초록 ▼

A memory configuration scheme that enables parallel decoding of a single block of turbo-encoded data is described. In this scheme a single code block is divided into multiple subblocks and decoding is performed on subblocks in parallel. The turbo decoder memory is configured so that subblock decoder

A memory configuration scheme that enables parallel decoding of a single block of turbo-encoded data is described. In this scheme a single code block is divided into multiple subblocks and decoding is performed on subblocks in parallel. The turbo decoder memory is configured so that subblock decoders can access the common memory resources independently of each other. This scheme is different from existing parallel decoding schemes in that it achieves the parallel implementation by applying multiple decoders to a single code block, not by assigning multiple decoders to multiple code blocks. The advantages of this scheme include minimum memory requirement and minimum decoding latency. The minimum memory requirement results from the fact that it needs memory resources only for a single code block regardless of the number of decoders used. The decoding latency is minimum since decoding of a code block is over when decoding on subblocks is completed.

대표청구항 ▼

We claim: 1. A method of decoding a parallelly concatentated constituent code (PCCC Code) block of N symbols using a turbo decoder operating in the outer mode wherein the turbo decoder comprises first and second component decoders and wherein the method comprises: partitioning the PCCC block into a

We claim: 1. A method of decoding a parallelly concatentated constituent code (PCCC Code) block of N symbols using a turbo decoder operating in the outer mode wherein the turbo decoder comprises first and second component decoders and wherein the method comprises: partitioning the PCCC block into a first input block of M symbols and a second input block of N-M symbols, where M equals one half (N+1); decoding the first input block at the first component decoder; decoding the second input block at the second component decoder; and wherein decoding the first input block and decoding the second input block is performed concurrently and in parallel. 2. A method of decoding a parallelly concatentated constituent code (PCCC Code) block of N symbols using a turbo decoder operating in the inner mode wherein the turbo decoder comprises first and second interleaver memories and first and second component decoders, and wherein the method comprises: storing in sequential order the first interleaver address memory a first index value associated with each of M symbols of the PCCC block, where M equals one half (N+1) and wherein the memory address associates with a first index value; storing in sequential order in the second interleaver address memory a second index value associated with each of N-M symbols of the PCCC block; processing at the first component decoder at interval K the symbol associated with the Kth first index value, where K is an integer in the range of 0 to M-1; processing at the second component decoder the symbol associated with the Kth second index value; and wherein the processing at the first component decoder and the second component decoder is performed in parallel. 3. The method as in claim 2 wherein the decoder further comprises a local memory and wherein the method further comprises storing in the local memory the interleaved symbol data associated with the unique first index value and wherein processing at the first component decoder at interval K the symbol associated with Kth first index value comprises reading at interval K the interleaved symbol data stored in the local memory associated with the Kth first index value. 4. A method of decoding a parallelly concatentated constituent code (PCCC Code) block of N symbols using a turbo decoder operating in an outer mode, wherein the turbo decoder comprises first and second input memories, first and second extrinsic memories, and first and second component decoders, and wherein the method comprises: partitioning the PCCC block into a first input block of M symbols and a second input block of N-M symbols, where M equals one half (N+1); storing the first input block in the first input memory and the second input block in the second input memory; processing at the first component decoder the first input block read from the first input memory and a first set of extrinsic data read from the first extrinsic memory; writing the output of the first component decoder to the first extrinsic memory; processing at the second component decoder the second input block read from the second input memory, a second set of extrinsic data read from the second extrinsic data memory; writing the output of the second component decoder to the second extrinsic memory, and wherein the processing at the first and second component decoders is performed in parallel. 5. A method of decoding a parallelly concatentated constituent code (PCCC Code) block of N symbols using a turbo decoder operating in an inner mode, wherein the turbo decoder comprises first and second input memories, first and second outer extrinsic memories, first and second inner extrinsic memories, wherein the first and second inner extrinsic memories comprise dual port memories, first and second interleaver address memories, local memory, and first and second component decoders, and wherein the method comprises: partitioning the PCCC block received at the decoder into a first input block of M symbols and a second input block of N-M symbols, where M equals one half (N+1); storing the first block in the first input memory and the second block in the second input memory; associating with each of M symbols stored in the first and second input memories a unique first index value and storing in sequential order the unique first index value in the first interleaver address memory; associating with each of N-M symbols stored in the first and second input memories a unique second index value and storing in sequential order the unique second index value in the second interleaver address memory; storing in the local memory the interleaved symbol data associated with the unique first index value; processing at the first component decoder at interval K the symbols associated with the Kth first index value in the local memory and a first set of extrinsic data associated with the Kth first index value wherein the first set of extrinsic data is read from the first outer extrinsic memory if the Kth first index value is less than M and is read from the second outer extrinsic memory if the Kth first index value is equal to or greater than M and where K is an integer in the range of 0 to M-1; writing the output of the first component decoder to the first inner extrinsic memory if the Kth first index value is less than M and writing to the second inner extrinsic memory if the Kth first index value is equal to or greater than M; and processing at the second component decoder at interval K each of N-M symbols associated with the Kth second index value and a second set of extrinsic associated with the Kth second index value wherein the N-M symbol is read from the first input memory if the Kth second index value is less than M and is read from the second input memory if the Kth second index value is equal to or greater than M, and wherein the second set of extrinsic data is read from the first outer extrinsic memory if the Kth second index value is less than M and is read from the second outer extrinsic memory if the Kth second index value is equal to or greater than M; writing the output of the second component decoder to the first inner extrinsic memory if the Kth second index value is less than M and writing to the second inner extrinsic memory if the Kth second index value is greater than or equal to M; and wherein the processing at the first and second component decoders is performed in parallel.