Locally invertible convolutional encoders, table-based decoding, and their applications to high speed communications

Reliable high speed digital communication requires high speed error control techniques. Convolutional coding approaches to Forward Error Correction (FEC), hybrid Automatic Repeat Request (ARQ), and error recovery in high speed networks are examined. FEC using convolutional coding at very high data rate requires good convolutional encoders and fast decoding schemes. A new class of locally invertible convolutional encoders is defined. A one-to-one mapping between information and encoded blocks of equal length is the key property of encoders in this class. A new interpretation of the syndrome sequence resulting from this property is given. It is shown that a rate-1/n convolutional encoder is both nondegenerate and noncatastrophic if and only if it is locally invertible. Table-based convolutional decoding techniques are developed. In table-based error correction, a finite length syndrome is used as an address to access precomputed correction information stored in a table. Correction table generation is described and table entry conflict resolution strategies are discussed. Performance of this scheme is analyzed and simulation results are presented. Applications of table-based decoding and locally invertible convolutional encoders for error control and error recovery in high speed networks are developed. Hybrid ARQ error control protocols combine FEC with error detection and retransmission in order to benefit from the advantages of both FEC and ARQ schemes. A type-I table-based convolutionally encoded hybrid ARQ scheme is described. The performance of this protocol can be improved by finely tuning retransmission requests so that throughput losses are minimized. A variable-redundancy hybrid ARQ scheme using locally invertible convolutional encoders is developed. A solution for lost packet recovery in high speed networks using locally invertible convolutional encoders is described.