Implementation of approximate decoding algorithm over TCP/IP

In this paper, we consider an implementation of approximate decoding algorithm that enables receivers to recover source data from insufficient number of network coded symbols. We reevaluate several theoretical properties of approximate decoding algorithms in the perspective of actual implementation over TCP/IP and we quantify the impact of packet sizes and the sizes of finite field on the performance of the algorithm. Our experiment results confirm that the theoretical properties are valid in the actual implementation, however, the optimal GF size that minimizes the decoding errors is determined by the packet size.