The decision-making behaviors of drivers have been clarified, as various researches on traffic behavior have been conducted until today. However, existing models set limits to examine non-linearity, ambiguity and the errors, which are occurred by variety of space-time contained in human conscious structure and process of decision-making. In particular, it is true that the route choice behavior of drivers is affected by external factors, such as past experience and change in diverse road environment information. However, taking into account the fact that various traffic problems are not being solved, it is proven that there are limitations in clearly describing traffic behavior. Accordingly, in order to construct a model that considers obscurity and non-linearity in the decision-making process of drivers, a model through neural network linkage and genetic algorithm was proposed in this study. In addition, this study was focused on proving the effectiveness of neural network theory and genetic algorithm, and studying the appropriateness of applying them into the field of transportation.

In iterated product of single parity check (SPC) codes, weight distribution is an important factor for the performance against transmission errors. A product code with Gaussian weight distribution should have a good performance. We present a closed-form solution for the weight distribution of a recursive SPC product code. We show that the code weights for this code are symmetrically distributed at (N plusmn radic(N)/2), where N is the full-length of a codeword. Though this code does not have a Gaussian weight distribution, it has better performance than conventional product codes. When soft-output iterative decoding is applied, the performance is away from the Shannon capacity limit by only 0.95 dB. Hence, we conclude that Gaussian weight distribution is not a necessary condition for a good performance.