Development and Study of a Class of Generic Reliability Based Direct Decoding Approaches for Turbo Codes

Abstract

Rapid advancement in wireless communication systems demand energy efficient and performance enhanced error correction schemes. Error correcting codes have become an integral part of many wireless communication systems. The Shannon performance approaching codes like turbo codes, LDPC codes, polar codes and repeat-accumulate (RA) codes are now being widely used in many of the practical channels. Turbo codes invented by Claude Berrou in 1993 are efficiently decodable using iterative techniques. Turbo codes have the dual benefit of having a code structure that can achieve newlinenear Shannon performance and an efficient decoder design. In spite of the complexity newlineassociated with iterative decoding of turbo codes, it has become a code of choice for many applications like satellite communication, mobile communication, WiMAX, and wireless sensor networks (WSNs). The conventional iterative turbo decoder which uses maximum a posteriori (MAP) or its variants like log-MAP or Max-log-MAP as the constituent decoder performs affixed number of iterations irrespective of the channel conditions and information block size. Reliable and low latency communication has gained paramount significance in the recent years with the increased interest in current and emerging communication applications like satellite telecommand, machine to machine communication, cellular networks, WSNs, smart buildings, and ultra reliable low latency communication. Practically, transmission with long data stream introduces decoding delay and complexity. To achieve reliable and low latency communication, information transmission using short block length messages is required. However as information block length reduces, conventional iterative turbo decoding results in performance degradation and high error floor .Developing soft decision decoding algorithms to achieve reliable and low latency communication with short block length messages still remains an open research problem. This research work focuses on enhancing the error correction performance...