JP4395040B2 - Communication apparatus and error correction packet insertion ratio control method - Google Patents

Description

  The present invention relates to a communication device and an error correction packet insertion ratio control method, and in particular, distributes data packets and error correction packets from a transmission side to a network, receives those packets on a reception side, and performs image processing in real time. The present invention relates to a communication apparatus and an error correction packet insertion ratio control method that are preferably applied to a communication system that performs processing.

  FIG. 7 shows a configuration example of a communication system to which the present invention is applied. As shown in the figure, an MPEG stream packet (hereinafter referred to as an MPEG stream packet) output from the encoder 7-2 is encoded by an encoder 7-2 with a moving image signal captured by the camera 7-1. "Data packet") is distributed over the network 7-3 and distributed in real time, and the receiving side receives the data packet and decodes and outputs the moving image / sound output by the decoder 7-4. 5 to display / output.

  FIGS. 8A and 8B show an outline of the device configuration of the encoder and decoder. The encoder has an NTSC / PAL video / audio signal input unit and an IP interface, encodes the input video / audio signal into MPEG data by the encoding unit 8-1, and transmits the transmission unit 8-2. 1 generates an error correction packet (hereinafter referred to as “FEC packet”) from the data packet of MPEG data, and distributes the data packet and the FEC packet on the network.

  The decoder has an NTSC / PAL video / audio signal output unit and an IP interface. The decoder 8-4 decodes the data packet received from the network by the decoder 8-4, and restores it by the decoding. Output the image / sound signal. Further, an error correction process is performed on the data packet in which an error failure has occurred in transmission on the network by using the FEC packet in the receiving unit 8-3 to restore the data packet.

  Next, an error correction function using FEC packets will be described. In the communication system described above, the data packet distributed from the encoder often causes an error failure in the network and cannot be normally received by the decoder. Therefore, there is an FEC function as a function for performing error correction using an FEC packet.

  As shown in FIG. 9A, the FEC function performs an exclusive OR operation on a plurality of data packets and transmits an FEC packet storing the exclusive OR information, as shown in FIG. A plurality of data packets are created at a certain rate and distributed to the network. In the decoder, as shown in FIG. 9B, the error packet 3 is recovered by performing an exclusive OR operation on the received data packet other than the error packet 3 and the FEC packet, for example. Do.

  However, error correction cannot be performed for the occurrence of an error packet that exceeds the FEC packet insertion rate. For example, when FEC packets are inserted at a ratio of 1 to 10 data packets, recovery is not possible if an error occurs in 2 or more packets among 11 packets.

  Next, the MPEG stream packet configuration will be described. FIG. 10 shows the configuration of a data packet distributed from the encoder. A data packet is roughly divided into two elements, an RTP header part and a payload part, and sequence number information indicating the delivery order of the packets is stored in the RTP header. The data size of this sequence number is 16 bits, and the sequence number is cyclically assigned in the range of 0 to 65535. Further, the reception unit of the decoder recognizes an error (loss) of the received packet based on the information of the sequence number.

  FIG. 11 shows a format configuration of an FEC packet distributed from the encoder for error correction. An FEC packet is roughly divided into three elements, an RTP header part, an FEC header part, and a payload part. LengthRecovery, PTR (PayloadTypeRecovery), TimestampRecovery, and FECPayload in the FEC header store operation result information obtained by performing an exclusive OR operation on the packet size of the data packet, PayloadType, Timestamp, and the payload portion, respectively.

  Mask has a data size of 24 bits, and here, the same number of bits “1” as the number of data packets subjected to exclusive OR operation is set. For example, when an exclusive OR operation is performed on 10 data packets, a bit pattern of “00000000000011111111111 (0x0003FF)” is stored.

  Further, similarly to the data packet, information on a sequence number (0 to 65535 number independent of the sequence number of the data packet) indicating the delivery order of the FEC packet is stored in the RTP header of the FEC packet. In the decoder, the receiving unit recognizes an error (loss) of the FEC packet.

  Conventionally, FEC packets are distributed at a fixed insertion rate set in advance by a maintenance person, and always use a fixed bandwidth. Therefore, in a network where the error occurrence rate is small relative to the FEC packet insertion rate, bandwidth is wasted, and conversely, if an error that exceeds the FEC packet insertion rate temporarily occurs, Packets cannot be recovered, and in this case as well, useless bandwidth due to invalid FEC packets is used.

  As described above, the distribution of FEC packets at a fixed insertion rate set by the maintainer is inefficient in terms of effective use of bandwidth, and further requires network monitoring by the maintainer. It had been. In response to such a problem, the following patent document proposes a technique for notifying the transmission side of error occurrence information from the reception side when an error occurs and performing data transmission control according to the error occurrence state on the transmission side. Yes.

The following Patent Document 1 describes a communication apparatus that transmits an ECC check bit for each transmission data unit length that changes according to an increase / decrease in error based on an error notification related to transmission data returned from the receiving side. ing. Further, Patent Document 2 calculates an index such as a packet loss rate and a transmission delay fluctuation from the monitor information of the communication state monitor unit in the receiving terminal, receives the data in the transmission control unit in the data server, and receives the data from the index. A data transmission device that determines transmission parameters and reduces degradation of reproduction quality is described. Patent Document 3 describes a speech coding apparatus that transmits on the transmission side by changing the bit allocation of speech codes and error correction codes according to the state of the error rate of the transmission path, and suppresses degradation of speech quality. Yes.
JP 2001-24627 A Japanese Patent Laid-Open No. 11-177623 Japanese Patent Laid-Open No. 9-297599

  As described above, conventionally, FEC packets are distributed at a fixed insertion rate set in advance by the maintenance person, or error occurrence information is notified from the reception side to the transmission side when an error occurs, and an error occurs on the transmission side. A technique for controlling transmission of error correction data according to the situation has been proposed. However, conventionally, sufficient countermeasures have not been taken against fluctuations in the error occurrence rate that changes every moment on the network.

  For this reason, a network with a low error rate relative to the FEC packet insertion rate uses a bandwidth wastefully, and if the error rate fluctuates, the error correction packet delivered from the transmission side cannot perform error correction. In some cases, there is a problem that the bandwidth is wasted due to an inappropriate number of error correction packets to be distributed. Also, when error information is sent from the receiving side to the sending side and error correction data transmission control is performed on the sending side according to the error status, error information is sent to the sending side every time the error rate is observed on the receiving side. In order to notify, a fixed bandwidth was always used, and effective use of the bandwidth was not attempted.

  According to the present invention, by performing transmission control that inserts an appropriate number of error correction packets in response to fluctuations in the error rate on the network, conventionally, error correction processing cannot be performed due to fluctuations in the error rate. An object of the present invention is to prevent useless bandwidth use due to transmission of such an error correction packet.

  The communication apparatus of the present invention (1) receives a data packet and an error correction packet distributed from the transmission side to the network, and recovers the error packet generated on the network by using the error correction packet. An error information storage means for storing a plurality of error rates of received packets observed over a plurality of times at predetermined time intervals, and a plurality of errors stored in the error information storage means Based on the rate information, error correction packet insertion rate determination means for determining an optimal error correction packet insertion rate, and information on the error correction packet insertion rate determined by the error correction packet insertion rate determination means to the transmission side Error information notifying means for notifying is provided.

  (2) The error correction packet insertion ratio determining means selects the error rate information of the largest value from the plurality of error rate information stored in the error information storage means, and the largest value Based on the error rate information, means for determining the insertion rate of error correction packets is provided.

  (3) The error information storage means includes means for storing observation information of the latest error rate at predetermined time intervals and discarding the oldest error rate information. Further, (4) the error information notifying means compares the error correction packet insertion rate notified last time with the error correction packet insertion rate determined this time for each predetermined time interval, and changes between the two. Only when there is, there is provided means for notifying the transmission side of the information of the error correction packet insertion ratio.

  The error correction packet insertion ratio control method according to the present invention includes (5) a data packet and an error correction packet distributed from the transmission side to the network, and an error generated on the network using the error correction packet on the reception side. In a communication system that recovers a packet and restores a data packet, the receiving side stores a plurality of error rates of received packets observed over a plurality of times at predetermined time intervals, and stores the plurality of stored errors. The error rate information of the largest value is selected from the rate information, the error correction packet insertion rate is determined based on the error rate information of the largest value, and the error correction packet insertion rate information thus determined Is transmitted to the transmission side, and the transmission side sends an error correction packet in accordance with the error correction packet insertion rate information notified from the reception side. And transmits inserted between Tapaketto.

  According to the present invention, the error rate information is statistically managed on the receiving side with respect to fluctuations in the error rate, and the optimal insertion rate of the error correction packet is determined based on the statistical information on the error rate. By determining, it becomes possible to control the insertion of the optimum number of error correction packets corresponding to the fluctuation of the error occurrence rate. As a result, invalid error correction packet insertion control processing due to error rate fluctuations and useless bandwidth use due to transmission of the error correction packet are improved.

  In the present invention, in order to cope with fluctuations in the error rate on the network, error rate information is statistically managed on the receiving side, and an FEC that is an error correction packet is based on the statistical error rate information. Control the rate of packet insertion. Hereinafter, (1) an error management table, (2) an error information processing flow, and (3) an information notification packet for performing these controls will be described in detail. Hereinafter, the error occurrence rate is simply referred to as an error rate.

(1) Error management table An error management table as shown in FIG. 1 is provided in the decoder reception unit for statistically managing error rate information collectively. In the error management table, the data packet reception number (N_DATA) indicates the number of received data packets received at said decoders, FEC packet reception number (n_fec), the receiving of the FEC packets received by said decoders Indicates a number.

Packet error number (n_error) indicates the total value of the number of error packets of data packets and FEC packets to be received by the decoders. Each of these packets is counted up according to the packet reception processing flow shown in FIG. The packet reception processing flow in FIG. 2 will be described in detail later.

  The information notification interval (t_cycle) in the error management table indicates the time interval of error information notification to the encoder. For each time interval, the number of received data packets (n_data), the number of received FEC packets (n_fec), and the packet The information on the number of errors (n_error) is cleared, and error information for the current cycle is collected.

  The error rate protection stage number (N) in the error management table indicates the holding number when each error rate information collected for each time interval is held. The previous information notification time (t_pre) in the error management table indicates the time when error information notification is performed to the encoder.

  The setting insertion rate (fec_set) in the FEC insertion rate column in the error management table indicates the FEC packet insertion rate notified to the transmission side in the previous cycle. Similarly, the calculated value insertion rate (fec_cal) in the FEC insertion rate column is The optimum FEC packet insertion ratio of the current cycle calculated from each error rate of the retained error rate protection stage number (N) is shown.

  The error rates e_1 to e_N in the error information column in the error management table are error rate information for the number of error rate protection stages (N) collected at each information notification interval (t_cycle). When calculating the FEC packet insertion ratio, the worst error rate is selected from the N error rates e_1 to e_N held.

(2) Information notification processing flow In order to notify the encoder of the FEC packet insertion rate information from the decoder to the encoder, the information in the error management table is used in the decoder according to the error information notification processing flow shown in FIG. The FEC packet insertion ratio corresponding to the fluctuation of the error rate is automatically calculated, and an information notification packet including the calculation result is transmitted to the encoder. The error information notification processing flow of FIG. 3 will be described in detail later.

(3) Error Information Notification Packet In order to notify the information on the FEC packet insertion ratio from the decoder to the encoder, for example, an RTCP Application defined (APP) packet shown in FIG. 4 is used. The ASCII character “RINF” is stored in Name in the packet, and information (1 to 24) of the calculated FEC packet insertion ratio is stored in FecRate.

  FIG. 5 shows an example of changing the FEC packet insertion ratio according to the present invention. Now, as shown in FIG. 5 (a), in a state where FEC packets are transmitted at a rate of 1 for 10 data packets, numbers 3 and 8 are shown as shown in FIG. 5 (b). Suppose an error occurs in two packets.

  The decoder calculates the error rate from the received packet, performs fluctuation protection over a plurality of stages to cope with fluctuations in the error rate, and determines the optimum FEC packet insertion rate. Information on the insertion ratio is notified to the encoder by an RTCP packet. The encoder changes the FEC packet insertion rate according to the FEC packet insertion rate information notified from the decoder, and distributes the FEC packet with a new insertion rate as shown in FIG.

  Here, the packet reception processing flow of FIG. 2 will be described in detail below. When a packet arrives at the receiving unit of the decoder from the network, a packet receiving process is first performed (step 2-1). Next, it is determined whether the received packet is a data packet or an FEC packet based on the information of PayloadType in the RTP header of the received packet (step 2-2).

  If the received packet is a data packet, the number of received data packets (n_data) is counted up (step 2-3), and the count value is stored in the error management table. If the received packet is an FEC packet, the number of received FEC packets (n_fec) is counted up as in the data packet, and the count value is stored in the error management table (step 2-4).

  Next, the occurrence of an error (loss) is checked by comparing and comparing the sequence number of the previously received packet with the sequence number of the currently received packet (step 2-5). If an error (loss) has occurred, the number of packet errors (n_error) is counted up and the count value is stored in the error management table (step 2-6). In this manner, processing is performed according to the flow of FIG. 2 every time a packet is received, information on the number of received data packets, the number of received FEC packets, and the number of packet errors is collected and stored in the error management table.

  In addition to the packet reception process described above, the decoder reception unit determines the FEC packet insertion ratio and performs the information notification process to the encoder according to the error information processing flow of FIG. First, an information notification interval (t_cycle) is set (step 3-1). Next, the error rate protection stage number (N) is set (step 3-2).

  Next, the previous information notification time (t_pre) in the error management table is compared with the current time (step 3-3), and whether or not a time longer than the set information notification interval (t_cycle) has elapsed. Until the time equal to or longer than the information notification interval (t_cycle) has elapsed, the process waits without proceeding to the next process.

When a time equal to or longer than the information notification interval (t_cycle) has elapsed from the previous information notification time (t_pre), the number of received data packets (n_data), the number of received FEC packets (n_fec), and the number of packet errors ( From the information of n_error), an error rate is calculated by the following equation (1) (step 3-4).
Error rate = {n_error / (n_data + n_fec + n_error)} × 100
... (1)

  The error rate calculated by the above equation (1) is stored as error information in the error management table (step 3-5). However, the information of the error rates e_1, e_2 to e_N-2, e_N-1 that have already been stored is shifted and stored in the information column of e_2, e_3 to e_N-1, e_N, respectively, and the latest error is stored. The rate is stored in the information column of e_1.

  In this way, information on the error rates e_1 to e_N corresponding to N error rate protection stages is stored. At this time, the previous oldest error rate e_N is discarded. That is, the latest error rate e_1 is stored in the error management table every information notification interval (t_cycle), and the oldest error rate e_N is discarded from the error management table.

  Next, the time when the error information is stored is stored as the previous information notification time (t_pre), and the previous information notification time (t_pre) is updated (step 3-6). Next, in order to obtain the error rate of the information notification interval (t_cycle) of the next cycle, information on the number of received data packets (n_data), the number of received FEC packets (n_fec), and the number of packet errors (n_error) in the error management table Is initialized (step 3-7).

Next, the worst error rate e_x is selected from the information of the error rates e_1 to e_N (step 3-8). This process enables protection against temporary error rate fluctuations. Next, the calculated value insertion ratio (fec_cal) is calculated by the following equation (2) based on the selected worst error rate e_x (step 3-9). However, in equation (2), the part after the decimal point is ignored, and the calculation result is stored in the information column of the calculation value insertion ratio (fec_cal) in the error management table.
Calculated value insertion ratio (fec_cal) = (100 / e_x) −1 (2)

Next, a comparison is made between the setting insertion ratio (fec_set) in the error information management table and the calculated value insertion ratio (fec_cal) (step 3-10). If the values are the same, the subsequent processing is not performed. Then, returning to the above-described step 3-3, the same processing is performed. When the setting insertion ratio (fec_set) and the calculated value insertion ratio (fec_cal) are different, the calculated value insertion ratio (fec_cal) is stored in the setting insertion ratio (fec_set), and the setting insertion ratio (fec_set) is stored. to update (step 3-11). Next, the information notification packet storing the setting insertion ratio (fec_set) in FecRate in the RTCP APP packet shown in FIG. 4 is distributed to the encoder (step 3-12).

  FIG. 6 shows a specific example of an error rate transition graph. FIG. 6 shows a transition graph of the error rate with respect to time, with the horizontal axis representing time elapsed (sec) and the vertical axis representing error rate (%). The present invention will be described in the case where the error rate transitions with respect to time as in the graph shown in FIG. However, the information notification interval (t_cycle) = 1 sec and the error rate protection stage number N = 5.

  In FIG. 6, the error rate is 25% in 0 to 1 second of the start portion of the graph, and 25% is stored as the error rate e_1 according to the error information notification processing flow of FIG. Value insertion ratio (fec_cal) = 3. Since it is in the initial state, the setting insertion ratio (fec_set) = 0, and the calculated value insertion ratio (fec_cal) is different from the setting insertion ratio (fec_set), so the setting insertion ratio (fec_set) is updated to 3. Then, an information notification packet storing 3 in FecRate in the RTCP APP packet shown in FIG. 4 is transmitted to the encoder. The encoder distributes the FEC packet by setting the FEC packet insertion ratio to 3 upon reception of the information notification packet.

  Next, the error rate is 23% for 1 to 2 seconds in the graph, and 23 is stored in the error rate e_1 and 25 is stored in the error rate e_2 when processing is performed according to the error information notification processing flow of FIG. The worst error rate is 25% of e_2, and the calculated value insertion ratio (fec_cal) and the setting insertion ratio (fec_set) are the same value, so information is not notified to the encoder.

  Next, the error rate is 20% for 2 to 3 seconds in the graph. When processing is performed according to the error information notification processing flow of FIG. 3, 20 is stored in the error rate e_1, 23 is stored in the error rate e_2, and 25 is stored in the error rate e_3. The worst error rate is 25% of e_3, and information notification is not performed as in the previous case.

  Next, the error rate is 24% in 3 to 4 seconds of the graph, and when processed according to the error information notification processing flow of FIG. 3, the error rate e_1 is 24, the error rate e_2 is 20, the error rate e_3 is 23, the error rate 25 is stored in e_4. The worst error rate is 25% of e_4, and information notification is not performed as in the previous case.

  Next, the error rate is 24% for 4 to 5 seconds in the graph, and when processed according to the information notification processing flow of FIG. 3, the error rate e_1 is 24, the error rate e_2 is 24, the error rate e_3 is 20, and the error rate e_4 is 23, 25 is stored in the error rate e_5. Since the worst error rate is 25% of e_5, information notification is not performed as in the previous case.

  Here, assuming that the FEC insertion ratio is always changed by the information notification packet at intervals of 1 second as in the past with respect to the error rate transition of 0 to 5 seconds, the error rate in 2 to 3 seconds is 20 Therefore, the FEC packet insertion ratio is set to 5. However, since the error rate is 24% for the next 3 to 4 seconds, the state where error recovery by the FEC packet cannot be performed continues. This is because the FEC packet insertion ratio is changed without dealing with fluctuations in the error rate, and a wasteful bandwidth is used by transmitting an invalid number of FEC packets in 3 to 4 seconds.

  However, in the present invention, in order to cope with fluctuations in the error rate, the worst error rate 25% is selected from the error rates e_1 to e_5 of the plurality of stages and the FEC packet insertion rate is determined. Recovery by the FEC packet is always possible for the error of the received packet for 2 seconds, and useless bandwidth use by the FEC packet is reduced.

  Similarly, according to the present invention, the recovery by the FEC packet is always possible even for 5 to 30 seconds. On the other hand, when the FEC packet insertion rate is always set by the error rate of each time as in the prior art, When the error rate at the time of current reception is higher than the error rate of the FEC packet insertion rate set immediately before, such as 11 seconds, 14-15 seconds, 18-19 seconds, 24-25 seconds, 27-28 seconds, etc. In addition, since recovery using FEC packets cannot be performed, useless bandwidth due to transmission of an invalid number of FEC packets is used.

It is a figure which shows the error management table of this invention. It is a figure which shows the packet reception processing flow of this invention. It is a figure which shows the error information notification processing flow of this invention. It is a figure which shows the format of the information notification packet of this invention. It is a figure which shows an example of the change of the FEC packet insertion ratio by this invention. It is a figure which shows the specific example of the transition graph of an error rate. It is a figure which shows the structural example of the communication system with which this invention is applied. It is a figure which shows the outline | summary of the apparatus structure of an encoder and a decoder. It is explanatory drawing of the error correction function using a FEC packet. It is a figure which shows the format of an MPEG stream packet. It is a figure which shows the format of a FEC packet.

Explanation of symbols

n_data Number of received data packets n_fec Number of received FEC packets n_error Number of packet errors t_cycle Information notification interval N Number of error rate protection stages t_pre Previous information notification time fec_set Insertion rate during setting fec_cal Calculation value insertion rate e_1 to e_N Error rate

Claims (4)

In a communication apparatus that receives a data packet and an error correction packet distributed on a network from a transmission side, and recovers an error packet generated on the network by using the error correction packet to restore a data packet.
Error information storage means for storing a plurality of error rates of received packets observed over a plurality of times at predetermined time intervals;
The error rate information of the largest value is selected from the plurality of error rate information stored in the error information storage means, and the insertion rate of the error correction packet is determined based on the error rate information of the largest value. An error correction packet insertion ratio determining means for determining;
A communication apparatus comprising: error information notifying means for notifying the transmitting side of information on the error correction packet insertion ratio determined by the error correction packet insertion ratio determining means. 2. The communication apparatus according to claim 1 , wherein the error information storage means includes means for storing observation information of the latest error rate at every predetermined time interval and discarding the oldest error rate information. . The error information notifying means compares the error correction packet insertion rate notified last time with the error correction packet insertion rate determined this time every predetermined time interval, and only when there is a change between the two the communication apparatus according to claim 1 or 2, characterized in that it comprises means for notifying the information of the error correction packet insertion ratio to the sender. In a communication system in which a data packet and an error correction packet are distributed from the transmission side to the network, and the error packet generated on the network is recovered by using the error correction packet on the reception side to restore the data packet.
On the receiving side, a plurality of error rates of received packets observed over a plurality of times at predetermined time intervals are stored, and the error rate having the largest value among the stored information on the plurality of error rates is stored. Information is selected, based on the error rate information of the largest value, the error correction packet insertion rate is determined, the determined error correction packet insertion rate information is notified to the transmission side,
An error correction packet insertion ratio control method, wherein a transmission side inserts an error correction packet between data packets according to the error correction packet insertion ratio information notified from the reception side.

Images