Evaluation of a Compound Node Selection Function for a Greedy Routing in Urban Vehicular Ad Hoc Networks.

The greedy routing that reduces the Euclidean distance to destination continuously to the possible maximum in every forwarding is suitable for vehicular ad hoc networks with high node mobility due to operation simplicity and low overhead. However, the selection of a next forwarding node based on the forward progress distance alone suffers from varying link conditions by high node mobility and communication obstructions. Thus, a compound node selection function is developed to include link quality factor. In this function, the accuracy of the forward progress distance largely depends on the node location error that comes from either GPS measurement error or the elapse of time after measurement. Location prediction is conducted by the Kalman filter and link quality is assessed using the successfully transmitted Hello messages. The compound function is evaluated against variations of fading and shadowing effects and the injected GPS error and is applied to a greedy routing protocol to prove its superiority.