DART = Dynamic Address RouTing - for scalable ad hoc and mesh networks

Other Pages An ordinary morning in an extraordinary future. The grand vision behind DART. Project Participants Jakob Eriksson Michalis Faloutsos Srikanth Krishnamurthy Publications "Scalable Ad Hoc Routing: The Case for Dynamic Addressing " INFOCOM 2004. "DART: Dynamic Address Routing" ICNP 2003 (poster). "PeerNet: Pushing Peer-to-Peer down the Stack"* IPTPS 2003 Presentations INFOCOM'04 Presentation Keynote, PPT (Exported) * PeerNet is the old name for the DART protocol, which had to be changed due to people confusing DART with an application layer protocol.

Welcome to the DART project web page. The goal of the DART project is to develop a scalable network layer routing protocol for mobile ad hoc and mesh networks. While today's routing protocols perform adequately in networks of dozens, perhaps even hundreds of nodes, they do not scale to large networks. DART addresses this scalability problem by separating the address of a node into two separate numbers: a) a unique and static node identifier, serving the same purpose as today's IP addresses and b) a dynamic routing address, which  indicates the node's current position in the network topology. The use of dynamic routing addresses creates an opportunity for route aggregation, which, in the case of DART, greatly improves scalability. DART proactively maintains accurate routing addresses as well as O(log N) size routing tables on all nodes, where N is the number of nodes in the network. Through the use of dynamic routing addresses, routing is greatly simplified. However, this creates a new problem of address allocation. We describe our method of address allocation, which executes locally on each node, and relies only on routing updates from immediate neighbors to select an available and accurate routing address. DART does not require any geographical location information, nor does it make any assumption as to the underlying medium. Wireless omnidirectional links, as well as directional and even wired links are supported equally well. In addition, nodes participating in a DART network do not require any manual network configuration, making DART a strong candidate for future mesh networking applications in additional to current ad hoc networking applications. We are currently working on improving the performance of the protocol further, as well as introducing security measures into the protocol. It is our goal to create a prototype implementation of the protocol under Linux, and to set up a test bed of stationary wireless and mobile wireless terminals for experimentation and fine-tuning of the protocol.