Sanjit Biswas

Architecture and evaluation of an unplanned 802.11b mesh network

This paper evaluates the ability of a wireless mesh architecture to provide high performance Internet access while demanding little deployment planning or operational management. The architecture considered in this paper has unplanned node placement (rather than planned topology), omni-directional antennas (rather than directional links), and multi-hop routing (rather than single-hop base stations). These design decisions contribute to ease of deployment, an important requirement for community wireless networks. However, this architecture carries the risk that lack of planning might render the networks performance unusably low. For example, it might be necessary to place nodes carefully to ensure connectivity; the omni-directional antennas might provide uselessly short radio ranges; or the inefficiency of multi-hop forwarding might leave some users effectively disconnected.The paper evaluates this unplanned mesh architecture with a case study of the Roofnet 802.11b mesh network. Roofnet consists of 37 nodes spread over four square kilometers of an urban area. The network provides users with usable performance despite lack of planning: the average inter-node throughput is 627 kbits/second, even though the average route has three hops.The paper evaluates multiple aspects of the architecture: the effect of node density on connectivity and throughput; the characteristics of the links that the routing protocol elects to use; the usefulness of the highly connected mesh afforded by omni-directional antennas for robustness and throughput; and the potential performance of a single-hop network using the same nodes as Roofnet.

Opportunistic routing in multi-hop wireless networks

This paper describes Extremely Opportunistic Routing (ExOR), a new unicast routing technique for multi-hop wireless networks. ExOR forwards each packet through a sequence of nodes, deferring the choice of each node in the sequence until after the previous node has transmitted the packet on its radio. ExOR then determines which node, of all the nodes that successfully received that transmission, is the node closest to the destination. That closest node transmits the packet. The result is that each hop moves the packet farther (or average) than the hops of the best possible pre-determined route.The ExOR design addresses the challenge of choosing a forwarding node after transmission using a distributed algorithm. First, when a node transmits a packet, it includes in the packet a simple schedule describing the priority order in which the potential receivers should forward the packet. The node computes the schedule based on shared measurements of inter-node delivery rates. ExOR then uses a distributed slotted MAC protocol for acknowledgements to ensure that the receivers agree who the highest priority receiver was.The efficacy of ExOR depends mainly on the rate at which the reception probability falls off with distance. Simulations based on measured radio characteristics [6] suggest that ExOR reduces the total number of transmissions by nearly a factor of two over the best possible pre-determined route.