Distributed communication has a great potential to become a promising foundation for future cognitive radio networks regarding its ability to provide a reliable multimedia transmission under lossy and delay constrained conditions. In this paper, we explore a novel distributed system for cognitive radio networks relying on the concept of multiple description coding for the more realistic case where packets are dropped due either to primary interruptions or subchannel fading. In particular, the initial stream is distributed across multiple nodes, each will be responsible of conveying one entire description using a single subchannel without the need to coordinate its activity with the other sources and the common destination will merge the received descriptions to recover the delivered stream. Numerical simulations are performed to investigate the throughput benefits of the proposed scheme, we argue that there is an optimal number of nodes providing the maximum achieved throughput while protecting the secondary content against the loss pattern.