Pseudo Random Network Coding in Infrastructure to Vehicle Environment for Data Download

Abstract

This paper introduces a new approach for infrastructure based content distribution in a vehicular network. It is built on broadcasting and pseudo random network coding. Its main strength is that, being broadcast based, it does not need any feedback channel and thus uses less data rate. Data is transmitted exploiting network coding, multiple linear combinations of data are sent. A vehicle needs to receive a defined number of independent linear combinations to decode the data. The server will send a larger number of different linear combinations. The unreliability of broadcast is thus neutralized through a useful redundancy rather than through re-transmission. Finally, computation of the linear combination coefficients is done so that the overhead is the same as it would be without network coding. Depending on the infrastructure deployed, this technique can be a content distribution solution per se or the first step in a more general solution, the second step being based on collaborative download. The high diversity in transmissions will then be a key feature for the performance of such an application.