 may address transmit clock variations in the modulators 130. FIG. 5 shows transmit buffers 502, each of which may provide some predetermined depth, such as a depth at least that of the channel timing variation (e.g., 200 ms). As one example, the communication channels may be expected to have the same nominal payload rates, e.g., 38.71 Mb/s. Further, assume that the transmit clock in each modulator is independent, and can vary by plus or minus 200 ppm.

Thus, in the worst case, two channels in a bonded channel group may have a clock difference of 400 ppm. As shown in the example in FIG. 5, the timing different from channel 1 to channel 2 is 200 ppm, and the timing difference between channel 1 and channel ‘m’ is 400 ppm. The timing difference of 400 ppm may amount to as much as one 188 byte MPEG2 TS packet every 2500 outgoing packets.

Accordingly, the source 102 may insert a compensation packet (which may have NULL content) on channel ‘m’ every 2500 packets to cover the extra outgoing packet, and also insert a compensation packet on channel 2 every 5000 packets for the same reason. The compensation packet may appear, for example, just prior to the MP, or anywhere else in the outgoing data stream. The destination 104 may identify and discard compensation packets (or any other type of jitter/skew compensation packet).

The source 102 may implement a buffer feedback 504. The buffer feedback 504 informs the distributor 108 about buffer depths in the transmit buffers 502. When the buffers run empty, or at other times, the distributor 108 may insert compensation packets, e.g., before MPs.

FIG. 6 shows another example of a content delivery architecture 600 that employs channel bonding. In this second option, the architecture 600 includes a distributor 108 that sends data over the communication channels in communication units called chunks (but any other term may refer to the communication units). The chunks may include one or more packets from any of the program sources. For example, a chunk may be 1 packet, 10 packets, 100 packets, 27 packets, 10,000 packets, 100 ms of packets, 20 ms of packets, 30 ms of video data, 5 s of audio data, or any other number or timing of packets or audio/visual content.

The distributor 108 may use the same or different chunk size for any of the communication channels. Furthermore, the distributor 108 may change the chunk size at any time, in response to an analysis of any desired chunk size criteria. One example of a chunk size criteria is desired channel change speed at the destination 104. As the number of packets in a chunk increases, the destination 104 may need to drop more packets before reaching the next chunk boundary, finding the matching MPs, and being able to synchronize to the received communication channels. The chunk size may also depend on compressed video rate or frame size, as well as target, average, or worst case recovery time for recovering from transmission errors or other transmission issues.

In the example in FIG. 6, the statistical multiplexer 106 receives program packets from input sources 1 . . . ‘n’. The program packets may be MPEG2 TS packets, or any other type of packet. The statistical multiplexer 106 creates a STS 110 from the program packets, and the STS 110 therefore has a particular sequence of packets multiplexed into the STS 110 from the various input sources according to the statistical properties of the program streams.

For the purposes of illustration, FIG. 6 shows the first six chunks that the distributor 108 has decided to send over the communication channels. In particular, the first three chunks are two-packet chunks 602, 604, and 606. The next two chunks are one-packet chunks 608 and 610. The next chunk is a two-packet chunk 612.

The distributor 108 generates MPs that precede the chunks. Alternatives are possible, however, and some are described below with respect to FIGS. 15 and 16. The distributor 108 may communicate the MPs and the chunks (e.g., in a round-robin manner) across the communication channels. In the example of FIG. 6, the distributor 108 sends a MP (e.g., MP-0, MP-1, and MP-2) to each communication channel, followed by a two-packet chunk behind MP-0, MP-1, and MP-2, in round-robin sequence: channel 1, channel 2, channel m, and then returning to channel 1. The distributor 108 may start the sequence with any particular communication channel.

As is shown in FIG. 6, the communication channels receive MPs and chunks in round-robin manner starting with channel 1 as follows:

Channel 1: MP-0; Channel 2: MP-1; Channel 3: MP-2

Channel 1: chunk 602; Channel 2: chunk 604; Channel 3: chunk 606

Channel 1: MP-4; Channel 2: MP-5; Channel 3: MP-6

Channel 1: chunk 608; Channel 2: chunk 610; Channel 3: chunk 612

Because chunk boundaries are marked with MPs, the distributor 108 may insert compensation packets (e.g., NULL packets) without affecting the channel bonding. In other words, each communication channel may have its own unique payload rate. Furthermore, MPEG2 TS corruption during transmission does not affect other packets.

Each MP may include a channel number and a group number, as described above. The channel and group numbers may take a wide variety of forms, and in general provide sequence indicators. Take the example where the chunk size is 100 packets and there are three communication channels A, B, and C, with the distributor 108 proceeding in this order: C, B, A, C, B, A, . . . . The first set of MPs that come before the first 100 packet chunks may each specify group number zero. Within group zero, the first MP on communication channel C has a channel number of zero, the second MP on communication channel B has a channel number of one, and the third MP on the communication channel A has a channel number of two. For the next group of chunks of 100 packets, the MP group number for the next three MPs may increment to one, and the channel numbers run from zero to two again.

At the destination 104, the demodulators 116 receive the MPs and chunks from each communication channel. Again, individual FIFOs 204 may be provided to help compensate for jitter and skew.

The collator 118 receives the MPs, and synchronizes on the received data streams when the collator 118 finds MPs of the same group number and in sequence across the communication channels that are part of the bonded channel group 112. Once the collator 118 has synchronized, it obtains each chunk following the MPs in order of group number and channel number. In this manner, the collator 118 constructs the DTS 120 that corresponds to the STS 110. As described above, the TIP 122 executes PID filtering on the MPEG2 TS packets to recover any desired program j, and may discard the other packets.

FIG. 7 shows an example of logic 700 for content delivery using channel bonding, that may be implemented in hardware or software in the example architecture 600 described above. Input sources receive program data (702). The input sources provide the program data to the statistical multiplexer 106 (704), which multiplexes the program data to generate the source transport stream (STS) 110 (706). The distributor 108 receives the STS 110 (708).

The distributor 108 also reads bonding configuration parameters (710). The bonding configuration parameters may, for example, specify that the distributor 108 should take the round-robin distribution approach, and may specify round-robin distribution parameters. Examples of such parameters include the round-robin distribution chunk size, e.g., ‘r’ packets at a time per communication channel (e.g., r=100), chunk size per communication channel, or chunk size variation in time, or variation depending on chunk size factors that the source 102 may monitor and adapt to over time, in what situations the distributor 108 should execute the round-robin technique, or any other round-robin parameter. As noted above, the bonding configuration parameters may also specify the number of communication channels in the bonding channel group 112, the communication channels that may be included in the bonding channel group 112, the type of communication channels that may be included in the bonding channel group 112, the program sources eligible for bonding, when and for how long communication channels and program sources are available for channel bonding, and any other parameters that may influence how the distributor 108 pushes program data across the communication channels in the bonded channel group 112.

In this option, the distributor generates MPs (712) for the chunks of program packets that the distributor sends through the individual communication channels in the bonded channel group 112. Thus, for example, when the bonding configuration parameters indicate a chunk size of 100 packets, the distributor generates a MP for each 100 program packets communicated down the communication channel. As was explained above, a MP may include synchronization data, such as a group number and channel number. As another example, the MP may include timing data such as a timestamp, time code, or other timing reference measurement.

The distributor 108 sends the MPs and the program data to the communication channels in the bonded channel group 112 (714). The distributor 108 may send the MPs and program data in a round-robin manner by communication units of program packets (e.g., by chunks of program packets). The source 102 thereby communicates program data to the destination 104 across the multiple communication channels in the bonded channel group 112 (716).

More particularly, the distributor 108 may send the program packets to the communication channels in round-robin manner by chunk. In other words, the distributor 108 may take chunks of program packets from the STS 110 and send them to the next communication channel in the bonded channel group 112 in a predetermined round-robin sequence (e.g., as specified in the bonding configuration parameters). As such, an MP is distributed to a communication channel, and is followed by a chunk of program packets tagged by the MP in terms of group number and channel number. The program packets include PID information that identifies the program to which each packet belongs. The MPs thereby effectively flag for the destination 104 the program packets that follow the MPs. After each chunk of program packets, the distributor 108 provides another group of MPs that are then distributed across the communication channels, and the cycle repeats. The chunk size may vary in time and by communication channel. Furthermore, the source 102 may send configuration communications to the destination 104 to advise the destination 104 of the bonding configuration and changes to the bonding configuration, including chunk size.

At the destination 104, the demodulators 116 receive the program data over the communication channels (718). The demodulators 116 provide the recovered program data to buffers (e.g., the FIFOs 304) to help address jitter/skew (720) on the communication channels. The buffered data is provided to the collator 118, which may pull packets from the buffers to synchronize on MPs. The collator 118 analyzes group information, sequence information, PIDs, and any other desired information obtained from the MPs and program packets to synchronize on MPs. The synchronization may include finding sequential MPs of the same group number across each communication channel in the bonded channel group 112.

The collator 118 then creates a destination transport stream (DTS) 120 from the recovered program data (722) by adding packets to the DTS 120 in a round-robin manner across the communication channels in the bonded channel group 112. In particular, the collator 118 adds packets to the DTS 120 by chunk of program packets in a round-robin manner across the communication channels in the bonded channel group 112. Thus, the DTS 120 may convey the program packets to the TIP 122 in the same sequence as they were present in the STS 110, for example.

The collator 118 provides the DTS 120 to the TIP 122 (724), which reads channel selection parameters (726). The channel selection parameters may specify, for example, which program is desired, and may be obtained from viewer input, from automated selection programs or processes (e.g., in a smart phone content recording application), or in other ways. Accordingly, the TIP 122 filters the DTS 120 to recover the program packets that match the channel selection parameters (e.g., by PID filtering) (728). The TIP 122 thereby generates a content output that includes an output packet stream for the selected program. The TIP 122 delivers the generated content to any desired device 124 that consumes the content, such as televisions, smart phones, personal computers, or any other device.

Turning briefly to FIG. 15, that figure shows an example variation architecture 1500 of the content delivery architecture 600 in FIG. 6. In one variation, the distributor may instead issue MP generation signals (e.g., the MP generation signals 1502, 1504, 1506) to the modulators 130. The MP generation signal 1502 may be a command message, signal line, or other input that causes the receiving modulator to generate a MP for insertion into the packet stream, e.g., at chunk boundaries. The MP may include any desired synchronization information, including time stamps, time codes, group numbers, channel numbers, and the like. The modulator may generate the synchronization information, or the distributor 108 may provide the synchronization information to the modulator along with the MP generation signal.

In another variation, both the distributor 108 generates MPs and the modulators 130 generate MPs. For example, the distributor 108 may generate the MPs for the modulator for CH2 and send MP generation signals to the modulators for the other channels. Another alternative is for the distributor 108 to generate MPs for some modulators some of the time, and to send MP generation signals to those modulators at other times. Whether or not the distributor 108 generates the MPs may depend on MP capability information available to the distributor 108. For example, the bonding configuration parameters 710 may specify which modulators are capable of generating MPs, when, and under what conditions. Then, the distributor 108 may send the MP generation signal to those modulators at the corresponding times or under the corresponding conditions. Further, the modulator may communicate with the distributor 108 to specify MP generation capabilities, and the conditions on those capabilities, such as when and under what conditions the modulator can generate MPs, and also what information the modulator needs from the distributor 108 to generate the MPs.

Turning briefly to FIG. 16, that figure shows content delivery logic 1600 for the architectures described above. FIG. 16 shows again that, in the architectures described above (e.g., 1500 and 600), the distributor 108 may generate MPs, the modulators 130 may generate MPs, or both may generate MPs. For example, FIG. 16 shows that for the modulator for CH1, the distributor 108 generates the MPs (1602), e.g., at chunk boundaries. The distributor 108 also generates the MPs for the modulator for CHm (1606). However, for the modulator for CH2, the distributor 108 sends an MP generation signal and any desired synchronization information (1604) to the modulator for CH2. Accordingly, the modulator for CH2 generates its own MPs for the chunks it receives from the distributor 108. Note also that any modulator may communicate with the distributor 108 to specify MP generation capabilities, and the conditions on those capabilities, including when and under what conditions the modulator can generate MPs, as well as what information the modulator needs from the distributor 108 to generate the MPs (1608).

Turning now to FIG. 8, the figure shows an example implementation of a distributor 800. The distributor 108 includes an STS input interface 802, system logic 804, and a user interface 806. In addition, the distributor 800 includes modulator output interfaces, such as those labeled 808, 810, and 812. The STS input interface 802 may be a high bandwidth (e.g., optical fiber) input interface, for example. The modulator output interfaces 808-812 feed data to the modulators that drive data over the communication channels. The modulator output interfaces 808-812 may be serial or parallel bus interfaces, as examples.

The system logic 804 implements in hardware, software, or both, any of the logic described in connection with the operation of the distributor 108 (e.g., with respect to FIGS. 1-7 and 10). As one example, the system logic 804 may include one or more processors 814 and program and data memories 816. The program and data memories 816 hold, for example, packet distribution instructions 818 and the bonding configuration parameters 820.

The processors 814 execute the packet distribution instructions 818, and the bonding configuration parameters 820 inform the processor as to the type of channel bonding the processors 814 will perform. As a result, the processors 814 may implement the round-robin packet by packet distribution or round-robin chunk by chunk distribution described above, including MP generation, or any other channel bonding distribution pattern. The distributor 800 may accept input from the user interface 806 to change, view, add, or delete any of the bonding configuration parameters 820 or any channel bonding status information.

FIG. 9 shows an example implementation of a collator 900. The distributor 108 includes a DTS output interface 902, system logic 904, and a user interface 906. In addition, the collator 900 includes demodulator input interfaces, such as those labeled 908, 910, and 912. The DTS output interface 902 may be a high bandwidth (e.g., optical fiber) output interface to the TIP 122, for example. The demodulator output interfaces 908-912 feed data to the collator system logic which will create the DTS 120 from the data received from the demodulator input interfaces 908-912. The demodulator input interfaces 908-912 may be serial or parallel bus interfaces, as examples.

The system logic 904 implements in hardware, software, or both, any of the logic described in connection with the operation of the collator 118 (e.g., with respect to FIGS. 1-7 and 10). As one example, the system logic 904 may include one or more processors 914 and program and data memories 916. The program and data memories 916 hold, for example, packet recovery instructions 918 and the bonding configuration parameters 920.

The processors 914 execute the packet recovery instructions 918, and the bonding configuration parameters 920 inform the processor as to the type of channel bonding the processors 914 will handle. As a result, the processors 914 may implement the round-robin packet by packet reception or round-robin chunk by chunk reception described above, including MP synchronization, or any other channel bonding distribution recovery logic. The collator 900 may accept input from the user interface 906 to change, view, add, or delete any of the bonding configuration parameters 920, to specify which channels are eligible for channel bonding, or to set, view, or change any other channel bonding status information.

The architectures described above may also include network nodes between the source 102 and the destination 104. The network nodes may be type of packet switch, router, hub, or other data traffic handling logic. The network nodes may be aware of the communication channels that they are connected to, both on the inbound side, and on the outbound side. Accordingly, a network node may receive any particular set of communication channels in a channel bonding group, but need not have a matching set of communication channels in the outbound direction. In that case, the network node may filter the received communication channel traffic, to drop packets for which the network node does not have a corresponding outbound communication channel, while passing on the remaining traffic flow over the outbound communication channels to which it does have a connection.

In concert with the above, the channel bonding may happen in a broadcast, multicast, or even a unicast environment. In the broadcast environment, the source 102 may send the program packets and MPs to every endpoint attached to the communication channels, such as in a wide distribution home cable service. In a multicast environment, however, the source 102 may deliver the program packets and MPs to a specific group of endpoints connected to the communication channels. In this regard, the source 102 may include addressing information, such as Internet Protocol (IP) addresses or Ethernet addresses, in the packets to specifically identify the intended recipients. In the unicast environment, the source 102 may use addressing information to send the program packets and the MPs across the bonded channel group 112 to a single destination.

A third option is to add, at the source 102, channel bonding data fields to the program packets. The channel bonding data fields may be added to the packet header, payload, or both. The channel bonding data fields may identify for the destination 104 how to order received packets to create the DTS 120. In that regard, the channel bonding data fields may include PID information, sequence information, channel number information, group number information, or other data that the collator 118 may analyze to determine packet output order in the DTS 120.

In some implementations, a communication head-end may define the program packets that each source will employ, and therefore has the flexibility to create channel bonding fields in the program packets. In other implementations, the source 102 inserts channel bonding data into existing packet definitions (possibly using part of a conventional data field for this new purpose). For example, in some implementations, each program is formed from multiple MPEG2 PIDs, with each MPEG2 TS packet being 188 bytes in size. When packets from the same program will be routed across different communication channels, the source 102 may use header or payload fields in the MPEG2 TS packets to carry channel bonding fields (e.g., PID and sequence number) in the MPEG2 TS packets.

As one example, the source 102 may add, as channel bonding data, a program ID (PID) and sequence number to program packets. The PID may be a O-bit field that identifies one of 16 different programs. The sequence number may be a 12-bit field that identifies one of 4096 sequence values. In this implementation, the source 102 need not send MPs. Instead, the channel bonding information (e.g., PID and sequence number) inserted into the program packets provides the destination 104 with the information it uses to construct the DTS 120. More specially, the collator 118 identifies the PIDs and the packets with sequential sequence numbers for each PID, and creates the DTS 120 with the correct packet sequence.

Furthermore, the source 102 may also insert the channel bonding data into lower layer packets. For example, instead of (or in addition to for redundancy), sending MPs defined at or above the transport layer, the source 102 may instead insert the channel bonding data into frames defined below the transport layer, such as data-link layer frames or physical layer frames. As an example, the data-link layer frames may be Low Density Parity Check (LDPC) frames, and the physical layer frames may be Forward Error Correcting (FEC) frames.

Because such frames are defined at the data-link layer, higher layers may have no knowledge of these frames or their formats, and generally do not process such frames. Nevertheless, the higher level layers, including the transport layer, may provide bonding information to the data-link layer that facilitates data-link layer handling of the channel bonding data. Examples of such bonding information includes the amount and type of channel bonding data desired, including the definitions, sizes, and sequence numbering of channel number and sequence number fields, desired chunk size, number, identification and type of communication channels to bond, or any other channel bonding information.

In more detail, in some communication architectures, the data-link layer packets have spare, reserved, or otherwise ancillary bits. Instead of having the ancillary bit fields remain unused, the system 102 may insert the channel bonding data in those ancillary bit fields. In other implementations, the data-link layer may define its own particular packet format that includes bit fields specifically allocated for channel bonding data.

FIG. 10 shows an example of a content delivery architecture 1000 that performs channel bonding below the transport layer, e.g., at the data-link layer or physical layer. FIG. 10 extends the example of FIG. 6 for the purposes of discussion, but channel bonding at lower layers may occur in any content delivery architecture. In FIG. 10, a protocol stack at the distributor 108 includes multiple layers, including a Physical (PHY) layer 1002, a data-link layer 1004, a transport layer 1006, and any other layers desired 1008. The protocol stack may adhere to the Open Systems Interconnection (OSI) model, as one example, and the data-link layer and physical layer structures that may carry channel bonding information may include, as examples, Forward Error Correcting (FEC) frames, PHY frames, MAC frames, Low Density Parity Check (LDPC) frames, or IP Datagram frames. However any other protocol stack and structure types may instead be in place to handle channel bonding at a level below the level at which the program packets.

The STS 110 provides the program packets to the distributor 108. The protocol stack handles the program packets. In particular, the data-link layer 1004 constructs low level frames that encapsulate program packets and channel bonding data, and that are sent across the communication channels in the bonded channel group 112. One example of the low level frames is the data-link frame 1010. In this example, the data-link frame 1010 includes channel bonding (CB) data, data-link frame (DLF) data, and program packets (in particular, the first chunk 602). The DLF data may include the information fields in an already defined data-link layer packet format. The CB data may include channel number and group number information, or any other information that a MP might otherwise carry.

Higher level layers may (e.g., the transport layer 1006 or other layers 1008), as noted above, provide guidance to the data-link layer 1004 regarding what bonding information to include in the data-link layer frames. However, this is not required. The data-link layer may do its own analysis and makes its own decisions concerning what channel bonding data to add into the data-link layer frames. In that regard, the data-link layer may read the channel bonding configuration parameters. The data-link layer may also exchange the configuration communications 126 with the destination 104, including configuration communications 126 with the data-link layer, transport layer, or other layers at the destination 104.

At the destination 104, a protocol stack 1012 processes the data received from the demodulators 116. In particular, the protocol stack 1012 may include a data-link layer 1014. The data-link layer 1014 receives the data-link layer frames (e.g., the frame 1010) to extract the program packets and channel bonding data. The collator 118 may then process the channel bonding data as described above to synchronize the communication channels in the bonded channel group 112 and build the DTS 120.

FIG. 11 shows an example of channel bonding using data-link layer frames 1100. In FIG. 11, a data stream 1102 represents, for example, source data prior to packetization. The data stream 1102 may be as examples, data generated by a video camera, microphone, or bytes in a file on a disk drive. A content provider generates a packetized stream 1104, for example in the form of MPEG2 TS packets 1106. The packets 1106 may take many different forms, and in the example shown in FIG. 11, the packets 1106 include Cyclic Redundancy Check (CRC) data 1108 (e.g., in a header), and a payload 1110.

FIG. 11 also shows the data-link layer frames 1112. In this example, the data-link layer frames 1112 include a header 1114 and a payload 1116. The header 1114 may include fields in which, although they are pre-defined for other purposes, the data-link layer 1004 inserts channel bonding data, such as channel number and group number. FIG. 11 shows an example in which the data-link layer frame 1112 includes an MATYPE field 1118 (e.g., 2 bytes), a UPL field 1120 (e.g., 2 bytes), a DFL field 1122 (e.g., 2 bytes), a SYNC field 1124 (e.g., 1 byte), a SYNCD field 1126 (e.g., 2 bytes), and a CRC field 1128 (e.g., 1 byte). This particular frame format is further described in the DVB S2 coding and modulation standard, In particular, the data-link layer 1104 may insert the channel bonding information into the MATYPE field 1118.

The framing of the data-link layer 112 is such that program packets are generally encapsulated into the payload 1116 of the data-link frame 1112, while the channel bonding information is added to the header 1204. However, note the packetized stream 1104 does not necessarily line up with the data-link layer frames 1112. This is shown by the dashed lines in FIG. 11, with the data-link layer frame 1112 breaking across program packets. The lack of alignment may be due to timing and packet size mismatches between various layers in the protocol stack, and because the data stream 1102 does not necessarily adhere to any fixed timing parameters or data formats.

In some implementations, the architectures may facilitate alignment by inserting packets (e.g., NULL packets) of any desired length, padding program packets (e.g., with NULL data), truncating program packets (or otherwise dropping program packet data), dropping program packets altogether, or in other ways. The data-link layer 1004 may execute the alignment in order to fit an integer number of program packets into a data-link layer frame. In some implementations, the data-link layer 1004 may communicate with other layers in the protocol stack, or other logic in the source 102, to provide guidance on timing, alignment, chunk sizes, or other bonding parameters that may facilitate alignment and channel bonding at the data-link layer.

FIG. 12 shows an example of channel bonding using data-link layer frames 1200. As with FIG. 11, in FIG. 12 a data stream 1102 represents, for example, source data prior to packetization, and the packetized data stream 1104 arises from the data stream 1102. The data-link layer frame 1202 includes a header 1204 and a payload 1206. However, in FIG. 12, data-link layer frame 1202 has been designed to include fields specifically for channel bonding information. In the example in FIG. 12, the header 1204 includes the channel bonding field 11208 and the channel bonding field 21210. Other header fields 1212 carry other header information. Any number and length of channel bonding fields may be present in either headers or payload fields in the data-link layer frames to hold any desired channel bonding information.

FIG. 13 shows an example of logic 1300 that a data-link layer in the source 102 may implement for channel bonding at the data-link layer. The data-link layer may provide feedback to higher layers (1302). The feedback may inform the higher level layers about alignment, timing, or other considerations that affect how program packets break across or fit into data-link layer packets.

The data-link layer receives program packets from the higher level layers (1304). If the data-link layer will force alignment, then it may pad program packets, insert alignment packets, or even drop packets or parts of packets, so that the program packets fit within the data-link layer frame (1306) in a way that corresponds to the selected channel bonding configuration, including, for example, the chunk size. The data-link layer inserts channel bonding information into data-link layer frames (1308). In some implementations, the protocol stack at the source 102 does not generate separate marker packets for the channel bonding information. That is, the low level communication frames (e.g., the data-link layer frames) carry the channel bonding information in specific fields defined in the communication frames, so that no separate encapsulation of the channel bonding information (into marker packets, for example), is needed. Expressed yet another way, the data-link layer frames may have one less layer of encapsulation, e.g., encapsulating the channel bonding information directly into the low level communication frame, rather than multiple levels of encapsulation, e.g., encapsulating the channel bonding information first into a MP defined, e.g., at the same protocol level as a program packet, and then the MP into the communication frame.

The data-link layer also inserts program packets or chunks of packets into data-link layer frames. For example, the program packets may exist in the payload field of the data-link layer frames. The marker information may specify which packets are present in the data-link layer frame with the marker information (1310). The data-link layer then transmits the data-link layer frames over a communication channel that is part of a bonded channel group 112.

FIG. 14 shows an example of logic 1400 that a data-link layer in the source 102 may implement for channel debonding at the data-link layer. The data-link layer receives data-link layer frames (1402). The data-link layer extracts the program packets and the channel bonding information from the data-link layer frames (1404). Any padding data in the program frames, or padding packets may be discarded (1406).

The destination 104 analyzes the channel bonding information to synchronize across multiple communication channels, as described above (1408). Accordingly, for example, the destination may align to channel bonding sequence information across multiple communication channels. Once synchronized, the destination 104 may construct the DTS 120, for example by round-robin adding chunks to the DTS 120 from the data-link layer frames, informed by the channel bonding information in the data-link layer frames (1410).

One example format for a MP is the following:

CBM_PID: ChannelBondingMarker PID, which may be a reserved PID value for a marker packet. In some implementations, MPs may include adaptation layer information and follow the MPEG2 TS packet structure, although some or all of the content of the packet will be specific to MP data instead of, e.g., program data. The bytes in the MP may be assigned as follows (as just one example):

Byte #1: 0x47 (MPEG2 TS pre-defined sync byte)

Byte #2/3: CBM_PID+TEI=0, PUSI=0, priority=1

Byte #4: SC=′b00, AFC=′b11 (no payload), CC=0x0

Byte #5: Adaptation_length=′d183

Byte #6: Flags=0x02, e.g., only private data is present

Byte #7: Private data_length=′d181

Byte #8: Number of channels in Channel Bonding group

Byte #9/10: CBM_Sequence_Number (CBM_SN)

Byte #11/12/13/14: CBM_SIZE

This generic MPEG2 TS packet syntax is further explained in ISO/IEC 13818-1, section 2.4.3.2, “Transport Stream packet layer”.

Some implementations may facilitate bonding channels that are clocked independently. Often each modulator may be driven by a separate clock. Each clock can vary based on manufacturing tolerances or even system design. In certain implementations, advantages may be obtained through synchronizing bonded channels to reduce jitter or delay at the receiving device.

Several channel bonding processing options have been discussed herein. In the first option, the distributor 108 adds marker packets on a per-channel basis, for example in a round-robin manner. In the second option, the distributor 108 generates and adds markers on a per-chunk basis, for example in a round-robin manner at chunk boundaries. In the third option, when packets from the same program will be routed to multiple communication channels, each packet receives a program ID and a sequence ID, and no marker packets are needed. In the fourth option, spare bits in network frames defined below the network layer, e.g., at the data-link layer, carry channel bonding information to the source 104. Further other options may exist. However, any of the architectures or features of these techniques may be used together in conjunction with the discussed implementations for synchronizing bonded channels.

FIG. 17 is a schematic diagram of a system 1700 for synchronizing channels in a channel bonding system. Synchronizing the channels may be particularly helpful when the distributor is distributing packets using the round robin technique described above. A data stream 1712 is provided to a distributor 1710. The distributor is connected to multiple communication channels 1720, 1722, 1724 for transmitting the data stream to an end device 1748.

The distributor 1710 may assign the packets of the data stream 1712 to channels and may add marker packets as described elsewhere in this disclosure. In some implementations, the marker packets may be assigned to communication unit (e.g. chunk). The distributor 1710 then distributes packets to a modulator 1720, 1722, 1724 for each communication channel. In some implementations, the marker packets may be assigned to a communication unit, which are then distributed to the modulators 1720, 1722, 1724.

Each modulator 1720, 1722, 1724 has a corresponding buffer 1730, 1732, and 1734. The buffers 1730, 1732, and 1734 hold the packets and/or communication units provided to the communication channel until they can be transmitted to the receiving device 1748. In addition, each channel 1720, 1722, 1724 are driven by corresponding clocks 1740, 1742, 1744. Each clock 1740, 1742, 1744 may be independent and, therefore, may vary slightly due to manufacturing variability or even system design.

In some implementations, the clock 1740 may be slightly faster than clock 1742. As such, the buffer 1730 may fill up faster than the buffer 1732. The modulator 1720 may be configured to monitor the buffer depth (e.g. the fullness of the buffer) and send a signal to the distributor 1710 indicating that the buffer depth has exceeded a predefined buffer level. The distributor may then adjust processing to synchronize the channels.

In some implementations, modulator 1720 may send a signal 1760 to the distributor 1710 indicating a value of the buffer depth. In this scenario, modulator 1722 and modulator 1724 may also send signals 1762, 1764 to the distributor 1710 indicating a value of the buffer depth. The distributor 1710 may then compare the buffer depth of each buffer 1730, 1732, 1734 and adjust processing based on the comparison.

In this manner, the signals 1760, 1762, 1764 may operate as a feedback signal from the modulator for synchronizing the communication channels. The distributor 1710 may receive the feedback signal in streaming fashion or at regular intervals. Alternatively, the distributor 1710 may request the feedback signal from each modulator 1720, 1722, 1724, as needed.

The distributor 1710 may adjust the processing in many ways to synchronize the communication channels. Some of the ways in which the distributor 1710 may adjust the processing are noted below. Further, the distributor 1710 may utilize the below described methods in any combination to synchronize the communication channels.

In some implementations, the distributor 1710 may send an extra packet to a modulator that is running faster (e.g. has a lower buffer depth) thereby resynchronizing the data. The extra packet may be a junk packet. The junk packet may include null data or other identifiable data that will be automatically filtered by the receiver device 1748. Alternatively, the extra packet may include data from the data stream. The extra packet may be a duplicate packet that is also sent over other communication channels. In certain implementations, the extra packet may be identified as a high priority packet that would benefit from redundant transmission over multiple communication channels. The extra packet may be sent on a marker boundary (e.g. just before or just after a marker packet).

The distributor 1710 may also modify the communication unit size for each channel to adjust the processing based on the feedback signals. If the modulator is running faster than other modulators (e.g. the buffer has a lower buffer depth) the distributor 1710 may increase the size of the communication provided to that communication channel. The distributor 1710 may increase the size of the communication unit by adding packets to the communication unit. With regard to adding packets and/or communication units the slowest communication channel will often dictate adding data to each of the other channels.

If the modulator is running slower than other modulators (e.g. the buffer has a higher buffer depth) the distributor 1710 may decrease the size of the communication unit provided to that communication channel. The distributor 1710 may decrease the size of the communication unit by removing packets from the communication unit.

In some implementations, distributor 1710 may send a redundant communication unit to one or more of the other modulator 1722, 1724 thereby resynchronizing the channels. In this scenario, the distributor 1710 may identify communication units that include high priority data from the data stream. In some implementations, the distributor 1710 may skip sending a communication unit or packet (e.g. in a round robin) to a slower communication channel, although this may result in some lost data.

In some implementations, the distributor 1710 may adjust the clock speed to synchronize the communication channels. The distributor 1710 may increase the clock speed for slower communication channels. Further, the distributor 1710 may decrease the clock speed for faster communication channels.

FIG. 18 is a schematic diagram of a system 1800 for synchronizing channels in a channel bonding system. In system 1800, the modulator 1720 monitors the fullness of the buffer 1730. The modulator 1720 may include logic or circuitry to adjust the clock speed of the clock 1740 driving the modulator 1720. The modulator 1720 may decrease the clock speed, if the buffer 1730 exceeds an upper buffer fullness threshold. Further, the modulator 1720 may increase the clock speed, if the buffer 1730 is below a lower buffer fullness threshold.

In a similar manner, the modulators 1722 and 1724 monitor the fullness of buffers 1732 and 1734, respectively. The modulator 1722 may include logic or circuitry to adjust the clock speed of clock 1742 driving the modulator 1722. Modulator 1724 may also include logic or circuitry to adjust the clock speed of clock 1744 driving the modulator 1742. The modulators may decrease the clock speed, if the corresponding buffer exceeds an upper buffer fullness threshold. Further, the modulators may increase the clock speed, if the corresponding buffer is below a lower buffer fullness threshold.

FIG. 19 is a schematic diagram of a system 1900 for synchronizing channels in a channel bonding system. In system 1900, modulator 1720 may send a signal 1760 to the distributor 1710 indicating a value of the buffer depth. In this scenario, modulator 1722 and modulator 1724 may also send signals 1762, 1764 to the distributor 1710 indicating a value of the buffer depth. The distributor 1710 may then compare the buffer depth of each buffer 1730, 1732, 1734 and adjust processing based on the comparison.

In this manner, the signals 1760, 1762, 1764 may operate as a feedback signal from the modulator for synchronizing the communication channels. The distributor 1710 may receive the feedback signal in streaming fashion or at regular intervals. Alternatively, the distributor 1710 may request the feedback signal from each modulator 1720, 1722, 1724, as needed.

In addition, the modulator 1720 may also include logic or circuitry to adjust the clock speed of the clock 1740 driving the modulator 1720. The modulator 1720 may decrease the clock speed, if the buffer 1730 exceeds an upper buffer fullness threshold. Further, the modulator 1720 may increase the clock speed, if the buffer 1730 is below a lower buffer fullness threshold.

In a similar manner, the modulators 1722 and 1724 monitor the fullness of buffers 1732 and 1734, respectively. The modulator 1722 may include logic or circuitry to adjust the clock speed of clock 1742 driving the modulator 1722. Modulator 1724 may also include logic or circuitry to adjust the clock speed of clock 1744 driving the modulator 1724. The modulators may decrease the clock speed, if the corresponding buffer exceeds an upper buffer fullness threshold. Further, the modulators may increase the clock speed, if the corresponding buffer is below a lower buffer fullness threshold.

The distributor 1710 and modulators 1720, 1722, 1724 may work in combination to adjust the processing in many ways to synchronize the communication channels. Some of the ways in which the distributor 1710 may adjust the processing are noted below. Further, the below described methods may be used in any combination to synchronize the communication channels.

In some implementations, the distributor 1710 may send an extra packet to a modulator that is running faster (e.g. has a lower buffer depth) thereby resynchronizing the data. The extra packet may be a junk packet. The junk packet may include null data or other identifiable data that will be automatically filtered by the receiver device 1748. Alternatively, the extra packet may include data from the data stream. The extra packet may be a duplicate packet that is also sent over other communication channels. In certain implementations, the extra packet may be identified as a high priority packet that would benefit from redundant transmission over multiple communication channels. The extra packet may be send on a marker boundary.

The distributor 1710 may also modify the communication unit size for each channel to adjust the processing based on the feedback signals. If the modulator is running faster than other modulators (e.g. the buffer has a lower buffer depth) the distributor 1710 may increase the size of the communication provided to that communication channel. The distributor 1710 may increase the size of the communication unit by adding packets to the communication unit. With regard to adding packets and/or communication units the slowest communication channel will often dictate adding data to each of the other channels.

If the modulator is running slower than other modulators (e.g. the buffer has a higher buffer depth) the distributor 1710 may decrease the size of the communication unit provided to that communication channel. The distributor 1710 may decrease the size of the communication unit by removing packets from the communication unit.

In some implementations, distributor 1710 may send a redundant communication unit to one or more of the other channels 1722, 1724 thereby resynchronizing the channels. In this scenario, the distributor 1710 may identify communication units that include high priority data from the data stream. In some implementations, the distributor 1710 may skip sending a communication unit or packet (e.g. in a round robin) to a slower communication channel.

In some implementations, the distributor 1710 may adjust the clock speed to increase the clock speed for slower communication channels and/or decrease the clock speed for faster communication channels. The modulator may also include logic or circuitry to increase or decrease the clock speed directly based on the buffer depth (e.g. fullness of the buffer).

FIG. 20 is a schematic diagram of a system 2000 for synchronizing channels in a channel bonding system. In the system 2000, the distributor 1710 is configured to receive a signal from the receiving device 1748 indicative of the channel speed for one or more of the communication channels. The receiving device 1748 may have a buffer for each communication channel. The receiving device 1748 may monitor buffer fullness of each buffer in the receiving device and generate a signal 2010 that is transmitted to the distributor 1710.

The distributor 1710 may adjust the processing in many ways to synchronize the communication channels based on the signal 2010. Some of the ways in which the distributor 1710 may adjust the processing are noted below. Further, the distributor 1710 may utilize the below described methods in any combination synchronize the communication channels.

In some implementations, the distributor 1710 may send an extra packet to a modulator that is running faster (e.g. has a lower buffer depth) thereby resynchronizing the data. The extra packet may be a junk packet. The junk packet may include null data or other identifiable data that will be automatically filtered by the receiver device 1748. Alternatively, the extra packet may include data from the data stream. The extra packet may be a duplicate packet that is also sent over other communication channels. In certain implementations, the extra packet may be identified as a high priority packet that would benefit from redundant transmission over multiple communication channels. The extra packet may be sent on a marker boundary.

The distributor 1710 may also modify the communication unit size for each channel to adjust the processing based on the feedback signals. If the modulator is running faster than other modulators (e.g. the buffer has a lower buffer depth) the distributor 1710 may increase the size of the communication provided to that communication channel. The distributor 1710 may increase the size of the communication unit by adding packets to the communication unit. With regard to adding packets and/or communication units the slowest communication channel will often dictate adding data to each of the other channels.

If the modulator is running slower than other modulators (e.g. the buffer has a higher buffer depth) the distributor 1710 may decrease the size of the communication unit provided to that communication channel. The distributor 1710 may decrease the size of the communication unit by removing packets from the communication unit.

In some implementations, distributor 1710 may send a redundant communication unit to one or more of the other channels 1722, 1724 thereby resynchronizing the channels. In this scenario, the distributor 1710 may identify communication units that include high priority data from the data stream. In some implementations, the distributor 1710 may skip sending a communication unit or packet (e.g. in a round robin) to a slower communication channel.

In some implementations, the distributor 1710 may adjust the clock speed to increase the clock speed for slower communication channels and/or decrease the clock speed for faster communication channels. The modulator may also include logic or circuitry to increase or decrease the clock speed directly based on the buffer depth (e.g. fullness of the buffer).

It is also contemplated that the distributor 1710 may adjust the synchronization as described above based on a combination of the signal 2010 and signals from the modulators 1720, 1722, 1724. Further, it is contemplated that any of the synchronization methods described herein may be used in combination with any of the architectures or features described above.

The methods, devices, and logic described above may be implemented in many different ways in many different combinations of hardware, software or both hardware and software. For example, all or parts of the system may include circuitry in a controller, a microprocessor, or an application specific integrated circuit (ASIC), or may be implemented with discrete logic or components, or a combination of other types of analog or digital circuitry, combined on a single integrated circuit or distributed among multiple integrated circuits. All or part of the logic described above may be implemented as instructions for execution by a processor, controller, or other processing device and may be stored in a tangible or non-transitory machine-readable or computer-readable medium such as flash memory, random access memory (RAM) or read only memory (ROM), erasable programmable read only memory (EPROM) or other machine-readable medium such as a compact disc read only memory (CDROM), or magnetic or optical disk. Thus, a product, such as a computer program product, may include a storage medium and computer readable instructions stored on the medium, which when executed in an endpoint, computer system, or other device, cause the device to perform operations according to any of the description above.

The processing capability of the architectures may be distributed among multiple system components, such as among multiple processors and memories, optionally including multiple distributed processing systems. Parameters, databases, and other data structures may be separately stored and managed, may be incorporated into a single memory or database, may be logically and physically organized in many different ways, and may implemented in many ways, including data structures such as linked lists, hash tables, or implicit storage mechanisms. Programs may be parts (e.g., subroutines) of a single program, separate programs, distributed across several memories and processors, or implemented in many different ways, such as in a library, such as a shared library (e.g., a dynamic link library (DLL)). The DLL, for example, may store code that performs any of the processing described above. While various embodiments of the invention have been described, it will be apparent to those of ordinary skill in the art that many more embodiments and implementations are possible within the scope of the invention. Accordingly, the invention is not to be restricted except in light of the attached claims and their equivalents.

Claims

1. A system comprising: an input interface;output interfaces to communication channels; andat least one buffer;a processor in communication with the input interface and the output interfaces, the processor being configured to: obtain source data from the input interface;distribute the source data through the at least one buffer across the communication channels in a bonded channel group; andmonitor and adjust synchronization of the communication channels based on a fullness of the at least one buffer.

2. The system of claim 1, the processor being configured to monitor fullness of the at least one buffer and adjust an amount of data to distributed to each communication channel.

3. The system of claim 1, the processor being configured to assign the source data into communication units, distribute the communication units to the communication channels, monitor fullness of the at least one buffer, and adjust a number of the communication units to distribute to a communication channel based on the fullness of the at least one buffer.

4. The system of claim 1, the processor being configured to assign the source data into communication units, distribute the communication units to the communication channels, monitor fullness of the at least one buffer, and adjust a size of the communication units to distribute to a communication channel based on the fullness of the at least one buffer.

5. The system of claim 1, the processor being configured to monitor fullness of the at least one buffer and adjust a clock speed of a communication channel based on the fullness of the at least one buffer.

6. The system of claim 1, the processor being configured to monitor fullness of the at least one buffer and adjust an amount of data to distribute to a communication channel and a clock speed of the communication channel based on the fullness of the at least one buffer.

7. The system of claim 1, wherein the processor is configured to receive a feedback signal from a receiving device and control an amount of data to distribute to a communication channel or a clock speed of the communication channel based on the fullness of the at least one buffer.

8. The system of claim 7, wherein the processor is configured to assign the source data to communication units, distribute the communication units to the communication channels, and adjust a size of the communication units or a number of communication units to distribute to a communication channel based on the feedback signal.

9. The system of claim 1, wherein the processor is configured synchronize a first clock of a first channel to a second clock of a second channel.

10. The system of claim 1, wherein the processor is configured to distribute the data in a round-robin manner across the bonded channel group.

11. A method comprising: obtaining a packet stream;dividing the packet stream into communication units;distributing the communication units across communication channels of a bonded channel group;monitoring fullness of a buffer; andadjusting synchronization of the communication channels based on the fullness of the buffer.

12. The method of claim 11, further comprising adjusting a number of the communication units to distribute to a communication channel based on the fullness of the buffer.

13. The method of claim 11, further comprising adjusting a size of the communication units to distribute to a communication channel based on the fullness of the buffer.

14. The method of claim 11, further comprising adjusting a clock speed of a communication channel based on the fullness of the buffer.

15. The method of claim 11, further comprising receiving a feedback signal from a receiving device and controlling an amount of data to distributed to a communication channel or a clock speed of the communication channel based on the fullness of the buffer.

16. The method of claim 15, further comprising adjusting a size of the communication units or a number of communication units to distribute to a communication channel based on the feedback signal.

17. A system comprising: program input sources;a statistical multiplexer connected to the program input sources, the statistical multiplexer comprising a source transport stream output configured to carry a program packet stream of program packets obtained from the program input sources;a distributor in communication with the statistical multiplexer, the distributor configured to: read bonding configuration parameters that identify a bonded channel group of communication channels among a set of available communication channels;break the packet stream into individual program packets;distribute the individual program packets across the bonded channel group; andmonitor and adjust synchronization of the communication channels.

18. The system of claim 17, the distributor being configured to monitor a fullness of a buffer and adjust a size or number of the program packets to distribute to a communication channel based on the fullness of the buffer.

19. The system of claim 17, the distributor being configured to receive feedback from a receiving device and adjust a size or number of the program packets to distribute to a communication channel based on the feedback from the receiving device.

20. The system of claim 17, further comprising the distributor being configured to monitor fullness of a buffer and adjust a clock speed of a communication channel based on the fullness of the buffer.