Error Protection Method for Multimedia Data

The present invention relates to an error prevention method for multimedia, and more particularly, to a method for improving data recovery and channel throughput in channels wherein a random error and a burst error occur by using a rate compatible punctured convolutional code (RCPC) and an automatic retransmission on request (ARQ). Further, this invention has been adopted by the ITUT/SG16/Q11Mobile Group.

Let us consider multimedia terminals which transmit and receive arbitrary packets of data (video, audio, data, or a mixed form of any of those three). The transmitter transmits information packets, for example, I, J, and other packets. For each information packet, the transmitter forms N-sized bit-streams which are different representations of given information packets. For example, the transmitter can generate a packet A (B, C, or D) for given information packet Type-1 and Type-2 are different in that they use different retransmission methods. The packets to be transmitted are formed using either convolutional code or RCPC.

FIG. 1 is a block diagram showing a general situation of data transmitted and received using ARQ. The basic concept of Type-1 ARQ will be described as follows with reference to FIG. 1. When a transmitter transmits a packet A having a length of N, a packet decoder 120 in a receiver starts decoding the received packet A 110. At this time, if errors occur in the packet A and no further decoding is possible, for example, channel coding is not employed, channel coding having a 1-bit error or more is employed or, more errors than a channel coder can detect and correct occur, the receiver asks the transmitter to send the same packet A again. Here, retransmission would be repeated either until the decoder 120 receives an error-free packet A, or for some specific number of iterations to perform transmission and receiving with respect to the next packet. Type-1 ARQ is very effective in burst-error containing channel. Next, Type-2 ARQ will be described. Up to now, there are three types of Type-2 ARQ, i.e., a basic type, a Class A and a Class B, each of which uses RCPC given information I (J,K, . . . ).

FIG. 2 is a conceptual diagram showing the operation of the basic type, wherein arrows represent combination. Here, given information I, the transmitter generates packets A and B using RCPC at a rate of 1/2 and transmits only the packet A. The decoder in the receiver attempts to decode the packet A. If successful, the decoder then attempts to decode the first packet of two for the next information J. Otherwise, the receiver asks the transmitter to send the packet B. Also, the decoder attempts decoding a combination of packets A and B. If successful, the decoder then attempts to decode the first packet of two for the next information J. Otherwise, the receiver asks the transmitter to send the packet A again and all of these procedures are repeated. The basic type has an advantage in that implementation is not so complicated.

FIG. 3 is a conceptual diagram showing the operation of the Class A packet (Lin-Yu), wherein * denotes self-decoding and arrows represent combination. The operational principle thereof is similar to the basic type except how to combine packets A and B when both packets fail to be decoded. That is, the decoder attempts decoding the combination of packets A and B, and if it fails, the receiver asks the transmitter to send the packet A again. Next, if the decoder succeeds decoding only the packet A, the next information J is processed, and if the decoder fails, the receiver combines the previously stored packet B and currently received packet A (i.e., in general, interleaves the two) to attempt decoding. This method is more effective in a random error containing channel rather than in a burst error containing channel.

Next, Class B is significantly more complicated than the basic type and the Class A. The basic concept thereof is based on the Class A. First, the Class A (Lin-Yu) is performed by generating the packets A and B given the information I using RCPC at a rate of 1/2. As described above, the Type-1 ARQ is greatly effective in the burst error containing channel. However, with Type-1 ARQ, retransmission would be more frequent in the random error containing channel, which causes drastically lower channel throughput. Even though Type-2 ARQ allows good performance in the random error containing channel, retransmission would be more frequent in the burst error containing channel; therefore, channel throughput can be lowered.

It is an object of the present invention to provide a method for maintaining channel throughput at a certain level in a random error containing channel and a burst error containing channel by operating like Type-1 in the burst error containing channel, while operating similarly to a basic type or a Class A of Type-2 in the random error containing channel.

To accomplish the above object, there is provided an error prevention method in a method for decoding a plurality of packets of given information, comprising the steps of a) decoding one of the plurality of packets, b) decoding another packet when an error occurs in the decoding in step a), c) decoding a combination of the decoding error packets when an error occurs in step b) or the third packet, and d) repeating step c) until the decoding error no longer occurs.

The above object and advantage of the present invention will become more apparent by describing in detail a preferred embodiment thereof with reference to the attached drawings in which:

FIG. 1 is a block diagram showing the general situation of data transmission and reception using an ARQ method;

FIG. 2 is a conceptual diagram showing the operation of a basic type;

FIG. 3 is a conceptual diagram showing the operation of a Class A;

FIG. 4 is a block diagram showing the structure of an apparatus for realizing error prevention according to the present invention;

FIG. 5 is a conceptual diagram showing a procedure for processing of received packets A, B, C and D in a decoder of a receiver shown in FIG. 4; and

FIG. 6 is a flow chart showing a procedure for processing received packets in a decoder according to the present invention.

The present invention is a method for performing a hybrid-type ARQ which mixes Type-1 and Type-2 methods.

Referring to FIG. 4, an error prevention apparatus includes: a transmitter provided with a packet buffer 430 for producing packets A, B, C and D using an RCPC 420 having a rate of 1/4 for given information packet I 410; an inverse RCPC 440; and a receiver provided with a buffer 450 for storing the received packet, for sending an ARQ and a packet number to the transmitter via a channel. In FIG. 4, an RCPC logic is fixed at a rate of 1/4. A portion comprised of four arbitrary polynomials which satisfies a locally invertible characteristic simultaneously produces RCPC-processed packets A, B, C and D. Also, the transmitter has a maximum of four retransmissions. Here, the local inversion in the RCPC denotes that original information I can be obtained with any one of the packets A and B and with a combination of the packets A and B.

FIG. 5 is a conceptual view showing a procedure for processing the received packets A, B, C and D in the decoder of the receiver shown in FIG. 4, wherein * indicates self-decoding and a bracket indicates a combination of packets (generally, an interleaving operation).

FIG. 6 is a flow chart outlining a procedure for processing the received packets in a decoder according to the present invention.

As shown in FIG. 6, the transmitter produces packets A, B, C and D using the RCPC 420 in step 612. The first packet A is transmitted to the receiver in step 614. The decoder attempts decoding the packet A in step 616. In step 616, if the packet A is decoded, the decoded results are stored in the buffer 450 (step 642) and the job for other information (e.g., information J) proceeds (step 644); otherwise, an ARQ signal is sent to the transmitter to request transmission of the packet B (step 618). The decoder attempts decoding only the packet B in step 620. If successful, the decoded results are stored in the buffer 450 in step 642 and the job for other information (e.g., information J) proceeds in step 644. If the decoder fails, it then attempts to decode the combination of the packets A and B in step 622, which is indicated by *AB in FIG. 5. At this time, if the combination of the packets A and B as shown in FIG. 5 are decoded, the decoded results are stored in the buffer 450 (step 642) and the job for other information (e.g., information J) proceeds (step 644). If the combination of the packets A and B is not decoded, the transmitter is requested to transmit the packet C by sending the ARQ signal thereto in step 624. The decoder then attempts decoding only packet C in step 626. If successful, the decoded results are stored in the buffer 450 in step 642 and the job for other information (e.g., information J) proceeds in step 644. Otherwise, the decoder attempts to decode the combination of the packets B and C in step 628, which is indicated by *BC in FIG. 5. If the combination of the packets B and C is successfully decoded, the decoded results are stored in the buffer 450 (step 642) and the job for other information (e.g., information a) proceeds (step 644). If the combination thereof is not decoded, the combination of packets B and C is combined with packet A as shown in FIG. 5 as *ABC and decoding is attempted in step 630. Here, if the combination of packets A, B and C is decoded, the decoded results are stored in the buffer 450 (step 642) and the job for other information (e.g., information J) proceeds (step 644). Otherwise, the transmitter is requested to transmit the packet D by sending the ARQ signal thereto in step 632. The decoder then attempts decoding only the packet D in step 634. If successful, the decoded results are stored in the buffer 450 in step 642 and the job for other information (e.g., information J) proceeds in step 644. Otherwise, the decoder attempts to decode the combination of packets C and D in step 636, which is indicated by *CD in FIG. S. If the combination of packets C and D is decoded, the receiver stores the decoded results in the buffer 450 (step 642) and performs the job for other information (e.g., information J) (step 644). Otherwise, the receiver combines packets C and D with packet B as indicated by *BCD in FIG. 5 and attempts decoding the combination in step 638. Here, if the combination of packets B, C and D is decoded, the decoded results are stored in the buffer 450 (step 642) and the job for other information (e.g., information J) proceeds (step 644). otherwise, the receiver combines packets B, C and D with packet A as indicated by *ABCD in FIG. 5 and attempts decoding combination in step 640. If the combination of packets A, B, C and D is decoded, the receiver stores the decoded results in the buffer 450 (step 642) and performs the job for other information (e.g., information J) (step 644). Otherwise, the process returns to step 614 to repeat all of these procedures until no errors occur. Meanwhile, the receiver stores the decoded results in the buffer 450 in step 642 and performs the job for the next information (e.g., information J, K, . . . ) in step 644.

As described above, the present invention has the characteristics of both Type-1 and Type-2 ARQ methods; therefore, one can obtain constant channel throughput in the burst error containing channel, the random error containing channel, and a channel where the two error patterns coexist simultaneously. In the burst error containing channel, the method of the present invention is performed nearly the same as or better than the Type-1 method and much better than the Type-2 method. As for the random error containing channel, since the method of the present invention is performed similar to the Type-2 method, it also performs almost the same as the Type-2 method, but much better than the Type-1 method.