JP4577341B2 - Relay device and autonomous error correction network - Google Patents

Description

  The present invention relates to a relay device and an autonomous error correction network provided at a connection point between one or a plurality of networks.

  In recent years, broadband subscribers are rapidly increasing in the Internet and intranets. Broadband applications that require certain bandwidth and quality, such as music and video download distribution and streaming distribution, and those that require realization of large-capacity and high-speed content distribution services such as real-time communication are increasing. It's getting on. These applications are expected to drive the market for higher speed broadband services such as FTTH [Fiber To The Home]. With this high-speed broadband service, high-speed download that can be downloaded within one minute, and the realization of a service level that does not stop during streaming or cause macroblock errors, etc. are ensured for high-quality content and digital works. Together with the standardization of rights management technology, it is even said to be the key to the development of Japan's broadband market.

  As a technology that processes such a large amount of information at high speed and in real time and responds to the quality requirements according to each application, the core network has a hardware level in Layer 2 using MPLS (Multi Protocol Label Switching). High-speed switching technology has been developed, and a technology for dividing and caching contents in an edge network has been put into practical use.

  However, at present, the technology for ensuring high speed and service level with the end-to-end on the Internet, which is a collection of network technologies of various eras, is not yet available. TCP / IP (Transmission Control Protocol / Internet Protocol), which forms the basis of the Internet, realizes highly reliable communication by flow control and retransmission request of lost packets. However, retransmission of lost packets is not an essential solution for ensuring the service level of broadband media because congestion due to concentration of access calls further congestion due to retransmission.

  In addition, FTP (File Transfer Protocol), which is widely used as a protocol for downloading, defines the receiving capability of each receiver side as a product of the transmission band and the delay time in order to prevent network congestion due to retransmission. Flow control is performed by keeping this constant. For this reason, when packet retransmission occurs and the delay time increases, the transmission band decreases, and high-speed downloading corresponding to broadband cannot be performed.

  In various media formats popular in streaming, various measures are taken to maintain image quality even on a network with poor QoS [Quality of Service], but packet loss of about 0.1 to 1% is achieved. Depending on the situation, a macroblock error or frame skip may occur in the reproduced video. In these streaming distributions, UDP (User Datagram Protocol) that does not perform flow control and retransmission of lost packets is often used to ensure throughput. Since UDP does not have a scheme that reduces the burden on the network against network congestion, a dedicated protocol is used in each application to reduce the resolution autonomously and lower the service level of each flow. Is realized. However, the follow-up sensitivity to network congestion is not sufficient, and in particular, there is a problem that the return of the service level after the congestion is relieved is very slow, and the quality of the entire program is lowered due to slight congestion.

  Therefore, in spite of the explosive spread of broadband by ADSL [Asymmetric Digital Subscriber Line] etc., it is difficult to spread full-scale broadband CDS [CircuitDesign Service] on the Internet. As a result, the current situation is one of the important rate-limiting factors for the increase in demand for FTTH in Japan.

On the other hand, as means for realizing QoS that does not depend on TCP / IP flow control or retransmission of lost packets, there is a method using UDP and forward error correction (hereinafter referred to as FEC [Forward Error Correction]).
JP 2001-189665 A US Pat. No. 6,307,487

  However, the conventional FEC has been mainly used for improving the S / N in the layer 1 of the OSI [Open Systems Interconnection] model. However, in order to increase the resistance to burst errors, the transmission data has randomness in advance. It is necessary to be creative. For this reason, a trade-off relationship arises between the resistance to burst errors and the delay time of transmission data. Therefore, it is not always necessary to use in a strict application such as VoIP [VoiceOver Internet Protocol] used for telephones. It was not appropriate.

  In addition, FEC performs an encoder assuming an end-to-end error rate and a packet loss rate in advance, but an end-to-end error rate and a packet in the Internet, which is essentially a collection of independent networks. It was difficult to assume a loss rate in advance. Even when an end-to-end error rate or packet loss rate can be assumed in advance, the transmission source transmits FEC data with a corresponding redundancy level increased, and the use efficiency of the transmission band of the network is reduced. Transmission time also increases. In particular, when an individual sends / receives data to / from an individual via a network using a public line, increasing the redundancy level in consideration of an error rate or packet loss rate in the middle of the network increases the amount of data to be transmitted. Increased personal costs. That is, the individual must bear the cost for the error rate, packet loss rate, etc. in the middle of the network.

  Referring to FIG. 6, when streaming content is transmitted from a transmission source 100 such as an ASP to a reception destination 101 such as a personal computer (hereinafter referred to as a personal computer) via the networks 102, 103, 104, and 105. An example will be described. When streaming content data is transmitted from the transmission source 100 to the reception destination 101 via the four networks 102, 103, 104, and 105, three relay devices 106 and 107 are connected to the connection points of the networks 102, 103, 104, and 105. , 108 exists. Here, the data capacity of the streaming content is set to 0.3 Mbps, the maximum packet loss rates in the networks 102, 103, 104, and 105 are set to 1%, 15%, 25%, and 1%, respectively, and the packet overhead is set to 1086/1024. The FEC overhead is 1.05. In this case, in order to decode the FEC data transmitted from the transmission source 100 without the loss packet at the reception destination 101, the maximum packet loss rate, the packet overhead, and the FEC overhead must be added to the streaming content data. Don't be. Therefore, in the transmission source 100, if the redundancy level (about 77%) is set in consideration of the QoS information, and FEC is performed to encode the data of the streaming content (0.3 Mbps), the transmission band is 0.53 Mbps is required. However, since the effective bandwidth of the network 102 connected to the transmission source 100 is 0.5 Mbps, FEC data having a transmission bandwidth of 0.53 Mbps cannot be transmitted. Therefore, if 0.5 Mbps FEC data is transmitted from the transmission source 100, a loss packet is generated at the reception destination 101, and the streaming content cannot be completely reproduced. Incidentally, in this example, since the redundancy level is set in consideration of packet loss in the networks 103 and 104 in the middle of data transmission, the transmission source 100 or the reception destination 101 uses the packets in these networks 103 and 104. Loss must be borne as a transmission cost.

  Furthermore, in the existing FEC technology such as Reed-Solomon code widely used for FEC such as CD-ROM and DVD, the calculation processing time is almost equal to the burst rate of the packet loss that is subject to error rate correction. Since it tends to be proportional to the square, it has not been easy to apply it to an application that distributes broadband data called rich content such as video and music over the Internet where bursty packet loss occurs.

  The above-described technique is disclosed in Patent Literature 1, Patent Literature 2, and the like, but no means for solving the above problems is disclosed.

  Therefore, an object of the present invention is to provide a relay device and a network excellent in use efficiency of a network transmission band, delay in data transmission, and QoS.

The relay device according to the present invention is provided at a connection point between one or a plurality of networks used when transmitting data from a transmission unit to a reception unit, receives data from an upstream device, and transmits the data to a downstream device. A relay device for transmitting data, which decodes forward error correction encoded received data from the upstream device, re-encodes the decoded data, and transmits the re-encoded data downstream A first transfer means for transmitting to the side apparatus, and a second transfer means for transferring the received data in units of packets forward-error-corrected from the upstream apparatus to the downstream apparatus as they are, When data in units of packets subjected to direction error correction coding is received, transfer by the second transfer means is executed and decoding by the first transfer means is started .

  According to this relay device, it is possible to improve redundancy in the middle of a network that transmits data. For this reason, when data with forward error correction performed with a small redundancy level is transmitted in the transmission unit, even if the redundancy level is lowered in the middle of the network, the redundancy can be improved by a relay device in the middle of the network. Packet loss does not occur in the part. Therefore, since data can be transmitted with a small redundancy and packet loss can be eliminated, the use efficiency of the transmission band can be improved, the transmission time can be shortened, and the QoS and service level in End-to-End can also be improved. To do. In particular, according to this relay apparatus, the forward error correction data is directly transferred downstream and the forward error correction data is decoded, so that the delay can be suppressed as much as possible.

  Incidentally, if the maximum packet loss rate of m partial networks (n) through which a certain flow passes is L (n)%, the packet loss rate of the entire network of this flow is 1-((1-L (1)). ) × (1-L (2)) ×... × (1-L (m))). When forward error correction is performed in consideration of this, the redundancy level of data transmitted from the transmission source becomes enormous, and the traffic in the partial network or relay device near the transmission source becomes enormous. Therefore, conventionally, it has been impossible to economically distribute rich content, and the service level has been lowered when packet loss more than expected has occurred. However, according to the relay device according to the present invention, since the redundancy lost for each section of the network is recovered and relayed to the next network, the service level is not lowered while effectively using the transmission band in each network. .

  The transmission unit is a transmission source that transmits data via a network. For example, an individual (including a transmission device such as a personal computer), an ISP [Internet Service Provider] (including a transmission device such as a server), an ASP, or the like. [Application Service Provider] (including a transmission device such as a server). The receiving unit is a receiving destination that receives data via a network, and is, for example, an individual or a corporation (including a receiving device such as a personal computer). The network is, for example, the Internet or an intranet. The relay device is a device that is provided at a connection point between networks and relays data from the upstream side to the downstream side, and is, for example, a router or a gateway placed in an ISP or ASP. Incidentally, the transmission unit may also perform processing in the relay device. The upstream side is the transmission unit side when data is transmitted. The downstream side is the receiving unit side when data is transmitted.

The relay device according to the present invention includes a receiving unit that receives a transmission stop request from the downstream device to the first transfer unit, and the first transfer unit recodes until receiving a transmission stop request from the downstream device. It is good also as a structure which performs the transmission to the downstream apparatus of the digitized data .

In the relay apparatus of the present invention, the downstream apparatus may transmit a transmission stop request to the first transfer means when the received data transferred by the second transfer means can be completely decoded. Furthermore, in the relay device of the present invention, the downstream device cannot completely decode the received data transferred by the second transfer means, and the re-encoding transmitted by the first transfer means following the received data. When the encoded data is received, the transmission stop request to the first transfer means may be transmitted when the re-encoded data can be completely decoded.

In the relay device of the present invention, the first transfer means has a buffer for temporarily storing the decoded data, holds the decoded data in the buffer until a transmission stop request is received, and transmits the transmission stop request. May be configured to delete the decrypted data held in the buffer. The relay apparatus of the present invention further includes a determination unit that determines whether or not the received data in the forward error correction encoded packet unit received from the upstream apparatus can be completely decoded. When it is determined that the received data in units of packets received from the side apparatus and encoded in the forward error correction can be completely decoded, transmission of data with improved redundancy over the received data following the received data The stop request may be transmitted to the upstream device.

In the relay device according to the present invention, the necessity of improving the redundancy of data and the redundancy based on the service quality information or secondary information of the service quality information of one or more networks located downstream of the relay device Determining means for determining a level, and determining whether or not to perform forward error correction, determining the type of forward error correction, determining the redundancy level, and / or according to the application and / or service level for each flow Alternatively, it may be configured to set parameters for forward error correction. Furthermore, in the relay apparatus of the present invention, the redundancy of data is improved by adding and re-encoding the insufficient redundancy without decoding all the data during the decoding of the flow block. It is good also as a structure.

  According to this relay apparatus, determination of necessity of forward error correction or determination of the type of forward error correction is performed according to the application and / or service level, so that the QoS required for various applications can be reduced. Accordingly, redundancy can be improved by forward error correction. In other words, there are applications that are sensitive to delay such as VoIP or applications that are insensitive to delay such as VOD but severe to packet loss. However, in the conventional technology, forward error correction is performed in a lower layer of OSI. For this reason, the QoS request for each application could not be handled. However, according to this relay apparatus, it is possible to perform appropriate forward error correction for each application having different QoS requirements such as usage time, allowable delay, allowable palette loss rate, and allowable packet loss burst number.

  The service quality information is information related to QoS in the network, and is, for example, packet loss rate, error rate, delay, jitter, and the like. The secondary information of the service quality information is information obtained by performing four arithmetic operations on one information or a plurality of information in the service quality information. The flow means a data flow that flows through a network composed of one or a plurality of blocks including all or part of transmission data such as content. As types of forward error correction, there are a Hamming code, a BCH [Bose Chaudhuri-Hocquenghem] code, a Reed-Solomon code, a tornado code, an LT [Luby Transform] code, and the like.

  In the relay device of the present invention, an application and / or service level may be set in a fixed-length data area of a packet. The relay apparatus of the present invention may handle a packet in which an application and / or service level is set in a fixed-length data area of the packet.

  According to this relay apparatus, by referring to data stored in a predetermined area of the packet head, the application and / or service level can be determined at high speed by hardware. Incidentally, in the conventional packet structure, since the application and / or service level is set in the forward error correction header of the variable length payload of the packet, the application and / or service level is determined by software processing. Processing time was required.

  In the relay device of the present invention, it is preferable that the type of forward error correction data is an LT code.

  Incidentally, in the conventional forward error correction, it is necessary to combine data scrambling and forward error correction in order to improve resistance to bursty packet loss and errors, and if the scramble width is increased, the resistance to burst errors is reduced. Improves, but delay increases. Therefore, if decoding and re-encoding are repeated in the relay device using the conventional forward error correction, the delay accumulates every time it passes through the relay device, and it cannot be used for VoIP such as a telephone or interactive game. Of course, there is a problem because the delay becomes too large even in applications such as VOD and live broadcasting. However, according to the relay device according to the present invention, it is possible to add and decode the lacking redundancy without decoding and re-encoding all the data at the time of relay, so that the delay is increased. Can be minimized.

An autonomous error correction network according to the present invention is an autonomous error correction network including a transmission unit, a reception unit, and one or a plurality of relay devices provided at connection points between one or a plurality of networks. Then, the relay device decodes the reception data in units of packets that has been forward error correction encoded from the transmission unit or the upstream relay device, re-encodes the decoded data, and converts the re-encoded data to First transfer means for transmitting to the downstream relay device or receiving unit, and the forward relay error correction encoded packet unit received data from the transmitting unit or upstream relay device as it is downstream of the relay device or receiving unit A second transfer means for transferring the packet data to the packet, and when the packet unit data forward-correction-corrected encoded from the transmission unit or the upstream relay device is received, the transfer by the second transfer means is executed. Together, characterized in that it starts decoding by the first transfer means.

  According to this autonomous error correction network, the redundancy can be improved at the time of data relay, so that the transmission band can be used effectively in each network, and the service level does not decrease.

In the above-described autonomous error correction network of the present invention, the relay device includes a reception unit that receives a transmission stop request for the first transfer unit from the downstream relay device or reception unit, and the first transfer unit includes the downstream transfer device. The transmission of the re-encoded data to the downstream relay device or reception unit may be performed until a transmission stop request from the relay device or reception unit is received .

  In the above-described autonomous error correction network of the present invention, the relay device determines whether or not to perform forward error correction, determines the type of forward error correction, and redundancy in accordance with the application and / or service level for each flow. It may be configured to determine the degree level and / or set parameters for forward error correction.

  The present invention can improve the redundancy at the time of relaying between networks even if the redundancy decreases in the middle of a network for transmitting data. Therefore, since data can be transmitted with a small redundancy, the use efficiency of the transmission band of the network can be improved and the transmission time can be shortened. Further, since packet loss can be eliminated or eliminated as much as possible, QoS in End-to-End can be reduced. Service level is also improved.

  In particular, in the present invention, since data transfer and data decoding and re-encoding are performed in parallel, the delay can be minimized. Further, in the present invention, even when packet loss or the like occurs and the transferred data cannot be 100% decoded to the original data, a packet that has been re-encoded and subjected to forward error correction is received from the upstream device. By doing so, the original data can be 100% decrypted.

  Furthermore, the present invention determines whether forward error correction needs to be performed or determines the type of forward error correction according to the application and / or service level, so that it can meet the QoS required for various applications. Therefore, redundancy can be improved by forward error correction, and the service level can also be satisfied.

  Embodiments of a relay device and an autonomous error correction network according to the present invention will be described below with reference to the drawings.

  In the present invention, in order to realize optimal transmission in consideration of transmission band, delay time, and QoS in transmission of data through a network, when the redundancy level is lowered during transmission, the transmitter, the network, The redundancy is improved based on the QoS information of the downstream network in the relay device provided at the connection point of the receiving unit. Furthermore, the present invention is based on applications and service levels in order to realize optimum transmission according to service levels set in applications and service contracts with different QoS requirements such as allowable delay and allowable pallet loss rate. The determination of whether or not to implement FEC, the determination of the type of FEC, the determination of the redundancy level, and / or the setting of FEC parameters are performed. In particular, in the present invention, in order to minimize the delay, the data received at the relay apparatus is transferred to the downstream side as it is and the data is decoded. Send the encoded data from the transmitting unit or relay device on the side to the shortage.

  In this embodiment, there are two embodiments. In the first embodiment, the relay apparatus performs decoding and FEC re-encoding to improve redundancy in serial with transmission data relay, In the second embodiment, decoding for improving redundancy and re-encoding by FEC are performed in parallel with relay of transmission data in the relay apparatus. In this embodiment, an LT code is used as FEC.

  First, the first embodiment will be described. With reference to FIG. 1, an overall configuration of a network according to the first embodiment will be described. FIG. 1 is a configuration diagram of the entire network according to the first embodiment.

  In the first embodiment, streaming content is distributed from the transmission source 1 to the reception destination 2 via the four networks 3, 4, 5, and 6, and the connection points of the four networks 3, 4, 5, and 6 are distributed. The data is relayed by the relay devices 7, 8, and 9 provided in FIG.

  The transmission source 1 is a provider that distributes content such as ASP, and includes a distribution server (not shown) for distributing data via a network. Further, the transmission source 1 includes the FEC encoder 1a in the distribution server or separately. In the transmission source 1, the original data of the streaming content is encoded into FEC data by the FEC encoder 1a, and the FEC data is distributed in units of packets. Incidentally, the transmission source 1 has a contract with the reception destination 2 (for example, an individual user), and the service level (service type) of the content distributed from the transmission source 1 in this contract, the type of application, the distribution fee, etc. Is set. The service level includes error rate priority, delay priority, SLA [Service Level Agreement] information, etc. as shown in FIG.

  In the FEC encoder 1a, the original data of the streaming content is encoded into FEC data at a predetermined redundancy level by FEC using the LT code. The FEC data is divided into packet payloads, stored, and transmitted. The predetermined redundancy level is based on packet overhead (eg, 1086/1024), FEC overhead (eg, 1.05), and maximum packet rate (1%) of network (S) 3 connected to source 1 For example, it is set to about 15%. In the case of transmitting 0.3 Mbps streaming content at a redundancy level of 15%, a transmission band of 0.345 Mbps is required, but the network (S) 3 can be sufficiently transmitted with an effective band of 0.5 Mbps. Incidentally, it is not necessary for the transmission source 1 to consider the maximum packet rate of the networks 103, 104, and 105 as in the prior art. The streaming content is composed of a large number of blocks, and each block is composed of a large number of packets.

  A packet configuration will be described with reference to FIG. FIG. 2 is an explanatory diagram of a packet, (a) is a configuration diagram of the packet, and (b) is a data configuration diagram of the IP header.

  The packet is composed of a fixed-length net header, an IP header, a UDP header, and a variable-length payload. FEC data is divided and stored in this payload. The IP header has a data structure as shown in FIG. 2B, and the service type is stored in the second byte from the top of the IP header. Therefore, the service type of the streaming content can be determined by referring to the 8-bit data of the second byte of the IP head of each packet. The network header or UDP head stores data indicating the type of application (streaming content or the like), and the type of application can be determined by referring to the data of the network header or UDP head. it can.

  With reference to FIG. 3, the FEC by the LT code performed by the FEC encoder 1a and the like will be described. Note that FEC using the LT code is also performed in the relay apparatuses 7, 8, and 9.

  FIG. 3 shows an example in which there are ten input data a to h corresponding to the original data of the streaming content. In FEC using the LT code, a random number is generated for each input data. Then, an equation (for example, aXORg in the case of the first column) is formed by exclusive OR for each column as output data, and this equation becomes FEC data. Basically, if there are equations (output data) for the number of input data, the equations can be solved. However, in consideration of packet loss during data network transmission, the number of equations (output data) is increased from the number of input data (that is, the redundancy level is increased). For example, if the redundancy level is 10%, 11 output data (equations) are generated for 10 input data. That is, even if 11 output data (FEC data) is transmitted from the transmission source and one output data (FEC data) is lost in the network, the equation is solved by 10 output data (FEC data) at the reception destination. be able to.

  Returning to FIG. 1, the receiving destination 2 is an individual user or the like, and includes a personal computer (not shown) or the like for receiving data via the network. The receiver 2 includes an FEC decoder 2a and a streaming viewer 2b in the personal computer or separately. The receiving destination 2 receives the FEC data in packet units and decodes it by the FEC decoder 2a. Then, at the receiving destination 2, the decoded data in units of packets is connected by the streaming viewer 2b and reproduced as streaming content.

  The networks 3, 4, 5, and 6 are networks such as the Internet and an intranet, and each have QoS information. As shown in FIG. 4, the QoS information includes a packet loss rate, an effective bandwidth, an error rate, a delay, a maximum burst length, jitter, and the like, and has different characteristics depending on the networks 3, 4, 5, and 6. For example, the maximum packet loss rate is 1% for the network (S) 3, 15% for the network (N-1) 4, 25% for the network (N) 5, and 1% for the network (D) 6. In the case of the effective bandwidth, the network (S) 3 is 0.5 Mbps, the network (N-1) 4 is 1000 Mbps, the network (N) 5 is 100 Mbps, and the network (D) 6 is 8 Mbps.

  The relay devices 7, 8, and 9 are devices that relay data between networks, and are, for example, routers, gateways, and the like provided in ISPs. In particular, when the redundancy level of the received FEC data is lowered, the relay devices 7, 8, and 9 decode the received FEC data and re-encode the decoded data in order to increase the redundancy level. Then send it downstream. For this purpose, the relay devices 7, 8, and 9 include packet switches 7a, 8a, and 9a, FEC decoders 7b, 8b, and 9b, buffer memories 7c, 8c, and 9c, and FEC encoders 7d, 8d, and 9d. Incidentally, the constituent means of these relay apparatuses 7, 8, 9 may be configured by hardware, or may be configured by software by executing a dedicated program by a computer. Since the FEC relay apparatuses 7, 8, and 9 have the same configuration, only the relay apparatus (n) 8 will be described in detail here.

  The relay device (n) 8 receives FEC data in units of packets from the network (N−1) 4 and takes the protocol of the network (N) 5 into consideration via the network (N) 5. FEC data is transmitted in units of packets to 9. Further, the relay device (n) 8 monitors the redundancy level of the received FEC data, and when the redundancy level is lowered, the relay device (n + 1) 9 may have insufficient redundancy. Decodes the received FEC data and re-encodes it to increase the redundancy level.

  The packet switch 8a checks the redundancy level of the received FEC data. In the packet switch 8a, based on the redundancy level of the received FEC data and the QoS information (packet loss rate etc.) of the downstream network (N) 5, the FEC data received as it is in the network (N) 5 is used. After transmission, it is determined whether or not the redundancy is insufficient in the relay device (n + 1) 9. If it is determined that the redundancy is not insufficient, the packet switch 8a determines that FEC is not performed, and transmits the received FEC data as it is to the relay device (n + 1) 9 via the network (N) 5 in units of packets. On the other hand, when it is determined that the redundancy is insufficient, the packet switch 8a determines to perform FEC, and sends the received FEC data to the FEC decoder 8b. Note that the packet switch 8a may determine whether or not the redundancy is insufficient at the destination 2 in consideration of the QoS information of not only the network (N) 5 but also the downstream network (N + I). .

  The FEC decoder 8b decodes the FEC data in units of packets and decodes the original data of the streaming content. Then, the FEC decoder 8b temporarily stores the original data in the buffer memory 8c.

  The FEC encoder 8d sets FEC parameters. Then, the FEC encoder 8d re-encodes the original data temporarily stored in the buffer memory 8c into FEC data by FEC using the LT code based on the set FEC parameter. Then, the FEC encoder 8d sends the re-encoded FEC data to the packet switch 8a in units of packets, and sends the FEC data in units of packets from the packet switch 8a to the relay apparatus (n + 1) 9 via the downstream network (N) 5. Send. In this encoding, all the packets in the block may be decoded and then decoded in units of blocks, or may be sequentially re-encoded in units of packets during decoding of the blocks.

  A method for setting FEC parameters will be described with reference to FIG. FIG. 4 is an explanatory diagram of FEC parameters.

  The FEC parameter is a parameter necessary for encoding by FEC, and includes an assumed maximum packet loss (redundancy level), a target packet reproduction error rate, an FEC overhead, a block length, a packet length, and the like. These FEC parameters include the redundancy level of the received FEC data, the packet header information of the received FEC data, the QoS information of the downstream network, and the resource information of the devices (transmission source, reception destination, relay device) of the entire network. And the reception information of the downstream relay device (n + i) and the like.

  As packet header information, there are error rate priority, delay priority, SLA information, etc. as service level information, and there are streaming content, telephone, interactive game, etc. as application types. By considering the service level information, the FEC parameter is set so as not to drop below the set service level. By considering the application type, the FEC parameters are set so as to satisfy the QoS required by each application such as an application sensitive to delay and an application sensitive to packet loss. Incidentally, since the service level information and application type are set in a fixed-length area of the packet, it can be recognized at high speed.

  The QoS information of the downstream network includes a packet loss rate, effective bandwidth, error rate, delay, maximum burst length, jitter, and the like. For example, if the redundancy level of the received FEC data is lowered, the redundancy level becomes less than 0% in the downstream relay device or the receiving destination due to the packet loss rate of the downstream network (the redundancy is insufficient). FEC parameters are set so as to increase the redundancy level. At this time, the FEC parameters are set so that the transmission band of the transmission data does not exceed the effective bandwidth of the downstream network.

  The device resource information includes a utilization rate of a CPU [Central Processing Unit], free memory, and the like. Looking over the entire network, FEC parameters are set so that decoding and re-encoding are performed in a device with sufficient CPU computing capacity or a device with sufficient memory.

  The reception information of the downstream relay device includes a reception packet rate or packet sufficiency information, delay, buffer memory, and the like. The FEC parameter is set in consideration of reception information of the downstream relay device such as whether packet loss occurs on the downstream side or whether the delay time is long.

  The operation of the relay device (n) 8 will be described with reference to FIGS. As a premise, the streaming content of 0.3 Mbps is encoded into FEC data in the transmission source 1, and the FEC data in packet units is transmitted over the network (S) 3. The packet-unit FEC data is received by the relay device (n−1) 7 via the network (S) 3. When the relay device (n-1) 7 determines whether or not to perform FEC based on the redundancy level of the FEC data in packet units, the QoS information of the network (N-1) 4, and the like, and does not perform FEC In this case, the received FEC data in packet units is transmitted as it is to the relay apparatus (n) 8 via the network (N-1) 4, and when the FEC is performed, the received FEC data in packet units is decoded and re-transmitted. The FEC data encoded and improved in redundancy is transmitted to the relay device (n) 8 via the network (N-1) 4. The packet unit FEC data is received by the relay device (n) 8.

  The relay device (n) 8 checks the redundancy level of the received FEC data in packet units, and determines whether or not to implement FEC based on the redundancy level and the QoS information of the network (N) 5. . For example, when the redundancy level of FEC data in units of packets is reduced to 5%, the maximum packet loss rate of the network (N) 5 is 25%, so that packet loss occurs in the network (N) 5. In the relay device (n-1) 9, the redundancy level is reduced to less than 0%. In such a case, the relay device (n) 8 implements FEC to improve the redundancy.

  When it is determined that FEC is not performed, the relay device (n) 8 transmits the received FEC data in units of packets to the relay device (n + 1) 9 as it is via the network (N) 5.

  On the other hand, if it is determined that FEC is to be performed, the relay device (n) 8 decodes the received FEC data in packet units. Further, the relay device (n) 8 sets FEC parameters based on the QoS information of the network (N) 5, the application, the service level, and the like (see FIG. 4). For example, since the maximum packet loss rate of the network (N) 5 is 25%, the redundancy level is set to about 30%. Then, the relay device (n) 8 re-encodes the decoded data into FEC data by FEC using the LT code based on the set FEC parameter. Subsequently, the relay apparatus (n) 8 transmits the re-encoded FEC data in packet units to the relay apparatus (n + 1) 9 via the network (N) 5.

  Thereafter, the FEC data in packet units is received by the relay apparatus (n + 1) 9 via the network (N) 5. The relay device (n + 1) 9 determines whether or not to implement FEC based on the redundancy level of the FEC data in packet units, the QoS information of the network (D) 6, and the like. When the FEC data in packet units is transmitted as it is to the receiver 2 via the network (D) 6 and the FEC is performed, the FEC data in which the received packet units are decoded and re-encoded to improve redundancy. Data is transmitted to the receiver 2 via the network (D) 6. Then, the FEC data in packet units is received at the receiving destination 2. At the receiving destination 2, the received FEC data in units of packets is sequentially decoded and reproduced as streaming content. In this decoding, since the redundancy is improved in the relay devices 7, 8, and 9, no errors such as loss packets and macroblock errors occur in the streaming content.

  According to the relay devices 7, 8, and 9, redundancy can be improved in the middle of the entire network, so that it is not necessary to set a large redundancy level in consideration of the QoS information of the entire network at the transmission source 1. The transmission bandwidth of transmission data can be suppressed. Further, according to the relay devices 7, 8, and 9, since the redundancy information is improved by considering the QoS information of the downstream network, the FEC data can be completely decoded into the original data at the receiving destination 2.

  Further, according to the relay devices 7, 8, and 9, since the FEC parameters are set according to the application and service level, the application can be transmitted while satisfying the QoS required by each application, and the service level is also improved. Can be maintained.

  Next, a second embodiment will be described. With reference to FIG. 5, an overall configuration of a network according to the second embodiment will be described. FIG. 5 is a configuration diagram of the entire network according to the second embodiment. In the second embodiment, the same reference numerals are given to the same components as those in the first embodiment, and the description thereof is omitted.

  In the second embodiment, streaming content is distributed from the transmission source 1 to the reception destination 2 via the four networks 3, 4, 5, and 6. The data is relayed by the relay devices 17, 18, and 19 provided in FIG.

  The relay devices 17, 18, and 19 are devices that relay data between networks, and are, for example, routers, gateways, and the like provided in ISPs. In particular, the relay devices 17, 18, 19 transfer the received FEC data in packet units to the downstream relay devices 18, 19 or the receiver 2 as they are, and simultaneously decode the received FEC data in packet units. Then, if the FEC data cannot be completely decoded in the relay devices 17, 18, and 19 (or the reception destination 2), the upstream relay devices 17, 18, and 19 or the transmission are used to compensate for the insufficient redundancy. The FEC data of the packet unit encoded from the original 1 is transmitted. When the relay devices 17, 18, 19 (or the receiver 2) can completely decode the FEC data in units of packets that have been transmitted, the upstream relay devices 17, 18, 19 or the transmission source 1 Stop sending from. Further, the relay devices 17, 18, 19 (or the transmission source 1) transmit the decoded data to the downstream relay devices 17, 18, 19 or the reception destination 2 in order to transmit the encoded FEC data in packet units. Re-encode. For this purpose, the relay devices 17, 18, and 19 include packet copies 17a, 18a, and 19a, FEC decoders 17b, 18b, and 19b, buffer memories 17c, 18c, and 19c, and FEC encoders 17d, 18d, and 19d. Incidentally, the configuration means of these relay devices 17, 18, and 19 may be configured by hardware, or may be configured by software by executing a dedicated program by a computer. Since the FEC relay apparatuses 17, 18, and 19 have the same configuration, only the relay apparatus (n) 18 will be described in detail here.

  The relay device (n) 18 receives the FEC data from the network (N-1) 4 in units of packets, and takes the FEC data in units of packets into the network (N) in consideration of the protocol of the network (N) 5. 5 to the relay device (n) 19. At the same time, the relay device (n) 18 decodes the received FEC data. When the relay device (n) 18 cannot completely decode the received FEC data, the relay device (n) 18 receives the re-encoded FEC data transmitted from the upstream relay device (n-1) 17 and re-encodes the FEC data. The FEC data is completely decoded based on the converted FEC data. When the FEC data can be completely decoded, the relay device (n) 18 stops the transmission from the relay device (n−1) 17. Further, in the relay device (n) 18, in order to cope with the case where the FEC data cannot be completely decoded in the downstream relay device (n + 1) 19, the decoded data is re-encoded and re-encoded for each packet unit. The FEC data is transmitted to the downstream relay device (n + 1) 19. Then, in response to the transmission stop request from the downstream relay device (n + 1) 19, the relay device (n) 18 stops transmission of FEC data and also stops decoding and re-encoding.

  When the packet copy 18a receives FEC data in units of packets from the network (N-1) 4, it transmits the FEC data in units of packets as it is to the relay apparatus (n + 1) 19 via the network (N), and also an FEC decoder. Send to 18b.

  The FEC decoder 18b decodes the FEC data in packet units and decodes the original data of the streaming content. Then, the FEC decoder 18b temporarily stores this original data in the buffer memory 18c. Further, the FEC decoder 18b determines whether or not the FEC data has been 100% decoded into the original data. When 100% decoding cannot be performed on the original data, the FEC decoder 18b transmits to the upstream relay device (n-1) 17 the FEC data re-encoded by the relay device (n-1) 17 to be transmitted. Send a request. Then, the FEC decoder 18b decodes the re-encoded FEC data from the relay device (n-1) 17 until the original data can be 100% decoded. At this time, the same re-encoded FEC data is repeatedly transmitted from the relay device (n−1) 17. On the other hand, if the original data is 100% decoded, the FEC decoder 18b transmits the FEC data re-encoded by the relay device (n-1) 17 to the upstream relay device (n-1) 17. A transmission stop request to be stopped is transmitted and the decoding is terminated.

  When the FEC encoder 18d receives a transmission request from the downstream relay device (n + 1) 19, the original data temporarily stored in the buffer memory 18c is converted into FEC data by FEC using the LT code based on the set FEC parameter. Re-encode. At this time, the FEC encoder 18d sets FEC parameters, but the FEC parameter setting method is the same as that in the first embodiment. Then, the FEC encoder 18d sends the re-encoded FEC data to the packet copy 18a in units of packets, and the re-encoded packet is transmitted from the packet copy 18a to the relay apparatus (n + 1) 19 via the downstream network (N) 5. The unit of FEC data is transmitted. At this time, the FEC encoder 18d continues to repeatedly transmit FEC data in the same packet unit until a transmission stop request is transmitted from the relay device (n + 1) 19. When the FEC encoder 18d receives a transmission stop request from the downstream relay device (n + 1) 19, the FEC encoder 18d erases the decoded data temporarily stored in the buffer memory 18c and ends the re-encoding. Note that the FEC encoder 18d may encode without changing the redundancy level.

  In the second embodiment, the transmission source 1 and the reception destination 2 perform the following processing in addition to the processing of the first embodiment. The transmission source 1 repeatedly transmits FEC data in units of packets in response to a transmission request from the relay device (n−1) 17 and stops transmission in response to a transmission cancellation request. In the destination 2, when the FEC data cannot be completely decoded into the original data, the transmission request is transmitted to the relay device (n + 1) 19 and the re-encoded FEC data is transmitted. When the original data is completely decoded by the re-encoded FEC data, a transmission stop request is transmitted to the relay device (n + 1) 19.

  The operation of the relay device (n) 18 will be described with reference to FIG. As a premise, the streaming content of 0.3 Mbps is encoded into FEC data in the transmission source 1, and the FEC data in packet units is transmitted over the network (S) 3. The packet-wise FEC data is received by the relay device (n−1) 17 via the network (S) 3. The relay device (n−1) 17 transmits the FEC data in packet units as it is to the relay device (n) 18 via the network (N−1) 4. Further, the relay device (n−1) 17 decodes the FEC data in packet units, re-encodes the decoded data in response to a transmission request from the relay device (n) 18, and re-encodes the FEC data. Data is transmitted to the relay device (n) 18 via the network (N-1) 4. The packet unit FEC data and the re-encoded FEC data are received by the relay device (n) 18.

  When receiving the packet-unit FEC data, the relay device (n) 18 transmits the packet-unit FEC data as it is to the relay device 19 via the network (N) 5 and decodes the packet-unit FEC data.

  Then, the relay device (n) 18 determines whether or not the FEC data has been completely decoded into the original data. When it is determined that the decoding cannot be completely performed, the relay device (n) 18 transmits a transmission request for transmitting the re-encoded FEC data to the relay device (n−1) 17, and the original data is completely transmitted. Decoding is performed using the re-encoded FEC data from the relay device (n−1) 17 until it can be decoded. On the other hand, if it is determined that the decoding has been completed completely, the relay device (n) 18 transmits a transmission stop request for stopping the transmission of the re-encoded FEC data to the relay device (n-1) 17, and the decoding is performed. End conversion.

  Further, when receiving the transmission request from the relay device (n + 1) 19, the relay device (n) 18 re-encodes the data decoded by FEC using the LT code into FEC data. Then, the relay device (n) 18 transmits the re-encoded FEC data to the relay device (n + 1) 19 via the network (N) 5 in units of packets. In addition, when the relay device (n) 18 receives the transmission stop request from the relay device (n + 1) 19, the relay device (n) 18 deletes the decoded data temporarily stored in the buffer memory 18c and ends the re-encoding.

  Thereafter, the FEC data in units of packets and the re-encoded FEC data according to the request from the relay device (n + 1) 19 are received by the relay device (n + 1) 19 via the network (N) 5. The relay device (n + 1) 19 transmits the packet-unit FEC data as it is to the receiver 2 via the network (D) 6. Further, the relay device (n + 1) 19 decodes the FEC data in packet units, re-encodes the decoded data in response to the transmission request from the receiver 2, and transmits the re-encoded FEC data to the network (D ) To the receiver 2 via 6. Then, the FEC data in units of packets and the re-encoded FEC data are received by the receiving destination 2. The reception destination 2 sequentially decodes the received FEC data in units of packets. If the FEC data cannot be completely decoded, the FEC data is decoded using the re-encoded FEC data. Then, the receiver 2 reproduces the streaming content using the decrypted data. In this decoding, since the redundancy is improved by completely decoding the FEC data in the relay apparatuses 17, 18, and 19, no error such as a loss packet or a macro block error occurs in the streaming content.

  According to the relay devices 17, 18, and 19, since the FEC data is transferred downstream as it is and the FEC data is decoded, the delay can be suppressed as much as possible. Further, since the relay devices 17, 18, and 19 supplement the FEC data re-encoded from the upstream until the FEC data can be 100% decoded into the original data, the original data can be decoded 100% and eventually received. The original data can be decrypted 100% in the destination 2. Therefore, it is not necessary for the transmission source 1 to set the redundancy level in consideration of the QoS information of the entire network, and the transmission data transmission band can be suppressed.

  As mentioned above, although embodiment which concerns on this invention was described, this invention is implemented in various forms, without being limited to the said embodiment.

  For example, in this embodiment, the present invention is applied to the case where streaming content is distributed from an ISP or the like to individuals. However, in the case of individuals and individuals, applications are also communicated via various networks in the case of telephones, competitive games, and the like. It is applicable to.

  In the present embodiment, all the relay devices on the network are configured to perform the processing according to the present invention. However, some relay devices on the network may be configured to perform the processing according to the present invention. Even when processing is performed in all the above relay devices, a relay device having a sufficient processing capacity may be configured to perform processing preferentially.

  In this embodiment, the LT code is used as the FEC, but other FECs such as a Hamming code, a BCH code, a Reed-Solomon code, and a tornado code may be used.

  In this embodiment, the case of one route from the transmission source to the reception destination has been described. However, the present invention can also be applied to a case where the relay device or the reception source branches to a plurality of routes (networks). An optimum route is also determined based on QoS information of each route (network) to be branched.

  In this embodiment, the service level (service type) is set in the IP header. However, the service level (service type) may be set in another fixed-length data area such as a UDP header. .

  Further, although the network QoS information is directly used in the present embodiment, secondary information such as four arithmetic operations may be used for arbitrary information in the QoS information.

It is a block diagram of the whole network which concerns on the 1st Embodiment of this invention. It is explanatory drawing of the packet which concerns on embodiment of this invention, (a) is a packet block diagram, (b) is a data block diagram of an IP header. It is a conceptual diagram of the FEC LT code. It is explanatory drawing of the parameter of FEC which concerns on embodiment of this invention. It is a block diagram of the whole network based on the 2nd Embodiment of this invention. It is a block diagram of the whole conventional network.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Transmission origin, 1a ... FEC encoder, 2 ... Reception destination, 2a ... FEC decoder, 2b ... Streaming viewer, 3, 4, 5, 6 ... Network, 7, 8, 9, 17, 18, 19 ... Relay apparatus, 7a, 8a, 9a ... packet switch, 7b, 8b, 9b, 17b, 18b, 19b ... FEC decoder, 7c, 8c, 9c, 17c, 18c, 19c ... buffer memory, 7d, 8d, 9d, 17d, 18d, 19d ... FEC encoder, 17a, 18a, 19a ... Packet copy

Claims (14)

A relay device that is provided at one or more network connection points used when transmitting data from a transmission unit to a reception unit, receives data from an upstream device, and transmits the data to a downstream device. And
A first transfer that decodes forward error correction encoded packet-unit received data from the upstream apparatus, re-encodes the decoded data, and transmits the re-encoded data to the downstream apparatus Means,
Second transfer means for transferring the packet-unit received data that has been forward error correction encoded from the upstream device as it is to the downstream device
With
When receiving forward error correction encoded packet unit data from the upstream side apparatus, the transfer by the second transfer means is executed and the decoding by the first transfer means is started. Relay device. Receiving means for receiving a transmission stop request for the first transfer means from the downstream device;
2. The relay according to claim 1, wherein the first transfer unit executes transmission of the re-encoded data to the downstream apparatus until a transmission stop request is received from the downstream apparatus. apparatus. 3. The relay according to claim 2, wherein the downstream device transmits a transmission stop request to the first transfer unit when the received data transferred by the second transfer unit can be completely decoded. apparatus. The downstream apparatus cannot completely decode the received data transferred by the second transfer unit, and has received the re-encoded data transmitted by the first transfer unit following the received data. 4. The relay apparatus according to claim 3, wherein, when the re-encoded data can be completely decoded, a transmission stop request to the first transfer unit is transmitted. The first transfer means has a buffer for temporarily storing the decoded data, holds the decoded data in the buffer until the transmission stop request is received, and receives the transmission stop request, The relay apparatus according to any one of claims 2 to 4, wherein the decrypted data held in the buffer is erased. A determination means for determining whether or not the received data in units of packets that are forward error correction encoded from the upstream device can be completely decoded;
When it is determined by the determination means that the received data in the forward error correction encoded packet unit received from the upstream side apparatus can be completely decoded, the received data is followed by the improved redundancy compared to the received data. The relay apparatus according to claim 1, wherein the transmission stop request for transmitted data is transmitted to the upstream apparatus. Determining means for determining whether or not to improve the redundancy of data and the redundancy level based on service quality information of one or a plurality of networks located downstream of the relay device or secondary information of the service quality information; Prepared,
Depending on the application and / or service level of each flow, determination of necessity of forward error correction, determination of type of forward error correction, determination of redundancy level and / or setting of parameters for forward error correction The relay apparatus according to claim 1, wherein the relay apparatus performs the following. 8. The redundancy of data is improved by adding and re-encoding the insufficient redundancy without decoding all the data in the middle of decoding of the block of the flow. Relay device. 9. The relay apparatus according to claim 7, wherein an application and / or service level is set in a fixed-length data area of the packet. 9. The relay apparatus according to claim 7, wherein a packet in which an application and / or service level is set in a fixed-length data area of the packet is handled. The relay apparatus according to claim 1, wherein the type of forward error correction data is an LT code. An autonomous error correction network consisting of a transmission unit, a reception unit, and one or a plurality of relay devices provided at a connection point between one or a plurality of networks,
The relay device decodes received data in units of packets that have been forward error correction encoded from the transmitter or the upstream relay device, re-encodes the decoded data, and re-encodes the data First transfer means for transmitting the data to the downstream relay device or the reception unit, and the forward error correction encoded packet unit received data from the transmission unit or the upstream relay device as it is to the downstream side A second transfer unit configured to transfer to the relay device or the reception unit; and when receiving data in units of packets subjected to forward error correction coding from the transmission unit or the upstream relay device, the second transfer unit The autonomous error correction network according to claim 1, wherein the transfer by the first transfer means is executed and the decoding by the first transfer means is started. The relay device includes a receiving unit that receives a transmission stop request for the first transfer unit from the relay device or the receiving unit on the downstream side, and the first transfer unit is configured to receive the relay device or the reception unit on the downstream side. 13. The autonomous error correction network according to claim 12, wherein transmission of the re-encoded data to the relay device or the receiving unit on the downstream side is executed until a transmission stop request from the unit is received. . In the relay device, depending on the application and / or service level for each flow, determination of necessity of forward error correction, determination of type of forward error correction, determination of redundancy level and / or forward error 14. The autonomous error correction network according to claim 12, wherein correction parameters are set.

Images