We present a novel active queue management (AQM) technique to demonstrate the efficacy of practically harnessing the predictive utility of SSR indications for improved video communication. We consider a network within which corrupted packets are relayed over multiple hops, but a certain percentage of packets needs to be dropped at an intermediate node due to congestion. We propose an AQM technique,survival of the fittest(SOTF), to be employed at the relay node, within which we use packet state information, available from SSR indications and checksums, to drop packets with the highest corruption levels. On the basis of actual 802.11b measurements we show that such a side information (SI) aware processing within the network can provide significant performance benefits over an SI-unaware scheme,random queue management(RQM), which is forced to randomly discard packets. With trace-based simulations, we show the utility of the proposed AQM technique in improving the error recovery performance of cross-layer FEC schemes. Finally, with the help of H.264-based video simulations these improvements are shown to translate into a significant improvement in video quality.