In the context of mobile sensor networks we leverage the connectivity between travelling vehicles in an urban area to propagate traffic information that is generated by sensors carried by vehicles. Due to the nature of network connectivity in this scenario, i.e. intermittent connections and absence of contemporaneous end-to-end paths, our primary concern is to establish a reliable and efficient routing layer.
In the scenario we are investigating, the network consists of vehicles travelling in an urban environment and several stationary gateways spread across the city that do not provide full city coverage as shown on the figure. When traffic density is low or when only few vehicles carry a wireless transceiver, the vehicular network often becomes disconnected. Carry-and-forward protocols are required for the reliable delivery of messages between vehicles in dynamically changing network partitions.
We aim to devise carry-and-forward algorithms that leverage knowledge of traffic statistics in an urban setting to enable timely delivery of messages from vehicles to stationary gateways, whilst minimizing message transmissions and optimizing bandwidth utilization. Our algorithms intend to minimize the number of transmissions while forwarding a message to a gateway within the message-specific delay threshold. To do so, they proactively alternate between two forwarding strategies: Multihop Forwarding, which refers to the aggressive forwarding of messages to vehicles that are better positioned to deliver them to an access point and Data Muling, which refers to buffering messages in local memory and carrying them at the vehicle's speed.
Consider the example on the right. A message is generated on a vehicle southeast (red circle) and we need to forward that message to one of the gateways (green circles). When two vehicles are in range, the routing algorithm decides which vehicle is more suitable to carry the packet to the destination, resulting in the packet's trasmission hop by hop towards a gateway(multihop forwarding). Sometimes vehicles are travelling on the right path towards a gateway and it makes sense to buffer the packet instead (data muling). You can see a short description of the algorithm here.