The cluster formation strategies for approximate decoding in IoT networks

In this paper, we consider delay-constrained data transmission based on network coding techniques over error-prone IoT networks. While network coding approaches can provide various advantages, there is a critical drawback referred to as all-or-nothing problem; the encoded source data cannot be recovered if a set of required number of data is not entirely received by a decoding deadline. As a solution, an approximate decoding approach has been proposed. In this paper, we quantify the performance of approximate decoding and show that the performance is determined only by the insufficient number of packets. Moreover, we analytically show the fundamental tradeoff between the performance of the approximate decoding and data transfer rate improvement; as the cluster size increases, data transfer rate is improved while the decoding performance is degraded. This tradeoff can lead to an optimal cluster size of network coding based networks that achieves the target decoding performance of applications. The analysis is confirmed by a set of experiments.