Applications of Iterative Soft Decision Decoding

Hard-decision decoders lose some useful information through quantization. Soft-decision decoders use side information from the demodulator to improve the coding gain. Thus, the softdecision decoders exhibit a better performance as compared to the contemporary hard-decision decoders. Soft-decision decoders for Turbo Codes and Low Density Parity Check codes have already been demonstrated to approach the Shannon limit. In this dissertation, some iterative softdecision decoding algorithms are proposed and their applications to communication protocols and data authentication in the presence of noise are investigated.

In the first part of this dissertation, a soft-decision decoding algorithm for Reed-Solomon (RS) codes is proposed. The algorithm iteratively erases a combination of least reliable symbols and flips another combination of least reliable bits. It is shown that for different code-rates and codelengths, a coding gain of up to 1 dB is obtained at a lower iteration complexity as compared to the other well known soft-decision decoders. Standard RS decoder may result in decoding errors (also known as miscorrections). A method is given to avoid the decoding errors, or minimize their probability of occurrence, as total prevention is not possible.

In the second part, some applications of soft-decision decoding to communication protocols are investigated. A soft-decision decoding algorithm, based on the Chase decoder, is introduced for the HARQ protocol. It is shown to improve the packet error rate. This method is then extended by integrating it with the multichannel transmission mechanism of WiMAX. The checksums of these packets are protected using RS codes. The erroneous packets are identified using the error locator polynomial of the RS decoder. These packets are then corrected using a soft-decision decoder. This is followed by a packet combining soft-decision decoder. Most of the existing packet combining methods can correct an odd number of errors. It is shown that the proposed method is able to correct both even and odd number of errors.

Finally, the application of soft-decision decoding to data authentication in the presence of noise is studied. Message can be authenticated part wise, even if the message as a whole could not be authenticated (fuzzy authentication). An algorithm is presented to decode and correct blocks of messages, protected by Noise Tolerant Message Authentications Codes (NTMACs). Higher protection for more important parts of a message using Weighted-NTMACs is also initiated. Error localization, correction and authentication are achieved at block level. These attributes of the proposed algorithms make them especially suitable for multimedia communications such as text, image, audio or video transmission.