by Leandros Tassiulas (University of Thessaly, Greece)

 

 

Recent advances in information and communication theory as well as networking suggest novel architectures where end-to-end information delivery is facilitated through cooperation and opportunistic caching at various stages during the information transport process. We consider a stochastic dynamic model of information transport in an architecture where various information streams are mixed in an intermediate relay node before delivery. The objective is to increase the efficiency of the system through judicious exploitation of transmission overhearing, inherent in radio broadcast. Other user overheard information is used as a key to extract own information in future transmissions. Algorithms for selecting mixing combinations to increase transport efficiency are considered. They rely on complete information about the state of overheard information.

 

When state information is not readily available but is gradually revealed through feedback, the mixing algorithm should account both for efficient state probing as well as for effective information recovery. System capacity characterization and efficient algorithms are obtained for the limited state information as well. The broadcast erasure channel is considered next. Its information theoretic capacity is characterized and capacity achieving schemes are proposed that rely on judicious exploitation of overheard information stored at the nodes.