LAUSR

As a critical part of public transportation systems, buses driving routinely in urban areas naturally form a bus ad-hoc network (BusNet). Different from conventional vehicular ad-hoc networks, the sparse distribution of buses magnifies the problem of intermittent network connectivity, making it more challenging for real-time routing. In this paper, we first present a novel architecture called software-defined BusNet, where the distributed controllers (i.e., base stations) collect traffic information and real-time requests from buses via vehicle-to-cellular communication. The collected information enables the logically centralized controller to make routing decisions with global traffic information. Then, the routing decisions are returned to roadside units (RSUs) and buses, which will instruct the data broadcast from RSUs to buses via vehicle-to-roadside (V2R) communication, and data forwarding among the buses through vehicle-to-vehicle (V2V) communication. On this basis, we formulate the problem of delay-constrained routing for BusNet (DCRB) which aims to maximize both the delivery ratio and the bandwidth efficiency. We build a probabilistic model to analyze the encounter probability and encounter delay between the relay buses based on the predicable bus traces. Further, we derive the expected reachability and expected delivery delay of the multi-hop routing path. The routing graph is constructed for searching the feasible routing paths via the modified Dijkstra algorithm. Four routing selection strategies are proposed based on different metrics: HRF (highest reachability first), SDF (shortest delivery delay first), MHF (minimum number of hops first), and RDH (combination of reachability, delay, and hop). Finally, we build the simulation model and implement the proposed routing algorithms for performance evaluation. The comprehensive simulation results demonstrate that RDH achieves the best overall performance on improving system delivery ratio and enhancing bandwidth efficiency.