Review on variants of network coding in wireless ad-hoc networks

Abstract

Network coding is a technique in which node is allowed to combine and encode one or more input packets into encoded packets instead of directly forwarding them. It increases throughput and delivery ratio. In this paper, we review basic linear network coding variants with their performance benefits and theoretical results. In practical setting, linear network coding requires central authority to control generation of meaningful encoding coefficients and for coordination between the nodes of network. In wireless network, due to dynamic nature of nodes and heterogeneity of network, centralized approach is not suitable. So, distributed approach should be used. We reviewed distributed linear network coding technique named as Random Linear Network Coding (RLNC). We also reviewed RLNC variants called Generation-by-Generation RLNC and RLNC with Multi Generation Mixing (MGM). We also reviewed the options to recover the lost encoded packets in networks where losses prevent efficient propagation of sender packets. MGM increases the decodable rate of encoded packets. We compared the performance of Generation-by-Generation Network Coding and Network coding with MGM.