| Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 98 === Turbo code is a capacity-approaching error correcting code. Due to its superior performance and reasonable decoding complexity, turbo code has been adopted as the standard in 3GPP-LTE. Iterative decoding method is usually used in turbo decoding, the more the iterations the better the performance. However, as the number of iterations increases, the performance improvement is reduced gradually and finally saturated soon. At this stage, further iterations will result in more power consumption and longer decoding delay instead of performance improvement. A stopping criterion is necessary to resolve this problem. In order to reduce the power consumption and extend the battery durability of a handheld device, the study of stopping criterion is an important issue for its commercialization and is also an active research topic.
So far, the stopping criteria of turbo decoding can be classified into two types of algorithms. One is soft information based algorithm including cross entropy (CE), sign change ratio (SCR), sign difference ratio (SDR), and so on. The other is hard information based algorithm including hard decision aided (HDA). Basically, all of previous works suffer from similar disadvantages including larger memory requirement, more complex computation units, and the calculation of a posteriori probability (APP) per iteration. In this thesis, we proposed a low-complexity of stopping criterion without the disadvantages of previous works. Simulation results show that the proposed algorithm achieve the same performance as previous works in error floor region meanwhile reduce the complexity of stopping criterion dramatically.
|
|---|
