Abstract
In this paper, we consider real-time voice transmission or speech communication systems, where voice information is encoded based on network coding techniques. For real-time delivery of data encoded by network coding techniques, the All-Or-Nothing problem of network coding is one of the most important challenges in order to guarantee quality of service (QoS) requirements. In order to overcome the problem, approximate decoding is used for immediate data recovery. In this paper, we focus on optimizing parameters for the best performance of approximate decoding algorithm by explicitly considering the information about source correlation. In particular, we consider the case where consecutive source data sets have symmetric distributions. We analytically show that the best strategy for the approximate decoding algorithm is to use mean of the distributions. Moreover, the performance of the proposed algorithm can improve as the variance of the distributions becomes lower.