Network Coding Strategies for Data Persistence in Static and Mobile Sensor Networks

Network coding promises to bring significant performance improvements to sensor network protocols but algorithms need to be designed to cope with the often very constrained resources of sensor nodes. Growth codes proposed by Kamra et al. (Kamra, 2006) are one such example aimed at improving sensor network data persistence. The codes use simple coding operations and require comparatively little memory. However, growth codes are based on the assumption of an extremely dynamic network topology and do not perform well in more stable settings. In this paper we propose modifications to growth codes that are able to achieve good performance over a wider range of static and dynamic scenarios. In particular, we investigate changes of how many and which symbols the transmitted information is coded over and how the decoding is performed. These modifications are analyzed in detail by means of simulations.