UEP network coding for scalable data
Apirath Limmanee, Werner Henkel · 2008
We analyze the effect of erasures on the recovery of scalable data when linear network coding is applied. We find that global encoding kernels (GEKs) describing linear network codes have different levels of built-in unequal-erasure-protecting (UEP) capability, allowing us to better protect the high-priority data when GEKs are wisely assigned. A sub-optimal assignment strategy is suggested to achieve fairness and optimality.