CAREER: Iterative Decoding Schemes For Channels With Memory: Application To Fading Channels

点击翻译按钮获取中文摘要

Error correcting codes (channel codes) are on e of the key elements to optimize digital communications systems. Traditionally, the design of good error correcting codes assumes memoryless channels. In this context, turbo codes and the rediscovery of low-density parity check (LDPC) codes represent two of the most significant advances in channel coding in recent years: Iterative decoding of these codes makes it possible to achieve performance close to the theoretical limits for memoryless channels. However, in most of the applications, the channel is not so simple. This research focuses on the study of iterative decoding for more realistic channels, such as wireless communications channels, characterized by having memory. The objective is to achieve reliable communications, close to theoretical limits, for these types of channels. This will have a direct application in the design of realistic communications systems (including wireless communications), allowing a reduction in the transmitter power requirements for a given quality of service, and a better use of the available bandwidth.In order to achieve the best possible performance when the channel has memory, the statistical properties of the channel must be exploited in the decoding process. This will be accomplished in a two-fold process: First, statistical models, such as hidden Markov models and stochastic grammars, the iterative decoding schemes will be modified to incorporate the statistical models in the decoding of turbo codes, LDPC codes, and concatenated space-time codes. Both steps are completely interwined. The idea is to jointly design the statistical models and the decoding modifications, taking the decoding performance for the real channel as the real channel as the optimization criterion. Moreover, when possible, this process should work adaptively, with no a priori knowledge of the channel required: when the communications system is used in an unknown channel, a convenient statistical model of the channel should be obtained jointly with decoding (in either a completely blind fashion if possible or by using pilots). In every iteration such a model should be used for the decoding and be conveniently refined.