Same thing, but use different format. maybe to P2P
1. add algorithm representation
2. add long distance neighbour. how to choose the long distance neighbour. in DHTs, keep neighbours half way to the remaining distance.
3. the overlay structure: how to maintain the membership, for example, how to improve the tree structure's robustness, how if nodes join and leave. the same to the RDV routing. To see if can borrow LiuXin's idea.
4. the whole system simulation, the overhead for maintaining the membership vs the saving for routing. The influce of 'chun' peers join and leave dynamiclly
5. learning, cache knowledge, learn from experiences
